Sedimentary Crisis at the Global Scale 1 : Large Rivers, from Abundance to Scarcity 9781119579861, 1119579864, 9781786303837, 1786303833

Table of contents :
Content: Cover
Half-Title Page
Dedication
Title Page
Copyright Page
Contents
Foreword
Preface
Acknowledgements
Introduction
1. The Torrential Crisis in the European Mountains (14th-19th Centuries)
1.1. Introductory generalities on global fluvial systems
1.2. Manifestations of the LIA crisis in the river valleys of Western Europe
1.2.1. Mountain crises
1.2.2. River crises and metamorphoses of the Drac and the Isère in Grenoble
1.2.3. Flooded piedmont plains in Switzerland
1.2.4. Sedimentation and large works in Italy
1.3. The difficult mastery of the Rhine delta in the modern era 1.3.1. Flow distribution between river branches: an age-old battle against the elements of nature1.3.2. Returns on destabilization
1.4. Observations on the torrentiality of the Southern Alps in the late 18th and 19th Centuries
1.4.1. A highly degraded state of affairs in the late 18th Century
1.4.2. Prefect Pierre-Henri Dugied's project (1819)
1.4.3. Alexandre Surell, author of the French policy for restoring mountain territories
1.4.4. The restoration of mountain land (RTM)
1.4.5. The Southern Prealps (Drôme): what kind of balance in torrential milieus? 1.5. The sediment conveyor belt, from torrents to outlets1.5.1. The forester Georges Fabre, from the Aigoual to the Gironde
1.5.2. The Rhône river trough
1.5.3. The redistribution of alluvia in the upper delta of the Rhône
1.5.4. Solid contributions to the Rhône outlet and progression of the Camargue delta
2. Continuity in European Hydraulic Science (16th-18th Centuries)
2.1. From hydraulic architecture to the fluvial system: transalpine preeminence
2.1.1. At the roots of European science
2.1.2. A great Italian scholar, Paolo Frisi 2.2. The first naturalist approaches to the water cycle in the Seine basin2.2.1. Pierre Perrault
2.2.2. Edme Mariotte
2.2.3. French hydraulic science in the 18th Century
2.2.4. Emergence of the natural state of rivers in the mid-18th Century
2.2.5. Jean-Antoine Fabre, the great naturalist engineer of Southern Alpine torrents
2.3. Conclusion
3. Exploited Nature and the River's Responses to the Globe's Surface
3.1. Mistreated soil and accelerated erosion
3.1.1. The Huang-He (Yellow River) basin: accelerated erosion in ahighly fragile milieu
3.1.2. Soil erosion in North America 3.1.3. Accelerated erosion on the Great Russian Plains, from Belarus to the Urals3.1.4. New Zealand, "destruction on the pretext of development" [WYN 02]
3.2. Mineral predation and river bursts
3.2.1. Lead and zinc in the Pennines: mines threatening dairy livestock
3.2.2. The "debris" from the gold-bearing alluvia of the Sierra Nevada (California)
3.2.3. The coal mines of the Loess Plateau, the Huang-He watershed
3.2.4. Mountaintop mining in the Appalachians at the risk of downstream reaches
3.3. Conclusion
4. From Hills to the Ocean: Production, Transfer and Trapping

Citation preview

Sedimentary Crisis at the Global Scale 1

To Geneviève, for her support

Series Editor Yves Lagabrielle

Sedimentary Crisis at the Global Scale 1 Large Rivers, from Abundance to Scarcity

Jean-Paul Bravard

First published 2019 in Great Britain and the United States by ISTE Ltd and John Wiley & Sons, Inc.

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms and licenses issued by the CLA. Enquiries concerning reproduction outside these terms should be sent to the publishers at the undermentioned address: ISTE Ltd 27–37 St George’s Road London SW19 4EU UK

John Wiley & Sons, Inc. 111 River Street Hoboken, NJ 07030 USA

www.iste.co.uk

www.wiley.com

© ISTE Ltd 2019 The rights of Jean-Paul Bravard to be identified as the author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988. Library of Congress Control Number: 2018963016 British Library Cataloguing-in-Publication Data A CIP record for this book is available from the British Library ISBN 978-1-78630-383-7

Contents

Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xv

Chapter 1. The Torrential Crisis in the European Mountains (14th–19th Centuries) . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

1.1. Introductory generalities on global fluvial systems . . . . . 1.2. Manifestations of the LIA crisis in the river valleys of Western Europe . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.1. Mountain crises . . . . . . . . . . . . . . . . . . . . . . 1.2.2. River crises and metamorphoses of the Drac and the Isère in Grenoble . . . . . . . . . . . . . . . . . . . . 1.2.3. Flooded piedmont plains in Switzerland . . . . . . . . 1.2.4. Sedimentation and large works in Italy . . . . . . . . . 1.3. The difficult mastery of the Rhine delta in the modern era 1.3.1. Flow distribution between river branches: an age-old battle against the elements of nature . . . . . . . . . 1.3.2. Returns on destabilization . . . . . . . . . . . . . . . . 1.4. Observations on the torrentiality of the Southern Alps in the late 18th and 19th Centuries . . . . . . . . . . . . . 1.4.1. A highly degraded state of affairs in the late 18th Century . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2. Prefect Pierre-Henri Dugied’s project (1819) . . . .

. . . . . . . . . . .

1

. . . . . . . . . . . . . . . . . . . . . .

5 5

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

7 11 12 17

. . . . . . . . . . . . . . . . . . . . . .

17 19

. . . . . . . . . . .

20

. . . . . . . . . . . . . . . . . . . . . .

20 23

vi

Sedimentary Crisis at the Global Scale 1

1.4.3. Alexandre Surell, author of the French policy for restoring mountain territories . . . . . . . . . . . . . 1.4.4. The restoration of mountain land (RTM) . . . . 1.4.5. The Southern Prealps (Drôme): what kind of balance in torrential milieus? . . . . . . . . . . . . . . 1.5. The sediment conveyor belt, from torrents to outlets . 1.5.1. The forester Georges Fabre, from the Aigoual to the Gironde . . . . . . . . . . . . . . . . . . . . . . . 1.5.2. The Rhône river trough . . . . . . . . . . . . . . . 1.5.3. The redistribution of alluvia in the upper delta of the Rhône . . . . . . . . . . . . . . . . . . . . . 1.5.4. Solid contributions to the Rhône outlet and progression of the Camargue delta . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

23 28

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

31 32

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

32 33

. . . . . . . . . . . . . .

35

. . . . . . . . . . . . . .

35

Chapter 2. Continuity in European Hydraulic Science (16th–18th Centuries) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39

2.1. From hydraulic architecture to the fluvial system: transalpine preeminence . . . . . . . . . . . . 2.1.1. At the roots of European science . . . . . . . 2.1.2. A great Italian scholar, Paolo Frisi . . . . . . 2.2. The first naturalist approaches to the water cycle in the Seine basin. . . . . . . . . . . . . . . . . . 2.2.1. Pierre Perrault . . . . . . . . . . . . . . . . . . 2.2.2. Edme Mariotte . . . . . . . . . . . . . . . . . 2.2.3. French hydraulic science in the 18th Century 2.2.4. Emergence of the natural state of rivers in the mid-18th Century . . . . . . . . . . . . . . . . . 2.2.5. Jean-Antoine Fabre, the great naturalist engineer of Southern Alpine torrents . . . . . . . . . 2.3. Conclusion . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41 42 44

. . . .

. . . .

50 51 52 53

. . . . . . . . . . . . . . . .

59

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

62 67

Chapter 3. Exploited Nature and the River’s Responses to the Globe’s Surface . . . . . . . . . . . . . . . . . . . . . . . . . .

69

. . . .

. . . .

3.1. Mistreated soil and accelerated erosion . . . . . . . . . 3.1.1. The Huang-He (Yellow River) basin: accelerated erosion in a highly fragile milieu . . . . . . . . . . . . . . 3.1.2. Soil erosion in North America . . . . . . . . . . . . 3.1.3. Accelerated erosion on the Great Russian Plains, from Belarus to the Urals . . . . . . . . . . . . . . . . . . 3.1.4. New Zealand, “destruction on the pretext of development” . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . . . . . .

71

. . . . . . . . . . . . . . . . . . . . . . . . . .

71 77

. . . . . . . . . . . . .

83

. . . . . . . . . . . . .

86

Contents

3.2. Mineral predation and river bursts . . . . . . . . 3.2.1. Lead and zinc in the Pennines: mines threatening dairy livestock . . . . . . . . . . . . . 3.2.2. The “debris” from the gold-bearing alluvia of the Sierra Nevada (California) . . . . . . . . . 3.2.3. The coal mines of the Loess Plateau, the Huang-He watershed . . . . . . . . . . . . . . . . . 3.2.4. Mountaintop mining in the Appalachians at the risk of downstream reaches . . . . . . . . . . . 3.3. Conclusion . . . . . . . . . . . . . . . . . . . . .

vii

. . . . . . . . . . . . . . . . .

91

. . . . . . . . . . . . . . . . .

92

. . . . . . . . . . . . . . . . .

93

. . . . . . . . . . . . . . . . .

100

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

101 104

Chapter 4. From Hills to the Ocean: Production, Transfer and Trapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

107

4.1. Global continental contributions to oceans . . . . . . . . . . 4.1.1. Continental denudation and sediment flux to river mouths . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2. Natural sediment interception on the way to oceans . . . 4.1.3. Disturbances in “geological” fluxes during the Anthropocene . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Selected case studies on the Earth’s surface . . . . . . . . . . 4.2.1. The Yangzi basin . . . . . . . . . . . . . . . . . . . . . . 4.2.2. The sediment load of rivers in mountain regions subject to tropical cyclones . . . . . . . . . . . . . . . . . . . . 4.2.3. The effects of the recent protection of degraded continental milieus . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4. Mining and the increase in river loads . . . . . . . . . . 4.3. Irreversible flux disturbances . . . . . . . . . . . . . . . . . . 4.3.1. The major role of artificial reservoirs . . . . . . . . . . . 4.3.2. Hydrological and sedimentary effects . . . . . . . . . . . 4.3.3. Trapping and effects on sediment transfer . . . . . . . . 4.3.4. River diversion, loss of transport capacity and trapping . 4.3.5. Predation of river resources: sand and gravel. . . . . . .

. . . . . . . . . .

107

. . . . . . . . . . . . . . . . . . . .

108 112

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

117 121 121

. . . . . . . . . .

121

. . . . . . . .

. . . . . . . .

122 125 126 126 127 129 131 135

Chapter 5. The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers . . . . . . . . . . . . . . . . . . . . . . .

145

5.1. A river in its natural state, the Amazon . . . . . . . . . 5.1.1. The river in its basin . . . . . . . . . . . . . . . . . 5.1.2. River function . . . . . . . . . . . . . . . . . . . . . 5.1.3. The threat of dams . . . . . . . . . . . . . . . . . . 5.2. Adjusted rivers in China and Southeast Asia . . . . . . 5.2.1. The Huang-He downstream of the Loess Plateau: contemporary generalities . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . . . . .

. . . . .

. . . . .

146 146 147 149 150

. . . . . . . . . . . . .

150

viii

Sedimentary Crisis at the Global Scale 1

5.2.2. The Yangtze and the Three Gorges Dam . . . . 5.2.3. The Mekong . . . . . . . . . . . . . . . . . . . . 5.3. The Mississippi, an altered river in a new country . 5.3.1. Basin and hydrology . . . . . . . . . . . . . . . 5.3.2. Geology of the Mississippi basin . . . . . . . . 5.3.3. Aspects of the river . . . . . . . . . . . . . . . . 5.3.4. Modifications to the sedimentary budget . . . . 5.4. Overexploited rivers in regions with a water deficit 5.4.1. The God River and the Aswan Dams . . . . . . 5.4.2. The Colorado River . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

155 160 166 166 167 168 170 176 176 182

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

189

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

193

Bibliography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

199

Index of Common Terms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

219

Index of Places . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

223

Index of Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

229

Foreword

Rivers and deltas have been essential to the human enterprise on planet Earth since the dawn of civilization, along the Tigris and the Euphrates, the Nile, the Yangtze and the Indus. Rivers provide domestic, commercial and industrial users with water, not to mention irrigation for agricultural purposes; they are a means of transport, a habitat for fauna, a source of hydroelectric production and fertile river sediment to enrich the floodplains. Some of them carry solid sediment in the form of tailings, which may affect aquatic ecosystems and be deposited on the floodplains. Transported sediment comes from the bed and banks of the river as well as the watershed of the highlands, which has consequences for the stability of the channel, every use of rivers and its floodplains. Rivers transport more sediment to the world’s oceans than all other agents, such as glaciers, groundwaters, wind, volcanoes and waves. When water rich in energy and sediment enters a low-energy water plane, low-altitude deltas are formed. These are as important for humans as they are for fauna. They can serve as a buffer against floods, but less and less efficiently as they sink and are occupied, over time, by humans. Rivers and deltas have been struck by natural disasters like hurricanes and floods. They have also been transformed by humans over the centuries, through direct manipulation in the form of dams, levees, diversions, bank stabilization, channeling/recovery, and the extraction of sand and gravel. Rivers and deltas are also indirectly affected in the watershed by human activities that modify the delicate balance between the water and sediment in the river system and that bring about change in the vegetation (deforestation/reforestation, grazing livestock, forest fires and agriculture). One of the greatest challenges facing scientists who study rivers and deltas lies in distinguishing the impacts of human activity from the changes that would have taken place without human interference. This challenge demands an integrated understanding of the hydrologic cycle, fluid dynamics, sediment cascades, water

x

Sedimentary Crisis at the Global Scale 1

resource engineering, the history of the use of land and water, water quality, and the transport of contaminants. Professor Bravard takes inspiration from all of these subjects, which contribute to a better understanding of the role of the river and the delta over time. This book begins with a study of the changes in European rivers, which began in the Middle Ages and continued throughout the Little Ice Age. In the late 19th Century, the impoverishment of river sediment became a major problem, with unexpected consequences for mankind. The reader will be impressed to discover to what extent rivers and deltas have changed over time, to what extent humans and natural factors have exercised their influence and to discover the consequences of these changes on human activities. It is difficult to imagine an author more qualified than Professor Jean-Paul Bravard to write a book detailing the geography and history of the “rolling carpet of sediment” from river sources, the transport through their “river gutters” and deposition at their deltas. In this book, Bravard compiles the lessons gained from his education and his own experience to discuss the change in character and behavior in rivers and deltas. Bravard describes and explains the different forms that this metamorphosis takes through the use of striking examples in Europe, Russia, China, North America, the Amazon, Egypt and Asia. Bravard has compiled a research file that has earned him international recognition for his work on rivers and deltas, detailed in multiple books and more than 350 articles in highly renowned national and international journals. This led to him being recruited as a consultant for the Rhône and Loire authorities and, more recently, the World Wildlife Fund for Nature in Southeast Asia. He has also received the CNRS1 Silver Medal in Human Sciences for his work. In this volume, his description, explanations and predictions concerning rivers and deltas are now synthesized for the informed public. Richard A. MARSTON University Distinguished Professor at Kansas State University Editor-in-Chief of Geomorphology, 1999–2018 Former President (2005–2006) of the American Association of Geographers Member of the AAAS, AAG, GSA, Explorer’s Club, Royal Geographical Society

1 National Center for Scientific Research (Centre national de la recherche scientifique).

Preface

For the past 20 years, scientists, NGOs and a portion of public opinion has been concerned with the threats weighing on the globe’s deltas. The most visible way in which this danger manifests itself is through the multiplication of disasters linked to cyclones; they cause considerable casualties and extensive damage to a number of coastal communities. In fact, coasts around the world seem to be more severely affected than they used to be by the violence of winds and storms resulting from the rise in the sea level. The issue of climate change must not, however, turn our attention from other crisis factors experienced by deltas. These actually accumulate major weaknesses that are important to their very nature: their flatness, their very limited altitude connected to their young age, a withdrawal from the coast and the increasing severity of natural recess. Deltas, which demonstrate fertile land and are highly populated, are, in many cases, also lands of suffering, because, in addition to their exposure to natural phenomena, they also suffer from the over-exploitation of their resources, a lack of water that seems paradoxical and, sometimes, war. However, deltas are not autonomous spaces. Owing to the very fact that they are at the river mouths of continental surfaces that have been occupied for millennia by dense human societies, they are also subject to dynamics much more directly connected to natural processes and human actions of a continental nature. Deltas actually owe their existence to the deposition of fine sediment carried by rivers to their outlets into the sea. These contributions have varied over the centuries and millennia as a response to changes in the climate and the effect of human actions on erosion in basins; river basins around the globe have experienced periods of intense sedimentary transport contrasting with periods of limited transport. Deltas are subject to the acceleration in the rise of the seas and climatic hazards, but to a large extent, they can respond and adjust on condition that the contribution of sediment with a continental origin maintains sufficient intensity. In the opposite case, these contributions reduce in intensity and deltas are or will no longer be able to

xii

Sedimentary Crisis at the Global Scale 1

compensate for their recess or the rise in the sea level, and this is the case for most deltas around the world today. Why are the links between continental spaces and deltas considered distended or cut off? If the need for sedimentary contribution to deltas has been proven today, why were these contributions unceremoniously intercepted in the 20th Century? Could the effects of the rising sea level have been misevaluated or neglected, leading to the burning issue today? Or is it rather because historic continuity, which leads sediment from hills and mountains to the sea, was not understood, or was forgotten, or even considered negligible, inconvenient for certain practices in place in river systems? In short, what are the root causes of the contemporary crisis of deltas? These are the questions this book will attempt to answer. Its goal is to raise awareness, beyond the scientific circle, of this complex phenomenon, because it lies at the intersection of natural processes and human choices. The situation is serious and implies understanding the nature of the exchanges between rivers and oceans around the globe, particularly at deltas, which are also affected by the effects of climate change. To respond to the questions from the preceding paragraph, we have made the methodological choice to first understand and present the history of rivers in order to better target the present and future of deltas, because they owe their existence to rivers. We will start by explaining where the contemporary landscape of Europe’s rivers comes from; then, we will ask what Europe has learned and passed on to the heirs of its remarkable intellectuals. We will note that the knowledge regarding the theoretical operations of rivers has not brought us to ethical practices, concerning both rivers and their natural extension, deltas. The chapters of this work are introduced below. Their order tends to suggest that the best river science does not necessarily bring about better practices, and that bad practices have a cost that society must always pay in one way or another. The delta, less frequently flooded, in channels less traveled for transportation, is abandoned to its fate when stability seemed certain in the long term. The progress of technology and practices seemed to have made upstream–downstream solidarity useless in the mid-20th Century. However, for the first time in human history, the deltas have brutally become the victims of the improper management of continental waters, for the sedimentary balance* is off. Hydroelectric development has been the major cause of the fragmentation of the elements of the hydrographic network and drainage basins, at the same time that the extraction of river resources was experiencing a considerable, uncontrolled rise, including in deltas themselves. It is rare to observe the convergence of bad policies, even though the responsibility for the current disorder weighs, first and foremost, on the management of continental river systems.

Preface

xiii

Acknowledgements The author would like to say a big thank you to Yves Bégin, Geneviève Bravard, Marc Goichot, Richard Marston, Michel Meybeck and Laurent Touchart for their shrewd advice in the editing phase; Thierry Sanjuan for his original suggestions; Colette Bedoin for her presence during a difficult technical stage and Emmanuelle Szychowiak for her translation of a German text. He would also like to express his gratitude to Richard Cosgrove (North Canterbury Fish & Game), Neil Cullen (New Zealand Farm Forestry Association), Marc Goichot (cover of volume 1), Robin Gruel, Allan James, Lois Koehnken, Ingrid E. Luffman, Georges Pichard, Peter Scott and Vivian Stockman (OVEC) for granting reproduction rights. The resources provided by Gallica were also extensively used in this work. Jean-Paul BRAVARD December 2018

Introduction

The first volume of this work starts with the history of a crisis that affected Europe in the late Middle Ages and the modern era. This crisis extended from the mid-14th Century to the mid-19th Century, five centuries dominated by misfortune, from the mountains to the mouths of the valleys. The mountains actually saw great torrential activity, whereas the rivers saw frequent and serious rises and floods; the damage they caused in a context of generalized cold seasons was considerable. However, crisis periods have luckily been broken up by dryer and warmer periods that have allowed the populations to grow again. We will seek evidence of the crisis from the Italian Apennines to Northern Europe via the Western Alps. There is no doubt that this crisis was under the primary influence of a damaged climate. This is attributed to the hydrological effects of the Little Ice Age* that affected harvests, brought about famines, caused high morbidity* and maybe even caused conflicts on a continent that otherwise saw a remarkable economic and intellectual boom. Climate is not everything, however, for the mountains were spaces of colonization and intense valorization over the centuries when the low agricultural productivity and the need for variety were the norm. The hydrological crisis of the Little Ice Age in the northern hemisphere stimulated both intellectual research and the boom in empirical knowledge. Chapter 2 deals with the slow conquest of hydraulic knowledge by the European powers, France, Italy and the Netherlands. The choice to use these countries as examples is partly guided by the fact that they have rivers that drain water and sediment from fragile mountains towards their deltas. Some basins are full-scale laboratories for the development of the erosive crisis that they experienced between the late Middle Ages and the 19th Century; they lend themselves remarkably to a cross-analysis of scientific progress. Italian engineers, with their mathematical training, were prematurely at the cutting edge in this domain in Europe because prosperous cities and their contado* directly confronted the risk and were to efficiently reduce their destructive effects to guarantee urban prosperity; France, on the other hand, had

xvi

Sedimentary Crisis at the Global Scale 1

more theoretical perspectives, primarily focused on calculations, most likely because wars and the need to make commerce prosper demanded other approaches, notably the construction of a network of canals1. We hope to show that European hydraulic science is constituted in the modern era and that it saw its apogee in the mid-19th Century. The notions of drainage basins* and upstream–downstream solidarity are derived therefrom and are, as we will observe, result including all the logic of the great erosive crisis from the Little Ice Age. A detailed understanding of the processes at work in drainage basins led to the deletion of some of the causes of the crisis, at least those that stemmed from human action. Thus, the mountains were protected from the effects of erosion, to the benefit of river mouths. In a context marked by instability, we feel that the excesses of the sky and waterways were feared by the affected societies. Mastery of the waters and land in motion was the obsession of societies in a state of crisis, an obsession that, incidentally, lasted well beyond the crisis itself in the mountain valleys. They are therefore places where the simple rise in the water level revives old fears and calls for arrangements that are the direct heritage of old practices. Mountains up to the sea, engineers and suffering populations have mobilized themselves to find means to escape misfortune. However, the river crisis, that of the risk of flood and that of damaged lands, of production, and of poverty, has also been a harbinger of collective progress. The solidarity between territories led to the appearance of the notion of the drainage basin, as well as the principle of the spatial uniqueness of water management. Never have industrializing agricultural societies better designed river management than in the 19th Century, fully aware of the connections between the mountains and the sea, like the interactions between slopes, rivers and deltas. Since the Neolithic agricultural revolution, which started in the Middle East and affected most of the regions of the globe, the continental balances have been affected and, in several large basins, the river imbalances were nearly contemporary to the construction of deltas that are hardly 6,000 years old. One powerful destabilization factor is the factor of agriculture: it involves clearing and scraping soil prepared for agriculture. Some specialists suggest that agriculture brought the Earth into a new geological period, the Anthropocene, characterized by formations of the Earth’s surface and geological formations produced by human action. This is a broad subject that does not lie in the direct perspectives of this work, but these goals largely overlap. Archeology and geoarcheology teach us a great deal about the chronology of the episodes of humans taking and losing influence at the surface of the land that has emerged, on the disappearance of protective forests, on the sedimentary flows that have overcome the oceans, built the floodplains, left proof of

1 Hydraulic science was not born in Western Europe, but in China and the Middle East; it is one of the facets of the Renaissance in Europe.

Introduction

xvii

human action in lakes and, thus, allowed this action to be quantified. Societies have been powerful destabilizing agents, to the extent that they profoundly changed the sediment balance around the globe and gained the upper hand over natural destabilization factors, at least during certain periods. The effects of several and long-lasting climatic crises superimposed themselves on a background of growing agricultural impacts – growth certainly unequaled in space and time – in certain periods of the humanized Holocene*. Periods of cold and heavy precipitation hit certain regions of the globe hard, notably in the northern hemisphere. Crises concerning production and even morbidity were accompanied by the loss of fertile land, materials being swept away by rivers that then rose while the adjacent plains were covered with marshes, the river dug into the earth and the deltas progressed. It is important to make a distinction between mountain erosion (concerning rocks and surface formations*) and the erosion of soil stricto sensu, which supports agriculture (even though the soil was the first to be eroded on the slopes of the Alps). The phenomenon of soil erosion is ubiquitous and we will only deal with a few examples of “accelerated” erosion. The change was brutal starting in the late 19th Century, a period in which a process of great sedimentary drying up concerning most rivers around the globe took place. To present this, it is necessary to understand, before and in light of the contemporary discoveries, the chain of processes that follow one another from the mountains to the sea. The numbers obtained through the extrapolation of values measured on small eroded surfaces in river basins are quite different from the values measured at the mouths of rivers. This is because drainage basins contain countless natural traps in which sediment transported by the waterways that drain them are deposited. The numbers confirm that agricultural development (as well as land abandonment) was indeed the major factor disturbing natural sedimentary balances, whether or not the climate played a magnifying role on weakened lands. New, powerful factors have appeared in the past 150 years, though. Around the world, the extraction of useful mineral resources is taking place in the very beds of rivers. It crosses channels, reduces the risk of floods and allows floodplains to be developed by agriculture, cities and industrial zones. However, the primary factor is the construction of large reservoirs that modify hydrology to the detriment of transportation and trap considerable quantities of sediment. The dam-reservoir has segmented the river space and led to the solidarities between upstream and downstream being forgotten. In Europe, the late 19th Century was a period of demographic decline in the mountains, spontaneous and planned reforestation, and the progressive mastery of erosion. With mountains whose erosion has largely been mastered through sections of rivers equipped with reservoirs, the river has lost the logic of upstream–downstream solidarity.

xviii

Sedimentary Crisis at the Global Scale 1

There are several models of river disturbance that share the general processes presented in this introduction, but they decline these specifically in the form of original geographic complexes influenced by the geology and topography of their drainage basin, by their climate and by the relative weight of the human pressures imposed upon them. We have chosen the Amazon as an example of a relatively undisturbed large river, even though the threat of large dams is becoming a reality. The situation of large rivers in China and Southeast Asia, where development is taking place at a rapid pace and producing brutal impacts, is another matter altogether. We have just left a century that saw the greatest breakthrough in the Holocene period. In the last millennia, which were those of the humanization of the Earth’s surface, reversibility was still present: instances of deciphering and climatic crises could destabilize the environment of river basins, but the crisis phases were followed by more or less long periods in which the previous balances were restored (or at least new balances, close or far from the previous, were put in place, the original state of reference not existing). Deltas have recorded the crises and remissions and progressed, taking each year as it came; there has been nothing of the sort in the last decades in which we have brought out a new world, that of delta penury. This first volume provides the foundation of the second, which will be dedicated to deltas.

1 The Torrential Crisis in the European Mountains (14th–19th Centuries)

1.1. Introductory generalities on global fluvial systems Fluvial systems function according to universal principles, governed by fluid mechanics, while remaining subject to regionalized constraints under climatic control; this provides partially distinct forms to northern, temperate and tropical rivers. In most matters and to stick to the circulation of water and sediment, the scientific literature distinguishes “sediment production” zones* (essentially localized in mountain and hill regions), “sediment transfer” zones* and “sedimentary deposit” zones downstream (alluvial plains, deltas, ocean margins). This longitudinal zonation is also governed all over the globe by secondary processes that nuance the general principle. First, river styles* are under the control of climate, vegetation that more or less protects slopes, and geology that conditions lithology and the ability to mobilize certain types of soft materials. If the balance or equilibrium between liquid flow (capable of ensuring sediment transport) and solid flow (sediment to be transported) is in favor of the former, the material is easily evacuated and the river presents a simple morphology, demonstrating a single winding channel, sometimes classified as a meandering* channel. These channels are found in regions that produce little sediment (mountains and hills with heavy rainfall, wooded, sparsely populated) and in the transfer zone. If, on the contrary, the liquid/solid fluxes balance is in favor of the latter, then the flows are not capable of transporting the entire load originating from the

Sedimentary Crisis at the Global Scale 1: Large Rivers, from Abundance to Scarcity, First Edition. Jean-Paul Bravard. © ISTE Ltd 2019. Published by ISTE Ltd and John Wiley & Sons, Inc.

2

Sedimentary Crisis at the Global Scale 1

production zone. This situation is very frequently seen in regions with semi-arid climates (the slopes there are badly protected), in regions of the globe with a contrasted relief and a fragile structure, and finally in elevated areas that have been cleared and over-exploited for pastures and agriculture (Figure 1.1).

Figure 1.1. The Selle torrent in the Massif des Écrins (France); it is fed here by an avalanche cone whose base is eroded by lateral erosion from the torrent. The bed comes loose downstream and adopts a braid style (source: J.-P. Bravard)

The production zone and sometimes even the transit zone are then saturated with rough material; in response, the river channel adopts a particular shape and mode of functioning, braiding*. The braid style is classically contrasted with the meander style, even though composite or hybrid styles are frequently seen. The excess sediment therefore has a descriptor, or marker, which is braiding, easy to diagnose and interpret based on the dynamic function of the basin. Without anticipating too much further, we can assume that braided channels reveal crises, which are the subject of the first chapters of this work.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

3

Figure 1.2. A river in British Colombia, the Squamish River. This wandering gravelbed river*, with a basin well provided in both water and sediment, meanders and braids at the same time in a former glacial basin [HIC 75] (source: J.-P. Bravard)

Some further questions have join those stated above, at the forefront of scientific investment. These concern: 1) the precise methods of mobilizing material on slopes until they enter channels (this is slope–river bed coupling* or “decoupling”); 2) revealing permanent deposit sites at the foot of eroded slopes far from active channels; these are long-lasting, for they are sheltered from erosion by channels (we generally speak of “colluvia”* to distinguish them from alluvia); 3) the importance of alluvial plains as deposit sites for alluvia brought when channels flood; the question also needs to be asked of these materials being recovered when the river channels move on the plain and recover these deposits to incorporate them into its load; 4) the chronology of sedimentary recovery that can ensure a transfer to the system’s traps downstream, even during phases with limited material entry into rivers from production zones. In short, the process chain involving production– transfer–deposit and the morphological “continuum”* involving mountain–piedmont

4

Sedimentary Crisis at the Global Scale 1

plain–valley are not sufficient to correctly grasp the load of a river that is building its delta. Finally, let us observe that the globe’s rivers, aside from increasingly rare exceptions, have stopped presenting the “pure” mechanisms, that were briefly described, for more than a century (sometimes longer). Sedimentary traps, the complexity of which was just explained, have moved to the background, behind the reality of artificial traps constituted by dams-reservoirs, the number and capacity of which exploded across the globe in the 20th Century. However, let us move to the heart of the subject of this first chapter, which is the sedimentary crisis of the Little Ice Age (abbreviated as LIA). The knowledge we have of the LIA is progressing in the fields of both hydrology and sediment transfer and deposit, not to exclude the mechanisms that connect these two components. Let us take the example of hydrology. A very in-depth study of the archives of cities on the Lower Rhone has provided great insight into the hydrological rhythm of the river at the outlet of its basin [PIC 14b]. The study of the floods on the Rhône, spread across decades and four degrees of gravity based on their manifestation in the riverbed and in its floodplain, revealed new elements, particularly their classification into two hydrological “hyperphases,” the first dating back to 1450–1599 and the second to 1647–1711. The hyperphases are separated from one another by a period of moderate hydrology (1600–1646). On an even finer level, the decade 1701–1710 was the hardest in the history of the river, and before it, the period from 1481 to 1500. During hyperphase 1, an isolated event, the 1548 flood, would surpass even that of 1856, despite it being considered the most significant flood in the Rhône’s history, i.e. the absolute “reference” for risk managers in France. Another discovery is the way in which the contemporary hydrological regime was established, with boosts in strong hydraulicity in the decades 1770–1780, 1801–1810 and 1841–1850. If the floods in 1840 and 1856 do not belong to these sequences, it is because these are isolated manifestations on a foundation of weak hydraulicity (Figure 1.3). This figure is one of the markers of the river’s new hydrology according to the authors of the study, G. Pichard and E. Roucaute. It is possible to evoke the reality and the gravity of the long crisis of the Little Ice Age (LIA), because the proofs of highly active processes that affected the torrents and torrential rivers are numerous enough to be convincing. These manifestations allow us to understand the extent of the means used by the inhabitants of slopes and floodable plains to find initially local and provisional solutions, then more or less definitive ones, occasionally benefitting from the help of the public authorities. We will consider the question of the torrential crisis from the LIA from the perspective of several European mountain ranges.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

5

Figure 1.3. Hydrological hyperphases (black curve) and flood periods (red curve) from the 14th to the 20th Century. The gray line emphasizes the LIA period (source: [PIC 14b]). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

1.2. Manifestations of the LIA crisis in the river valleys of Western Europe 1.2.1. Mountain crises Historians have collected testimonies in the high valleys of the Western Alps. Two works are interesting insofar as they pose the question of a change in the torrential and river landscapes during the Middle Ages based on texts and in little-known terms, without being able to reference the later notion of the Little Ice Age, which did not yet exist when they were written. The medievalist historian T. Sclaffert [SCL 26, SCL 59] observes that Seyssins, located near Grenoble, had a port on the Drac, from which, curiously, boats departed, whereas (later) iconography shows us that the Drac was a torrent congested with pebbles in the modern era. The historian hardly goes beyond the 14th Century to find torrential activity in the Northern and Southern Alps, but the crisis (in the broad sense) is indeed present. T. Sclaffert describes valley floors covered in fields, orchards and prairies on riverbanks; later, the rivers become torrential, capable of ravaging the valley floors during floods. The agricultural wealth went along with the existence of numerous mills and “artifices” that have since disappeared. This research provides raw facts that are all the more interesting in that they had no explanatory objective and that no manifestation of the change in landscape is sought to support a theory that did not yet exist.

6

Sedimentary Crisis at the Global Scale 1

Here, we will use the great work by forester Paul Mougin (1866–1939). It certainly comes after the laws voted on in France to restore mountain territories (1860 and 1882), but it provides a very interesting historical overview of the causes of deforestation in Savoy. P. Mougin [MOU 14] adopts the standpoint of his original post when he positively emphasizes the foresight of the mountains’ inhabitants (both monks and farmers) but negatively emphasizes their “spirit of profit and plunder” to explain elevated clearings and illegal deforestation as well as pastureland abuses in the forest, not to mention supplying mines, salt manufacturing and glass factories with lumber and fuel, or even the destructive effects of wars. The degradation of forests seems to be attested to in the Ancien Régime in France (1589–1789), but it is not possible to create a hierarchy of these factors. The author dates the first measures to ban deforestation back to the decisions of the Senate in Savoy (the first in 1559, renewed in 1594, 1654, 1666 and 1667); these decisions went into effect through a royal decree in 1723 applicable in the Duchy of Savoy, then by the Royal Constitutions of 1729 and 1770. P. Mougin accuses the confiscation of religious and noble property, following the annexation of the Sardinian province in 1792, of having allowed a loss of all control over the forests to exclusively favor the communes and their inhabitants; the end of the Empire (1815) is also a period considered by P. Mougin to be harmful for afforestation. The period of Sardinian restoration (1816–1860) did not bring about improvements, despite the restoration of the “royal regulations”. For our purposes, it is interesting to see how the Sardinian legislation dealt with the question of defending against torrential rivers. The Royal Constitutions from 1792 were meant to promote measures limiting “corrosion” and the meandering of rivers: it was forbidden to burn and uproot trees in a stretch measuring approximately 6 m wide, and it was necessary to plant them where there were none. The communes and riverside landowners were responsible for this work. A regulation from 1739 states that commune administrators can impose repair work for “washed away or damaged” river bottoms upon riverside inhabitants (at their cost!); this meant facilitating damaged lands to be farmed again in the interest of their owners and the tax authorities. These measures, reestablished in 1816 when Savoy returned to the kingdom of Piedmont-Sardinia, would remain in effect until 1860. It is surprising that these measures only concerned “rivers and torrents”, actually torrential rivers, without observing torrents per se, i.e. the sources of the denounced damage. One hypothesis would be that the steep slopes leading to the primary branches of the hydrological network provided too little return to be worth attention.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

7

There are also some testimonies – today we would classify them as “geoarcheological” – concerning the medieval and modern eras. One inhabitant of the Barcelonnette region provided a first-hand account dating back to the period of extensive forest restoration works. He described thick arable soil underneath the torrential blocks, free of stones, that was “certainly cultivated”: this soil contained tools and held a rock wall and poplar trunks in place. Fossilized under thick alluvial fans from the torrential cone of the Riou Bourdou measuring 2–8 m, this agricultural landscape dated back to the early 15th Century, at the latest. Without citing every element published by a wide range of authors, there is abundant proof of a significant rise in rivers due to coarse alluvium deposition: in the bed of the Giffre (Haute-Savoie), where the poles carved in the late 13th Century and sunk into a marsh are fossilized under pebbles [PEI 86]. In the Chautagne marsh, at the edge of the Rhône in Savoy, peat stopped forming around 1170±140 A.D. due to the massive arrival of overflowing silt deposited by the river’s flooding [BRA 87]. An attempt to map the first manifestations of alpine torrentiality based on the published literature provided us with the dates included in the range of 1336– 1471, but it goes without saying that such a map should be updated using new intensive research in the medieval archives. It would most likely differ from the prototype, with earlier beginnings and thus fewer dates dating back to the 15th Century. 1.2.2. River crises and metamorphoses of the Drac and the Isère in Grenoble In the 1920s, a pioneering work was written by the engineer Bouchayer regarding the lower Drac [BOU 25]. He showed that the floods in 1277, 1373–1377 and possibly 1414 did not change the morphology of the Drac, but rather that the numerous floods occurring between the late 15th Century and 1675, which we know about because Grenoble’s city defenses were directly affected before the works organized by Colbert starting in 1683, belong to a long series of floods in the city and stronghold of Grenoble (Figure 1.4). Furthermore, measures were taken against deforestation in the basin starting in 1565, because this was assumed to be responsible for torrentiality. Finally, the dykes on the Drac were extended to the foot of the Chartreuse Mountains in 1771–1782 in order to limit the damming effect produced on flooding Isère (see below) [SAL 91]. This research, using somewhat obsolete methods, was the first in France to pose the question of variations in the hydrology, solid transport and morphology of an important torrential river.

8

Sedimentary Crisis at the Global Scale 1

Figure 1.4. Map of the damage caused on the outskirts of Grenoble by the flooding of the Drac on October 5, 1616. The Drac braided through the pebble banks and threatened Grenoble’s fortifications (source: © Grenoble city archives)

The most convincing example is that of the Isère, upstream from Grenoble. This large river, which receives the Drac immediately downstream from Grenoble, does not seem to have been affected by floods before 1469 [CŒU 03], and the convergence of its flood with that of the Drac is recorded in 1524–1525. The water levels of the Isère (and that of the floods in the city of Grenoble) are affected by the hydraulic damming exercised by the Drac (1604). The writings of engineer Pierre

The Torrential Crisis in the European Mountains (14th–19th Centuries)

9

Rolland (1741) and his son Jacques (1787), as well as a map of the course of the Isère (Carte du cours de l’Isère; referenced in Figure 1.5), describe not only the flooding space [BRA 10a] but also the braiding downstream progression of the Isère in the Grésivaudan valley, to the detriment of Grenoble’s reach meanders, which were cut off. Also described are the filling of the bed and the reduction of the hydraulic capacity* at the crossing in Grenoble, as well as the aggravation of the overflow and speeds on the alluvial plain. P. Rolland observes that before the cutoff of the meander in Gières in 1731 (C), the riverside plains along the Isère were submerged: “[…] long before the territory of Grenoble had suffered the slightest overflow… whereas the plain and the territory of Grenoble, since the era of this cut-off, were covered with water long before those from the parishes below” (1741).

C

Figure 1.5. The Isère and the Drac in Grenoble in 1741. The Isère meanders and flows to the west (left); the Drac converges with the Isère downstream from the city (the Drac is dammed, but forms an alluvial fan in the northwest). The green area represents the spread of the flood on December 21, 1740; the outlet channel project bypassing Grenoble is represented (source: © Departmental archives of Isère). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

10

Sedimentary Crisis at the Global Scale 1

In an essay dating back to 1787, J. Rolland observes that: “ […] the floods… pouring over the plain where currents have been established by the height of the overflow fill the old passages with deposits of gravel and open new ones through the best grounds… The river destroying its banks enlarges its bed, divides its waters, and thus no longer has enough strength to carry the matter that the torrents deposit there, its bed rises to the point that the slightest floods will soon form an overflow there.” As for Grenoble: “[…] the Isère crosses it through a channel that can only transport about half the volume of water that it provides during floods, such that the insufficiency of the outlet produces cataracts and currents so large in the streets and around the perimeter of the city that there has always been scouring following which the walls were seen falling, the plantations ripped up, and the buildings collapsing…” This causes “humidity and deadly diseases”. The engineer J. Rolland, beyond the simple repetition of the floods and their effects, analyzed river metamorphosis* before his time, although without conceiving the relationship between the changes occurring on the Drac (the earliest of these) and on the Isère afterwards. The time gap between the reaction of one river or another is now explained by the faster spread of alluvia from the Drac downstream due to the proximity of the sedimentary sources from the upper basin, the very steep slope of the river and the limited storage capacity of the gorge sections. The two studies of the case of the Drac and the Isère are even more remarkable given that they are located at the large confluence in Grenoble. The example of the Isère in Grenoble shows the extent to which the knowledge of rivers was concrete; the large mechanisms were well understood and on the right scale in the late 18th Century. The engineers in the 18th Century did not conceptualize, but they acted empirically and intelligently. Today, we know that the recent concept of “river metamorphosis” applied to these alpine rivers where it had been understood. The Southern Alps were the subject of fewer works, but the testimonies collected by C. de Ribbe [RIB 57] come together to describe rising rivers. According to a manuscript from before the French Revolution, the Bléone rose three feet at the bridge of Digne; other cases of rapid rising are indicated in the Issole, the Asse, the Durance or even the Ubaye in Barcelonnette. Recent works performed on small territories confirm the state of advanced degradation of rural spaces.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

11

1.2.3. Flooded piedmont plains in Switzerland Grenoble’s situation in the 18th Century is particular in that the valley floors, formed by glaciers, are broad enough there to store alluvial sediment and waters. This is also the case in Switzerland, when torrential rivers in the Bernese Alps, loaded with gravel and murky waters, took material over once-flat glacial valley floors occupied by lakes and drained by rivers deprived of the energy necessary to evacuate the torrential material. The diversion of Swiss rivers from the 15th to the 19th Centuries demonstrates the intensity of the conflict between the humanization of the valley floors through clean-up and the growing constraint exercised by the torrents originating in the nearby mountains. Work has been done according to the possibilities, with no particular model (but possibly the Brenta, which drains Veneto, serves as an example) [SCH 92]. The first, modest deviation of a Swiss river took place starting in the 15th Century to protect Buochs, a village threatened by the Engelberger Aa, a tributary of Lake Lucerne, upstream from Lucerne. The first important correction is located in southeastern Bern. A tributary of Lake Thun, approximately 30 km southeast of Bern, the Kander was born of a glacier under a 3,650 m summit; its heavily loaded floods joined the Semme and flooded the low alluvial plain west of Thun, threatening to flow into the lake and block the course of the Aar downstream from the city. In 1711–1714, under the impulse and management of soldier and surveyor Samuel Bodmer (1652–1724), then Samuel Jenner, the Kander was diverted into two parallel tunnels dug into the rock of the secondary range in Strättligen using rough tools; this work then allows it to flow directly into Lake Thun, thereby relieving the city. The tunnels having collapsed, the river dug its bed in the rock to a depth of 40 m. Regressive erosion up to a depth of 5 m resulted from this; it spanned a length of 4–5 km, bringing about a loose alluvial substrate. Similarly, a delta composed of sand and gravel formed in Lake Thun, with a depth of approximately 60 meters. Nothing was truly regulated by these works, for the heavy inflow raised the level of the lake and threatened the banks due to a failure to enlarge the lake’s outlet [VIS 11]. The last example that we will use is the correction of the Linth, planned in 1784 and successfully executed between 1807 and 1816 according to a plan by the Badish engineer von Tulla, the same man who corrected the Rhine. The plain located between the outlet of Lake Walen and that of Zurich was occupied by a broad river, the Linth, whose floods caused great damage. Under the authority of the engineer Escher, the Linth was the subject of double channelization, one cutting through the river braids over a length of 15 km, the other an upstream derivation through Lake Walen (called the “Escher Canal”). These works were done using piles of fascines capable of holding the sediment percolating through the branches and creating a new

12

Sedimentary Crisis at the Global Scale 1

bank downstream step by step. This work was meant to calm the Linth’s flooding, before the plain was crossed by the Linth canal, leading to Lake Zurich. This is a great example of hydraulic protection, even though it required several decades of effort to obtain very high efficiency. 1.2.4. Sedimentation and large works in Italy It is undoubtedly in Italy that the problems of torrentiality were the most severe, possibly because the geographical configuration juxtaposes steep mountains (that are also densely populated and thus fragile) and low land with uncertain drainage. It is not excessive to make a comparison between certain alpine configurations and that of the Appennino Tosco-Emiliano, where the torrential rivers drain the mountain towards the Po plain on the one hand and the Tuscan plane on the other hand, the Arno serving as a primary drain there. In Florence, the rise in the Arno’s bed between the Saint Nicolas and Toussaints dykes became concerning in the 16th Century [FRI 74]. Viviani shows that the process takes place up to the sea and is due to the convergence of multiple factors acting in a cumulative way: a context of sedimentary overabundance (and rise) on the tributary branches draining the Apennine, notably on the Bisenzio and the Ombrone; the loss of head caused by mill thresholds; and finally, the narrowing of the bed caused by the arches of bridges. As an example, the large palace of the Uffizzi in Florence had been built in 1560 by Vasari; it had most likely been established above the level of the heavy floods, during an era when rising water levels were still limited, but in 1677, during the restoration of the palace’s foundations, Viviani had the low windows of the stables, which water and mud could enter, partially bricked up. In the late 16th Century, the projects to correct the Arno aimed to no longer leave “such a beautiful, rich and magnificent” city exposed (Figure 1.6). These projects can be divided into three categories: 1) by-passing Florence by digging a new flood bed, fed by a spillway, but with the risk connected to excessive speeds; 2) building up the Arno’s parapets in the city by closing all openings, including the drains that would be transferred downstream; it is true that this choice would have caused a “loss of the beautiful view over the Arno” and would have made it “an entire river seemingly suspended in the air between two walls”; 3) lowering the Arno’s bed as it crossed Florence by eliminating the works raising the shoreline in the city while preserving the large dyke upstream to catch gravel.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

13

In the late 17th Century and after great procrastination, the policy adopted to protect Florence was to catch the alluvia on the mountain. Torrents like the Nievole, the Serchio or the Era were outfitted with “dykes”, essentially small river dams built to block rocks and gravel; these works allowed sediment to pass, however, and Viviani recommended constantly raising all the Arno’s tributaries as well as the bed of the Arno itself, all while reforesting the tributaries’ basins. This example shows that the defense of Florence, initially planned in the 16th Century, adopted in the late 17th Century the principles that have become classics for combatting torrents and, further downstream, classic for protecting against the flooding of torrential rivers. Disputes are therefore lively between the followers of sedimentary retention and those who doubt the ability of dams to efficiently catch material. The policy adopted in Florence in the early 18th Century confirms the control of the Arno’s tributary torrents, but the opposite policy was adopted for the river, in order to improve the conditions for navigation and conquering lands. At great cost, the Arno on both sides of Florence is narrowed between the longitudinal dykes and straightened to reduce its length by 7 or 8 km, at the expense of broad sinuosities earlier on. The result of these works is a several-meter rise in the bed between the dykes and a descent of rocks and gravel further downstream, where they fill a large section that previously had always possessed a sandy bed (Figure 1.6). This policy of cutting sinuosities, pursued downstream between 1740 and 1760 to protect Pisa from the flooding Arno, was harshly condemned by the engineer P. Frisi [FRI 62] on the grounds that it was inefficient and harmful.

Figure 1.6. The Arno crossing Florence, looking downstream. This “view” of Florence shows the large transversal dyke retaining sediment and the extension of the river immediately upstream from the city; this is due to the erosion of the natural bank and significant deposits (source: Fondazione Cassa di Risparmio, Florence)

14

Sedimentary Crisis at the Global Scale 1

In Emilia-Romagna, the question of alluvium accretion on the Po and its tributaries originating in the Apennines was more serious than in Florence. We will see the way in which the disputes between cities concerning the lower reach of the Reno (this is the “Reno question”) were a factor in the progress of Italian hydraulic science (Chapter 2). In the late 18th Century, the bed of the Reno and that of its tributary, the Sammoggia, rose on the gravel descending from the mountain and progressed ever further, as on the Arno. The Po’s plain became increasingly marshy between Bologna and Ferrare. There are several competing drainage “lines”, from the direct route to the Po Grande, starting towards Ferrare in 1522 and sought after by the Este family who defended the interests of the city of Bologna, to routes leading to the Comacchio lagoon, borrowing the route of old branches of the Po. The definitive route (the cavo Benedettino) was created under Pope Benedict XIV in the 1760s. The channel combines several functions: directing the Reno’s waters, navigation and on-land deposits. It adopted an artificial route going through Argenta and leading to the Po Morte di Primaro before reaching the Adriatic Sea. Experts who were against the digging of a new straight bed for the Reno towards the Po Grande had claimed that the arrival of gravel from the mountain would make its bed sinuous and rise and threaten the dykes and the surrounding countryside located further down. This was actually the case during the 1648 rupture of the dykes at Cento, after which the plain was ravaged by the flood [LUG 13]. Another “classical” region regarding dangers connected to fluvial alluvium accretion is that of Venice, i.e. the lagoon, port and grau* through the lido* and thus “la Serenissima” itself, then considered to be threatened by death. The Venice lagoon appeared around 7000–6000 BP*, when the sea level stabilized to be around the current level. The regional coastal rivers brought the sand that the long-shore drift* modeled into an open lido through passes. The site in Venice was chosen in the 5th Century AD due to the protection that the island could ensure in the face of threats coming from both the sea and the continent. To ensure its longevity in the face of river backfilling caused by the coastal tributaries (the most important of which was the Brenta), the lagoon could hardly count solely on the slow tectonic subsidence* experienced by the Po’s low plain, to which the compaction of Holocene sediment deposited since the rise in the sea level is added. The Brenta is a 160 km coastal river that starts in Trentino, drains the Valsugana and leaves the Alps in Bassano; today, it is adopting a piedmont plain route that leads it to the Adriatic Sea south of Chioggia, avoiding the Venice lagoon to the east [BON 03]. A good geomorphological map reveals the spatial scope of the immense alluvial fan of the Brenta, which was noted thanks to old channels still filled with water; they are occupied by the phreatic flows from large sources originating at the edge of Pleistocene gravel from the high plain and fine sediment from the low Holocene plain. The river, random in its movements, has occupied the territory

The Torrential Crisis in the European Mountains (14th–19th Centuries)

15

between the fans of the Sile and the Piave to the northeast and the Euganean Hills and the fan of the Adige to the south [BON 08]. The successive channels of the Brenta River, dated Roman and Medieval, naturally fed the Venice lagoon, where river levees were observed and properly dated. The last natural branch (the Medoacus Major in texts) reveals a change in the Brenta on the northern blank of its fan in the 6th Century AD, possibly during catastrophic floods that occurred in the fall of 589, according to the chronicle written by a contemporary. In the Middle Ages, waterways played a remarkable role in the political, military and economic histories of the large cities of Veneto. In the 14th Century, Venice, a rival of Padua and Treviso, became not only a maritime power but also a land power. The density of the hydrographic network played an essential role in ensuring the economic prosperity of these cities, and Venice could legitimately claim that controlling the water was vital for it, as it was for the prosperity of Padua. The Brenta provided Padua’s economic outlet and allowed it to construct its contado; it also constituted the hinterland* of Venice, which the river was to provide with food and wood flowing downstream. The Brenta also provided Venice with potable water endowed with excellent organoleptic qualities complementing the scarce water supply provided by polluted wells below city plazas to drain rain waters; in 1425, the Brenta became the only source of water authorized by the city magistrates. If the little river brought prosperity to Venice, it was also the cause of its decline. The city always had to defend itself, but “the most formidable danger was not only the sea and the men it brought, but also powerful, disorganized forces, those of nature, which dominated the continental water, i.e. rivers” [BEV 95]. In 1141, the Paduans rerouted the Brenta directly towards the lagoon using the old route, the branch of the Medoacus Major [RIP 03]. However, the river naturally leads to Venice, which is thus threatened by the construction of an interior delta, by the deposition of sand at this site and by the closing off of access to the sea provided by the Canale Orfano through the lido. To ward off these threats weighing over the city, the Medoacus Major was closed in 1191 by the Venetians so that the Brenta would flow a bit further to the south. The hydraulic story is then one of debates and conflicts to remove the mouth of the Brenta from the Venice lagoon by constantly moving it further to the south, until the Brenta nova was opened in 1540. By proposing a new route for the river, the engineer Marco Cornaro (1412–1464) had thought up several radical solutions meant to stop the accumulation of sand and the paludification* of the lagoon at a time when the excessive deforestation in the Alps was already being denounced. This rerouting risked modifying the lagoon’s equilibrium, interrupting navigation and stopping mills and the provision of potable water. Debates were heated, and in 1501, the Republic’s Council of Ten named a Magistrate of Water; in 1505, he established a Collegio Solenne delle Acque. These measures alone speak to the extreme importance granted to water management. The final decision would be made to ensure the rerouting of the lagoon to the south to

16

Sedimentary Crisis at the Global Scale 1

make it operational in 1610. In 1613, to reduce floods and due to the insufficient size of the Brenta nova, a water collection channel (the Taglio nuovissimo), running north–south, was dug between the Brenta nova and the lagoon [BON 12]. A similar, though somewhat more modest story is that of the two other coastal rivers, the Sile and the Piave, that silted up massively in the northern part of the Venice lagoon during the great flood of 1533 [BON 04]. In the 16th Century, the engineers of the Republic of Venice forced the Piave and the Sile rivers to flow to the northeast, into the “old” Piave. In the end, Venice’s obsession with protecting itself from river contributions was translated by large works to push the threat of the Po to the south. The downstream course of the large river actually shifted to the north and threatened to capture the Adige and to fill the port of Chioggia. In 1600– 1604, Venice rerouted the Po di Tramontana towards the south into the Taglio di Porto Viro that had just been dug. At the end of these sometimes-lively debates, an exceptional corpus of hydraulic knowledge concerning the balance between rivers, the lagoon, the wind and the sea ended up being created, taking into consideration the currents of fresh water and the pressure of salt water. Modern water science was established in Venice, as well as other parts of Italy, but like nowhere else in the Western world. As a partial conclusion concerning mountain regions, the long erosive and river crisis of the Little Ice Age greatly affected Italy starting in the 16th Century and lasting until the mid-19th Century. France was prematurely affected, but the convincing testimonies are only significant in the 17th Century and particularly in the 18th Century. The Mediterranean and Apennine mountain climatic background and high human densities combined to fragilize these mountains. One of the favorite subjects of current research is actually attempting to take human and climatic factors into account for the onset of erosive crises across territories. The range of knowledge in the modern era never mentions the hypothesis of aggravated precipitation nor even natural factors encouraging erosion. Pragmatic solutions dealing with the anthropic causes of erosion result therefrom, but they would be no different if natural causality had played an effective role recognized as such. In any case, the policy to protect deforested slopes and correct torrents began actively in the late 16th Century in Tuscany, whereas nothing was to be observed in France or in the field (but this would need to be verified with precision, and the discoveries, if they are made one day, will likely be limited) nor in the scientific literature that is well identified. Jean-Antoine Fabre is most likely the first French scientist to tackle the question of torrentiality, both theoretically and practically, if we trust his 1797 treatise. His work comes more than a century after that of Tuscan

The Torrential Crisis in the European Mountains (14th–19th Centuries)

17

Vincenzo Viviani (Chapter 2). Italy was certainly at the forefront, but it is very clear that its cities, in a position of both political and commercial powers, had that ability to kindle the scientific spirit and mobilize it rapidly and intensely; a dense network of wealthy cities located on piedmont plains and near torrential waterways could attract the benefits of active patronage. Without excluding the ability of these cities to react, the precocity of their responses to crises on the Po’s plain and in Tuscany could also be an indication of the premature appearance of serious difficulties that could have made “systematization” necessary on the part of these cities; this was also true, at the level of the drainage basins, since the recommended and implemented measures extend from the high basins to the river, sometimes until the mouth, as can be seen from the example of the Arno. It is at the level of the mountain basin that the constraints are mastered and at the level of the river basin that we can hope to find the responses to the question of floods, drainage and improvement in a period characterized by a growing need for wheat. 1.3. The difficult mastery of the Rhine delta in the modern era Far from the Alps, whose sediments originating from the alpine Rhine are caught in Lake Constance or drained by the Bavarian upper Danube, the Rhine discharges the waters from a basin measuring 198,000 km2; as for the Meuse, it drains a surface area of 36,000 km2. A relief of hills, plateaus and old mountains provides a diversified load that feeds an active delta. The low valleys of the Rhine and the Meuse in their common delta somewhat resemble the fluvial Veneto in that the gentle slopes and high average discharges accommodated heavy floods and significant solid fluxes during the Little Ice Age. We will examine the way in which the dynamics of this climatic period manifested themselves concretely in the preparation of the fluxes into this delta over the last centuries. 1.3.1. Flow distribution between river branches: an age-old battle against the elements of nature The river network in the Netherlands has been organized since the Middle Ages; elevated dykes were built between 1050 and 1350 to set the route of river branches and protect the delta plain against floods. Nothing was gained, however. St. Elizabeth’s flood took place on November 18 and 19, 1421, in the regions of Holland and Zeeland, particularly in the Biesbosch, the space located between the Waal and the Meuse, moving through the city of Dordrecht. A storm that built up over the North Sea broke the poorly maintained dykes surrounding the polders, ravaged

18

Sedimentary Crisis at the Global Scale 1

dozens of villages and caused the death of thousands of inhabitants all the way up the Waal estuary (the downstream Rhine) affected by the storm surge (Figure 1.7). Although polderized, the Biesbosch remained a wetland after this catastrophe, with 7000 ha of artificial lakes having come into contact with the rising freshwater during high tide until 1970. The Biesbosch has been one of the most beautiful national parks in the Netherlands since 1994.

Figure 1.7. Extract from a map of Holland representing the downstream course of the Rhine and the Meuse. Drawn by Daniel de la Feuille in 1706, it was republished in 1747 by Paul de la Feuille in J. Ratelband’s atlas. The large lake is the Biesbosch, created by the bursting of the dykes on the Rhine in 1421 (source: Wikimedia Commons)

Since the works of the Medieval era, the course of the Rhine and its branches has been fixed without ever being secured again. Downstream from the city of Lobith, quite close to the modern-day border with Germany, the Rhine extends through its major branch, the Waal, below the village of Nijmegen. Another branch of the Rhine, parallel to the Waal but 10–20 km to the north, takes the name “Nederrijn” (lower Rhine) or “Lek” once it has passed Arnhem; on its right, it has lost the branch of the Ijssel that flows to the north towards the Ijsselmeer. As for the low Meuse, its course is parallel to the Waal, with which its bed is locally anastomosed. One hydraulic difficulty has shown itself through the centuries, primarily in the 16th Century. This is the relative reinforcement of the flow through the Waal to the detriment of the secondary branches, the Waal having captured 90% of the water

The Torrential Crisis in the European Mountains (14th–19th Centuries)

19

and the other distributary channels only 10%. The strong flow of the overflowing Waal threatened the country with floods, whereas the impoverished branches, notably the Ijssel, saw sedimentary accumulation, the current’s low speed no longer allowing the material to be evacuated. It was in 1707 that large works were performed to enlarge the stretch between the Waal and the bifurcation of the Ijssel in order to reinforce the flow leading to the Ijssel, thus to restore the balance of the flow distribution. This was performed through the digging of the Pannerdensch channel, in the bed of the river branch. However, this was also followed (rather logically) by the modification of the Nederrijn, which also reacted to the increase of the flow from the Waal. Next came ruptures of the dykes and serious floods, great instability of the stretches of the river that underwent sedimentation and incessant works to reconfigure the branches and bifurcations. In particular, low dykes and movable partitions were built to keep the low floods in the river channels. In 1745, the goal was set to give two-thirds of the Rhine’s water to the Waal, one-third remaining for the Pannerdensch channel and then for the Ijssel and the Nederrijn. This was a long and difficult task [HES 02b]. In 1753, the Dutch Academy of the Sciences in Haarlem proposed responses to the following questions: “To what extent have the rivers in the Netherlands risen on their sediment since the turn of the century? By what means can we remove the deposits of sand and mud formed in the river beds and how can we prevent sedimentation?”  [MID 97] These questions show that the problem was very well stated. The later works were crowned an empirical success, which was probably enough for their promoters. The desired result was achieved in the mid-1780s thanks to the placement of a new generation of spurs allowing another water partition, and “normalization” was also obtained in the mid-19th Century, after which harnessing became systematic and led to the definitive fixation of the river branches. 1.3.2. Returns on destabilization Recent research has allowed us to theoretically understand the profound causes of the changes recorded since the Middle Ages and the reasons for the recorded success. The medieval dykes, with their increased height, had not modified the river environment, but the Waal tended to slowly increase its part of the water and sediment flow, while the secondary branches were progressively blocked. The work performed between 1350 and 1707 returned balance to the flow distribution between the branches of the Rhine through the artificial reduction of the width of the branches. The consequence of this heavy modification of the hydraulic geometry

20

Sedimentary Crisis at the Global Scale 1

was the modification of the energy applied to the floor of the channels through surface unity, particularly in the annex branches. This work was not sustainable, for the same causes produced the same effects: despite the digging of the Pannerdensch channel, which was in use for a few decades, the Waal maintained its supremacy and its relative excess of energy restored the imbalance to the detriment of the secondary branches, endlessly subject to excessive sedimentation. As we saw above, it is the complete harnessing of the system that allows us to maintain a totally artificial balance, capable of balancing out the natural tendency to create a single bed through the Netherlands [HES 02a]. 1.4. Observations on the torrentiality of the Southern Alps in the late 18th and 19th Centuries Until recently, the French Southern Alps had been one of the chosen lands for manifestations of torrential erosion in Europe. They owe this to natural characteristics like the abundance of outcrops made of fragile marly rocks, tectonics disrupted by orogeny and, of course, the characteristics of the Mediterranean climate with strong autumnal water vapor, not forgetting the melting of mountain snow in the spring. These natural traits do not sufficiently explain unbridled torrentiality. In fact, it is best to consider the high human density of these mountains, as well as the methods of exploitation that made the slopes valuable, both for agricultural plowing as multiple forms of pastures. In this regard, transhumance, which involved welcoming large herds from the Provence lowlands in the summer, must have contributed considerably to the weakening of elevated lands leased by these communities. We have decided to consider the way in which the contemporaries, who were eyewitnesses, tackled this question. 1.4.1. A highly degraded state of affairs in the late 18th Century Thanks to Charles de Ribbe (1827–1899), we have a precious and most likely trustworthy work, written in 1857 by a legal expert from Provence [RIB 57]. “Attaching a singular value to tradition on taste and principle”, as he himself maintains, Charles de Ribbe, lawyer in the Imperial Court of Aix and legal historian, familiar with the archives of Provence, was the archetypical conservative legal expert in his youth, partisan to a coercive policy concerning the abuse practiced in the mountains of the Southern Alps. At the end of his life, he had moved towards the social ideas of engineer and sociologist Frédéric Le Play. Struck by the 1856 Rhône floods, the Durance and its tributaries, C. de Ribbe recounts, in detail, the story of the relationships between the mountain communities and the law under the Ancien

The Torrential Crisis in the European Mountains (14th–19th Centuries)

21

Régime, in Provence, which he called the “natural fatherland of torrents” (Figure 1.8). These abuses were followed by the absolute prohibition of deforestation, clearing and thinning or slash-and-burn*, ancestral practices that were motivated by a search for land to plant wheat and that were held responsible for the erosion of the thin, arid soil of the alpine slopes. The resistance of a population that considered itself impoverished and the calamities due to torrents hindered the strict application of the law; today, we would call these “humanitarian reasons”. The most serious cases were, according to Charles de Ribbe, those of the villages of the SaintLaurent-du-Var and Gréolières. However, the law had its say with the royal decree from April 12 1767 authorizing, clearing and even granting premiums; we can also mention the tolerance allowed by the law from Floréal 9, year XI1, the source of the accusations of laxness made following this period of French history. Charles de Ribbe finally denounced the later insufficiency of the 1827 Forest Code, which gave full liberty to the owners to use and abuse their forests both to collect wood and for pastures and burning. The opinion of this legal expert, who was certainly not an 18th-Century man, but who was capable of understanding the century according to the law that he had put in place, provided a counterpoint that was indispensable for analyses by naturalist foresters. During the French Revolution, engineer Jean-Antoine Fabre dedicated several pages of a treatise on the means of preventing destruction by torrents and their formation in the department of Basses-Alpes [FAB 97]. With benefit of inventory, it is the oldest published text that we have in France (Figure 1.9. (a)). Fabre recommends avoiding clearing or imposing restrictions on them, such as leaving cleared strips, since the walls (check dams) prescribed by law had not been built; reforesting the mountain by planting acorns and beechnuts to populate these areas with oaks and beeches and grassing over the land by planting the seeds of adapted plants. It is nearly certain that some of the practices recommended by Fabre and commonly used in the 19th Century were already in place. Doesn’t this forester report the only means known at the time to “destroy torrents”? “This method involves taking them from their origin and barricading the bed at intervals with stakes thrust into the ground, intertwined with trees placed across them and covered with stones. This will form an obstacle that will stop the water during storms and force them to deposit everything they are transporting. As the floor rises due to deposition, these works will also be raised until the torrent bed is entirely filled. Then, to prevent renewed formation, it will be greatly prudent to plant bushes at this site”.

1 Floréal, year XI: April, 1802–1803.

22

Sedimentary Crisis at the Global Scale 1

Figure 1.8. Extract from Trudaine’s atlas for the generality of Grenoble. HautDauphiné no. 7 (1745–1780). “Path through the Die Gap, Drôme, Hautes-Alpes”, along the torrent of the Petit Buëch. The alluvial fans from the mountain chains are very active; note the Buëch’s braid style (source: National archives, CP/F/14/8479)

Fabre does not recommend “walls”, for they are costly and responsible for scouring at their feet, creating a cascade. Walls already existed in the HauteProvence landscape, such as on the Mezel torrents that converge with the Asse; the bed there is large enough for the scour pool at the foot of the wall to be limited. Fabre’s negative experience regards the Cébière torrent near Castellane, into which a scoured wall collapsed. Fabre considers any effort to “destroy a torrent”2 useless and he believed that it is too late once the torrents have already dug deep “valleys” or when they are dug up to the rock. This minimalist position taken by Fabre, who is nevertheless an experienced observer, suggests that the practice of thresholds or dams (his “walls”) was still in its infancy, unless they were excessively prudent or the financial means were truly lacking, which is also possible.

2 In the 19th Century, we would speak of the “extinction” of a torrent.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

23

The oldest report we have found on the torrentiality of the Diois (Drôme) dates back to 1860, but it concerns an initiative taken in 1804. In the commune of Menglon, the Boidans torrent threatens to bury the hamlet of Gallands under gravel. The 1860 document3 demanded urgent action both to reforest the mountain and to build a dam to hold back the alluvia. He mentions a check dam made of dry rocks, measuring 7 m in height, built in 1804 with the same goal (in 1860, vestiges were present). This dam had produced erosion downstream, which destroyed the retaining walls of the village houses. This is only one example, and a systematic search would likely reveal this practice and show its use across the Alps. 1.4.2. Prefect Pierre-Henri Dugied’s project (1819) Soon after his arrival in the department of Basses-Alpes, Prefect Pierre-Henri Dugied presented an “afforestation project” to the Interior Minister. As he had just arrived from the subprefecture of Joigny, in Yonne (South of Paris basin), he had no experience with torrential activity, but he learned by taking a tour around his precinct [DUG 19]. The report that he wrote was meant to inform the departmental functionaries and French agricultural societies about the issue. In accordance with what a Restoration prefect was duty-bound to write, it was the Revolution and the laxism of the courts that were responsible for the excessive clearing and deforestation in the mountains. As corrective measures, Dugied proposed planting forest species, premiums to communes, and he showed himself to be very cautious concerning work on the mountain torrents, which he would delay for 20 years in his personal capacity. He was drawn in by techniques used by a landowner (and local representative) with a physiocratic leaning to conquer lands on the Asse River using dykes encouraging deposition. Following this visit, the prefect gladly dedicated more generous state funding, increased through owners’ contributions, to conquer 4000 ha of beautiful river land along the Durance, the Bléone and the Verdon. In order to “box in torrents”, it was rather a matter of reducing the influence of torrential rivers for the benefit of rich agriculture. Dugied’s concerns distanced themselves from the mountain source of the issues; the project is, as we can see, somewhat lacking in orientation. 1.4.3. Alexandre Surell, author of the French policy for restoring mountain territories The best-known specialist on alpine torrentiality is Alexandre Charles Surell (1813–1887). A graduate of the École polytechnique, he was named bridge and road

3 Archives of the Drôme, 57 S 46.

24

Sedimentary Crisis at the Global Scale 1

engineer for the department of the Hautes-Alpes in 1836 (at the age of 23), a position he would stay in for 7 years (Figure 1.9. (b)). Two years after his arrival, Surell wrote a famous text, the content of which is based on the effects of the floods that occurred in 1837 and 1838 [SUR 41]. This engineer is often presented as a forester due to the orientation of his work and the policy that he brought about, but he did not belong to the forester corps (he would later make a career with the railways). Surell presented himself as a successor to Fabre, with whom he likely shared the same hydraulic culture.

a)

b)

Figure 1.9. a) Photograph of Georges Fabre (1844–1911); b) Etching representing Alexandre Charles Surell (1813–1887) (source: Wikipedia)

In substance, the work must be merited on the great clarity and its attractive character for the general public, if only through its lack of hydraulic equations in the text. We can find in Fabre’s principles, with classical torrential terminology, the notion of the slope of equilibrium*, the definition of torrential lava* and the choice of check dams rather than more or less tightened longitudinal dykes4; without further delay, he maintains that the people in the country had long since adopted groynes, but they were rarely followed up on by the Administration. Surell summarized

4 Surell points out that these check dams were very widespread in his department and gave the example of those built under the Empire in the bed of the Romanche in Villars d’Arène.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

25

Fabre’s contributions, but he did not innovate with regard to torrential hydraulics; his contribution was quite different. We will start with his remark that “entire villages are on the verge of being engulfed”; we must admit that “the formation of these torrents came after the establishment of villages” [SUR 41, p. 54]. Herein lies the dynamic aspect of Surell’s work, the process being modern in that it had the dual vision of space and time (he was only 25 and this native of Bitche in Lorraine had just gotten to know the Hautes-Alpes!). If Surell placed geological causality ahead of climatic causality in his explanations, his work is full of history in its part dedicated to the formation and violence of torrents. The basic principle of Surell’s reflections is that there is a birth, life and death of torrents. In his words, torrents are born before inhabitants’ eyes (e.g. on the mountain of Orcières), develop (as with the mountain of Saint-Sauveur, facing Embrun) and are extinguished 5 (like in Savines, where Surell observed evidence of former, contemporary activity of a human establishment). His understanding of phenomena integrates time into a fragmented, multiple space. The extinction of torrents remains to be explained. When the torrent’s slope of equilibrium is reached, scouring stops and “everything becomes calm”: “Nothing will remain standing except solid rocks, everything scourable having been carried away… The torrent will find its origin at the food of a wall of steep rocks” [SUR 41, p. 143]. In the technical terms used by Surell, there are both torrents that are born in an upper valley and others that deviate from their course over alluvial fans in search of their slope of equilibrium, and yet others that have acquired a stable regime. Herein lies the approach that we call “synchronic”* today. From another perspective, known as “diachronic”*, the torrential river seeks a longitudinal profile (through “corrosion”, i.e. bed incision or thalweg aggradation), reaches its slope of equilibrium and wanders to finally acquire its stable regime (at the end of an undetermined period). Surell does not employ the term “cycle”, which can be found in the works of geographer W.M. Davis [DAV 99], but the concept is implicit in this statement: “Each started with a torrential era, and each ends or will end with a stable state” [SUR 41, p. 150]. “Nascent” torrents are attributed to deforestation, as attested to systematically by “the seniors in every commune”, but the process is reversible, since afforestation can

5 The volcano metaphor is implicit, as is the term “lava”, but Surell’s explanation definitely stems from the field of hydraulics.

26

Sedimentary Crisis at the Global Scale 1

contribute to their expansion, thanks to roots, to the shelter of canopy*, obstacles and the imbibition of humus. It is indeed the placement into a historic perspective across landscapes and mankind that lends strength to Surell’s work. As for the causes of the fragilization of slopes, Surell does not innovate and returns to Dugied’s ideas: the local farming community had cleared steep slopes covered with pastures despite the prohibition by Colbert’s 1669 ruling, and they were spared by the courts; the Revolution and the Empire were also responsible, starting with the spread of the law from Floréal 9, year XI, which only punished the clearing of wooded areas. Surell concludes: “after the plows come the herds”, both local and transhumant ones. An unpopulated mountain like the Dévoluy “abandons the plow” and “bases all of its resources on herds” that “hasten the ruination of the nation, which will perish at the hands of this very resource” [SUR 41, p. 185]. We could be tempted to see the parallel to the torrential cycle in this history of the humanized landscape. The logical consequence implemented by the Administration would be the policy to evict inhabitants from threatened slopes. To the number of solutions, Surell proposed creating reservations, even “no-go areas” on eroded slopes in order to allow the forest to rebuild itself; he advocated planting grass, with rapid effects, when afforestation was not possible, notably at higher altitudes. In the face of the opposition that manifested itself in the mountain inhabitants, the Administration would impose its views and hasten the process by protecting these spaces and planting them. Even though his discourse was more favorable to local farmers than to transhumant herders, Surell recommended recognizing the public utility of reforestation and practicing the seizure of property. One of his key statements is the following: “We must ask for help from living forces rather than fight the torrents by building up stone and earthworks at high costs” [SUR 41, p. 161]. Surell’s plan is thus to implement biological engineering and not to continue treating torrential beds, which he judges to be expensive and pointless. Alexandre Surell was not the only specialist in alpine torrentiality. Two names emerge from the period between 1850 and 1860, which was crucial for the maturation of the ideas that led to the law on the restoration of mountain lands (restauration des terrains de montagne, RTM): Auguste Blanqui in the political field [PON 95] and Joseph Scipion Gras in the technical domain. Auguste Blanqui (1805–1881) went down in history for his socialist, utopian ideas as well as his revolutionary stances [BLA 46]. A native of Puget-Théniers in the department of Alpes-Maritimes and therefore familiar with torrential matters,

The Torrential Crisis in the European Mountains (14th–19th Centuries)

27

Blanqui chose to support taking rights of ownership away from individuals using and abusing their freedom; this position is in line with his views on social equality: “What could be wrong with the owner of abandoned lands whose sterility he continues to maintain being bound to reforest them or concede the pointless possession thereof to the State, thereby acting for himself and all others?” Blanqui likely led a role in preparing people for the principle of expropriation that would be upheld in the RTM’s policy, launched by an 1860 law. Joseph Scipion Gras (1806–1873), born in Grenoble and a graduate of the École polytechnique, was named the engineer-in-chief of the mines in his hometown, where he was responsible for geological and meteorological studies on alpine torrents [GRA 50]. J.S. Gras published two important studies that led to progress in the understanding of solid transport in saturated flow and rehabilitated the principle of check dams; he made a case for high-altitude check dams capable of stopping “pebble fans” in a manner all the more stable as the deposition has gained vegetation. He also made a case for the creation of devices allowing sediment deposition at the top of alluvial fans. J.S. Gras applied his principles to the confluence of the Vénéon and the Romanche, in the upper part of the Bourg-d’Oisans basin; the Romanche actually aggraded notably between the dykes in the 1840s and caused increasingly serious floods. His 1850 work contains an appendix on the torrents that descend from the eastern flank of the Chartreuse Mountain upstream from Grenoble in a very particular geological and climatic context. The solution lies in the construction of stone or wooden dams capable of containing the pebbles descending from the mountain walls inside the upper catchments. His 1857 work expands these perspectives and presents very concrete solutions for building efficient dams (partial and complete containment, containment mazes, check dams) [GRA 57]. It is very likely that Joseph Scipion Gras’s notoriety weighed heavily on the decision to build torrential dams to complement the more “biological” choices that Alexandre Surell had defended earlier. The conceptual and technical tools were ready to implement the restoration of mountain land (RTM).

28

Sedimentary Crisis at the Global Scale 1

1.4.4. The restoration of mountain land (RTM) The State’s position, when it came to implementing generalized measures, was to negotiate with the general councils of the departments in question, these being divided between the support provided by its “elite”, on the one hand, at the recommendation of government functionaries, and the reluctance of elected officials not wishing to lose the votes of rural inhabitants threatened with the loss of their range areas and the benefits of transhumance, on the other hand. The 1927 Forest Code had freely allowed pasturelands on non-forested areas, and the doxa of the era was “for the progress of agriculture”, including in communities on slopes, judged to be capable of producing grains. Ministerial measures in the late 1830s, inspired by the ideas of the Restoration, invited the general councils to vote in favor of agriculture, and it was not a question of reforestation this time. The elected officials were in sync with the government’s position, since the Southern Alps had experienced a series of drought years between 1825 and 1840 that caused them to forget the torrential danger.

Figure 1.10. The state of very advanced degradation of a small basin in Basses-Alpes: the Valette torrent in the “Saint-Pons perimeter” around 1880 (source: [GAY 82], Archives RTM)

The Torrential Crisis in the European Mountains (14th–19th Centuries)

29

This political choice could only be provisional as long as the positions of some remained unstable. Multiple consecutive years with heavy precipitation ended the short dry period with violent storms between 1840 and 1842. In this short period, two notions immediately opposed one another in the department of the Drôme with regard to how destructive torrents were to be handled [BRA 02]: one involved “enclosing” rivers, the other “treating” slopes. The former option was preferred by landowners on the plain and the prefecture, which was in favor of dyke construction, whereas the latter option found significant support among the General Council. The Drôme’s prefecture then changed its position and hoped to solve the issue at the source, “in accordance with the suggestions of the engineer Surell”. Once the hydrological crisis had occurred, the General Council returned to its original position, despite the damage caused by a new, small torrential crisis in 1850–1852, and it was content with the State’s policy in favor of agriculture. Everything changed once again after the extreme damage brought about by the general flood in May 1856 and the heavy summer storms in the mountain (Figure 1.10). The general councils definitely pronounced their positions in favor of reforestation and the government prepared the 1860 law with a rather general consensus. The law from July 28, 1860 and that from June 8, 1864 are the legislative tools of the policy for the “restoration of mountain lands” [COR 87]. These laws emphasized planting grass and reforestation, in accordance with Surell’s recommendations6. They were met with varied opinions by mountain communes, which lost range areas, but benefitted from the jobs on construction sites. Ten years after the 1860 law, the results were clear to see and the administration considered them to be fixed, but later reports would show that the new practices were not actually respected. The law from April 4, 1882 would be necessary for the Administration and the communes to progressively reach a better agreement, thanks to the compensation for expropriated lands and because depopulation had already strongly affected the mountains and reduced opposition. The 1882 law proved to be more flexible than the previous one in that the “obligatory” areas of the 1860 law were restricted to sectors in “actual and current” danger, which is clearly less restrictive [LAN 99]. The Southern Alps were the subject of more significant works than the Northern Alps, where the problems were less severe; furthermore, the altitudes there were too high to allow reforestation and pastoral activity still too excessive for the communes to accept RTM control of their pastoral space (Figures 1.11 and 1.12).

6 Numerous works and updates were carried out by the RTM [DEM 82, GAY 82, KAL 85].

30

Sedimentary Crisis at the Global Scale 1

Figure 1.11. Stabilization of a basin head with a substrate composed of marl and limestone through the riverbanks’ shaping and the fixation of “living fascinings” (bundles of sticks meant to take root), the area of the Curusquet in Basses-Alpes (source: Archives RTM)

Figure 1.12. Dam no. 1 on the Pravert torrent, built in 1894 in Tréminis (Dévoluy, Southern Alps) (source: RTM 38)

The Torrential Crisis in the European Mountains (14th–19th Centuries)

31

The alpine policy concerning the control of eroded slopes, as well as mastery of mountain torrents and torrential rivers, came about following the foundation of hydraulics knowledge acquired in the 18th Century, as they were summarized in J.-A. Fabre’s treatise. It was Alexandre Surell, however, who first figured out how to combine the dynamic understanding of the mountain space through the integration of societal pressure on slopes, the history of forest loss and the formation of torrents and finally, biological restoration methods. The 1860 law on the restoration of mountain territories led to the complementarity of Fabre’s civil engineering methods (torrential check dams built of fascines) and Surell’s biological methods (grass and tree seedling, planting of young trees). The implementation of these measures, more or less coercive depending on the location and time period, was the job of the RTM. 1.4.5. The Southern Prealps (Drôme): what kind of balance in torrential milieus? Ongoing research has been conducted in the Diois and the Baronnies on the impact of the measures implemented by the RTM [LIE 03]. In the Drôme basin, which was the most restored in the Southern Alps after 1860, 23 national forests were created and categorized into six large areas, managed by the Office national des forêts (French national office of forests), such that 80% of the basin was concerned. The torrential beds were stabilized between 1863 and 1887, and the low-order ravines were “filled” between 1887 and 1914 (all in all, more than 200 dams and more than 200 check dams were built); regarding the afforestation rates, it went from 30% (in poor forests) in 1830 to nearly 60% in 2000 (in well-managed forests). A large-scale approach was used in the eastern part of the Drôme basin by combining the archives, the analyses of fluvial landforms and their dating based on the rings of forest pines present in the thalwegs. The forest colonization of the torrential beds started in the 1860s and extended downstream following the incision of the channels, which dried out the stony substrates. Climatic change, notably with the disappearance of summer floods, likely played an early role, at least prior to the major effects of the RTM’s campaigns [AST 11]. However, it is not possible to determine the respective role of those campaigns, of land abandonment related to depopulation (through the near elimination of pasturing) and of natural regeneration (or “spontaneous” reforestation) following the abandonment of slopes. Further downstream in the fluvial system and starting in the early 20th Century, the torrential rivers narrowed; their incision on the spot was greatly marked starting in the 1950s. At the same time, the torrential rivers on plains experienced significant contraction of their braids under the effect of the stabilization of sedimentary delivery areas and the incision of riverbeds descending from upstream, the effect of a sedimentary deficit and a rarification of heavy floods. In general, the transit of

32

Sedimentary Crisis at the Global Scale 1

course load in torrential rivers diminished less than could be expected, for the materials originating from the incision of tributaries continued to flow [PIÉ 04]. 1.5. The sediment conveyor belt, from torrents to outlets 1.5.1. The forester Georges Fabre, from the Aigoual to the Gironde We started with the principle of sediment continuity from sources to the outlet, a founding principle stated by J.-A. Fabre [FAB 97], even though it remains rather implicit in his treatise. What was the fate awaiting this principle in late 19th Century France? We must turn to a man with the same name, Georges Fabre (1844–1911), to find it in its original form. From 1868 to 1900, G. Fabre, a graduate from the École polytechnique and a forester, was the director of the Gard’s Reforestation Service, working in the Mont Aigoual range. Wishing to tackle the issue of erosion in ranges rather than in individual catchments, G. Fabre made a case for favoring the concept of “extended area” to seriously regulate the problems of excessive erosion and river load. In a report from 1895 [FRA 85], G. Fabre thus demanded an area of the Dourbie (tributary to the Tarn) on the grounds that the sand of the Aigoual, brought downstream, “fatally” obstructed the port in Bordeaux, in the Gironde estuary, at a rate of 600,000 m3/year (thus, 1/7 of the solid flow from the Garonne would come from this sector of the basin). “The forest restoration in the Hautes Cévennes of the Dourbie basin can only be truly efficient and useful through the extinguishing of the thousand ramifications above the large torrential rivers one by one […]. This result can only be obtained by covering all of the craggy slopes with a continuous forest layer”. By highlighting the threats weighing on the port of Bordeaux, G. Fabre would have masked his true goal, which was to create a production forest on the ruins of the old agrarian system of the Cévennes, but this forester knew how to perform a successful “state intervention with a human face” [NOU 88]. The extensive areas created in the Aigoual were likely examples of this without respecting the 1882 law (they went beyond it) and they were validated by the law of August 16, 1913, 2 years after G. Fabre’s death. Thus, we see that the implicit principle of continuity between the Aigoual and the outlet of the Gironde was an image meant to convince the central administration; this image does not seem to have been founded on scientific bases, but this point remains to be seen. Regardless of the case, the fact that this image had political power means that continuity was on the legislator’s mind.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

33

1.5.2. The Rhône river trough Around 1860, the Little Ice Age (LIA) reached its end and the restoration of mountain territories had not yet begun; land abandonment was too recent for these effects to be felt. Furthermore, considering the downstream transfer time of gravel sediment, the river received material that had left the mountain years, even decades before that date. How did the Rhône react to the crisis of the LIA throughout its course? Homogeneously or heterogeneously? This provides an opportunity for us to observe a river in its continuity, from the mountain to the delta, at a time when this approach was still relevant. The analysis of an atlas at a 1/10,000 scale allowed us to get an exact idea of the river’s upstream–downstream diversity [PON 57–66]. Three stretches were therefore highlighted (Figure 1.13). – The Upper Rhône7, from the Fier confluence to Vienna, braids intensely, except in two sections where it is closed in and constrained by its limestone surroundings. The bed load had been completely blocked for millennia in the Basses Terres basin, carved out by the last quaternary glacier at the foot of the Jura, unable to feed downstream; it was renewed by contributions from the Ain, after the interruption of transit for about 40 km. The tributaries (the Arve, Fier, Guiers and Ain, as well as the limited contribution of the Gier between Lyon and Vienna) played an important role in this procedure, because they contributed a significant volume of materials, but it was the Ain that was the sole sedimentary supplier between the bend of the Jura and Lyon. – The Middle Rhône, between Lyon and the confluence with the Isère, does not braid or at least very little so, because its bed load is greatly reduced; the influence of the Ain cannot be felt past Vienna, because all the material that comes from it is deposited in the river bed and on the alluvial plain. In this second portion, the Rhône carries pebbles, but not enough for this to translate into a braid style. – Finally, from the Isère to the sea, the braiding is moderate, fed by the tributaries thanks to Mediterranean rain (the Drôme, Ardèche, Durance and Gard). The river flows on a rocky floor inclined by tectonic tipping, which gives it great energy, both on the slope and in its flow [BRA 10b]. In 1860, the French Rhône was therefore a source of discontinuous bed load transport along its axis, contrary to what we might think. However, all in all, it is a braided river, contrary to its medieval functioning, which was characterized by meandering, as was proven by the paleo-environmental studies conducted on the alluvial plain. The Rhône transformed during the Little Ice Age, and we can assume that this change took place first downstream from the primary sources of pebbles and that it then moved downstream, although without the continuity having time to 7 The Upper Rhône extends officially from Switzerland to Lyon.

34

Sedimentary Crisis at the Global Scale 1

be fully ensured. If we must seek perfect continuity, this must be on the side of the suspended load, made of sand, silt and clay, because these types of material travel quickly and, even though they are deposited, significant quantities were able to reach the sea throughout the periods in question.

Figure 1.13. Longitudinal discontinuity of the Rhône around 1860. Discontinuous braiding developed downstream from the injections of gravel bed load by the tributaries (source: [BRA 10b])

The Torrential Crisis in the European Mountains (14th–19th Centuries)

35

1.5.3. The redistribution of alluvia in the upper delta of the Rhône The territory of an active delta generally starts at a point upstream of the space where the water divides into two or more channels distributing the water. The Rhône delta began at the Tarascon and Beaucaire, even though the diffluence has now moved downstream. The dykes built following the great flood of May–June 1856 no longer allow river construction to be observed across the entire Camargue region. Localized flooding takes place when breaches open in the levees. The observations reported by the Marquis A. de Roys are therefore even more remarkable. In 1851, A. de Roys measured the volume of alluvia deposited on the delta plain by the Rhône’s flooding in November 1840 [NOU 88]. He estimated a 6,000,000 m3 deposit of fertile silt over 20 days on the Tarascon plain and in the Bellegarde marshes to the west, over an area of 4,000 ha. The thickness of the deposition could have reached a depth of 1 m near the river bulge close to the river and was as thin as 1 mm on the land far from the Rhône (the concentration of suspended material* in the Rhône’s water was 4 g/liter, a high value so far downstream). De Roys claimed that the “fatal dyke construction” on the Rhône, performed 800 years earlier, slowed down the rise of the plain in the Languedoc to the benefit of the narrow strip of alluvial belt within the dykes, such that these dykes burst more frequently. Sand invaded the plain beyond the dykes, and is piled by the action of the north wind into “small mountains” measuring 7–8 m, fixed on obstacles (with a local height of 15–20 m). The thickness of the alluvia and their size reduce downstream as the slope of the plain decreases. However, as noted by M. Pardé [PAR 25], the Rhône channel had also risen between 1842 and 1886. Everything downstream of the Rhône fluvial system reacted to the sediment production of the Alps and the Massif Central; here, we can emphasize the importance then assumed by sand in the river load arriving to the sea. 1.5.4. Solid contributions to the Rhône outlet and progression of the Camargue delta During the LIA, the Rhône’s solid flux to its outlet was made up of the suspended load from the entire basin, with contributions from the tributary basins, whose volume depended on the geography and intensity of precipitation. The solid flux also incorporated a bed load* component originating less from the Upper Rhône than from the large tributaries along its southern course. Several studies have been devoted to this issue.

36

Sedimentary Crisis at the Global Scale 1

The information concerning the Holocene past is archived in the delta’s sediment. A core sampling in the Bras de Fer reveals that it was active as a distributary* of the Rhône between 1586 and 1712. Analyses allowed the hydrology and transport rhythm towards the sea to be reconstructed. The floods and sediment contributions were elevated and frequent, leading to a certain instability of the Rhône, since its bed was in braids at the very heart of the delta. Their heavy mineral composition shows that the majority of the sand came from the nearby and steep Cévennes tributaries. The basin’s upstream portion and the Durance provided their load to the delta with a certain delay. The hydrological history of the Rhône in Arles and a series of bathymetric maps showing sedimentary accumulations off the coast of the outlet speak to the great instability of the flood’s flow and the solid transport [PIC 95]. Surveys taken on the Bras de Fer’s backfill and its former banks showed the role played by coarse sand in a morphology of “delta braids”, highlighting the reality of metamorphosis of the Rhône channels in the Camargue region between 1700 and 1710. This example materializes and validates the observations made in the 18th Century, which bore witness to extensive activity at the Rhône outlets. The progression of the Bras de Fer’s outlet was, on average, 80 m/year between 1680 and 1710, and it reached its record of 180 m/year between 1700 and 1710, before the diffluence* that took place starting in 1711 towards the actual outlet. How did the distribution on the coast of material arriving to the sea in the Rhône’s unchanged delta take place? Pierre-Jean Bompar’s old map (1591) represents the mouths of the river and their entry into the sea, loaded with alluvia. The period 1580–1590 is considered to have experienced a “significant geomorphological crisis” in response to the torrentiality of the Southern Alps; the period 1601–1646 is alleged to have experienced a relative hydrological calm, favorable to the congestion of the river mouths before two new crises, dating back to 1651–1710, 1771–1780 and 1801–1810 and finally, the 1840s allowed sediment to be expelled under hydrological pressure from upstream. The desire to correct the Rhône had been manifest since 1587, when the river was put back into its former bed after the avulsion* of Fumemorte, and it would continue growing in the 18th and 19th Centuries [PIC 95]. The mouths of the river moved laterally along the coast due to changes in the inland routes, connected to the aggradation of the branches built above the delta plain. On the occasions of floods, the aggraded channel swung from one side to the other and opened a new outlet (the so-called grau, from the Latin gradus, “passage”). A lobe was built at the new outlet, and it was then destroyed by the action of long-shore drift; the long-term equilibrium of the coastline* was thus ensured [MAI 06]. In the early 18th Century, the meandering Bras de Fer, nearly filled, only progressed 50 m/year; the fluxes swung towards the current outlet, the Bras de Pégoulier. The new primary lobe progressed 150 m/year, before experiencing a decrease to 30 m/year between 1765 and 1840.

The Torrential Crisis in the European Mountains (14th–19th Centuries)

37

During the period of strong bed load contribution in the late LIA, the speed of coastline growth was 10 times greater than it is today. Progression owed very little to the Rhône, contrary to intuitive logic. The greater part of the river contributions were collected in the delta lobe*, situated at the current outlet of the Great Rhône, without lateral redistribution. How, then, can we explain the progression experienced by the coast? If the sea provided shelly sand with a biological origin (20–30% of the volume), most of this came from erosion by the long-short drift of old delta lobes (the lobes of the Bras de Fer, closed in 1711, and the Grau d’Orgon, before the Petit Rhône, etc.). In the first two chapters of this study, we insisted on the period of the Little Ice Age, which extended from the 15th Century to the 19th Century in Western Europe, affecting in particular mountains subject to growing human pressure at the same time. On site, we were able to follow the torrential and river responses to the mountain erosion crisis. In the Alps and the Apennines, this response was expressed from the mountain drainage basins to the sea, but with complex spatial and temporal nuances under the influence of secondary climatic fluctuations and the particularities of the piedmont plains. The concrete effects of the crisis are a discontinuous but sustained feeding of the river outlets, which is translated by limited control of the excessive sediment on the part of society. In the scientific literature, the gravity of the questions posed to urban societies has led to considerable progress, notably in Italy. This will be the subject of Chapter 2.

2 Continuity in European Hydraulic Science (16th–18th Centuries)

Chapter 1 showed that adaptation to hydroclimatic crises was based on pragmatic approaches and led to an empirical understanding of hydrosedimentary processes in watersheds and in the upstream–downstream continuity of river channels, but the period also witnessed remarkable theoretical efforts. The notion of continuity from mountains to river mouths is self-evident in the way that we conceive rivers today, even though the principle suffers from a blatant lack of application1. In a work that was famous at the time, Élisée Reclus popularized the progressive movement of water, from the source of the river to its mouth, at the end of the course, sometimes hesitant and sometimes rushed by slopes, of a river channel enlarged by its tributaries [REC 69]. É. Reclus only took water flow into account; in this partial conception, the storms that struck the Cévennes fed the plains and piedmont plateaus in the Gard or Ardèche, finally ending in the Mediterranean Sea at the end of an uneven route. In temperate climates, however, which guarantee the continuity of the water flow from sources to the outlet, this purely hydrological definition of continuity is not sufficient. In actuality, water transports dissolved substances and organic particles like tree trunks, branches and pieces of leaves. Most of all, it transports mineral matter, so clear to the eye that even the lay observer cannot ignore it; this is sediment that rolls along the channel floor or that the current’s turbulence raises up in the water mass as suspended matter during floods. The water assumes a beige or reddish color when the clay and silt leave the eroded soil and join the river. The impact of pebbles along the channel

1 Since 2006, a later and largely ill-adapted substitute has been to recreate sediment continuity in France by introducing the policy of systemic leveling of milldams; they are often several centuries old and no longer have retention capacities.

Sedimentary Crisis at the Global Scale 1: Large Rivers, from Abundance to Scarcity, First Edition. Jean-Paul Bravard. © ISTE Ltd 2019. Published by ISTE Ltd and John Wiley & Sons, Inc.

40

Sedimentary Crisis at the Global Scale 1

floor can be heard, as noted by sailors on barges docked along the banks of the Rhône in Lyon in the late 19th Century. We are now going to turn our sights to Italy to discover the pioneering work of a Renaissance intellectual world. It has been proven that this country’s scientists are the originators of European hydraulics2, even though it is not modern river science in a political, economic and cultural context unique to Europe, and even though specific works have highlighted other countries. We will start our analysis with Leonardo da Vinci, simply mentioning that the author of the Leicester/Hammer Codex, compiled between 1506 and 1513, deals with various aspects of water issues. Rivers are described in two ways starting in the Cinquecento. Science deals with the circulation of water flows and the role of transport agents that erode, carry and deposit material originating partly in the mountains and partly from the erosion of the banks and floor; a fraction of the material reaches the river mouth. It took several centuries for pioneering hydraulic engineers to move past sectorial views of rivers, based on the response to concrete questions, such as the fight against floods and bank erosion, to a broader understanding of floodplains3. Very few natural phenomena demonstrate the level of complexity found in sediment transport. As we understand it today, this manifests itself through multiple continuity and discontinuity phenomena. It is discontinuity, in the form of floods or localized erosion phenomena, which has motivated human intervention along rivers, which served as the experimental foundation of the first theoretical studies. The Italian peninsula was the theater of a remarkable boom in hydraulic science, which led to its considerable lead in this concern starting in the mid-16th Century. After the concrete summary of the crises experienced by sensitive floodplains and vulnerable cities (Chapter 1), in this chapter we will move onto the theoretical aspects of river functioning as it was analyzed at that time. The difficulty the author has run into is that the France-centered vision at our disposal masks the innovations made elsewhere in Europe, not to mention other cultures. The difficulties faced by the Italian states in terms of policy and economics, particularly after the French Revolution, may have harmed the development of hotbeds on transalpine hydraulics and also the longevity, not to mention the continuity, of this science that, in the 18th Century, capitalized considerable acquisitions; similarly,

2 There is likely a great deal to be learned from Dutch works, but these were focused more on hydraulic machines and canals than natural rivers. The Dutch were also experts in matters of protection (Volume 2, Chapter 2). 3 This chapter does not consider the works that were successfully performed in countries with a long scientific tradition, such as China. Thus, during the Song Dynasty, Shen Kuo formulated a theory on the formation of the Earth including soil erosion and silt deposition [SIV 95].

Continuity in European Hydraulic Science (16th–18th Centuries)

41

French science during the Enlightenment likely concealed, in a voluntary or involuntary way, the real scientific level achieved in the provinces of the Po and in Florence. 2.1. From hydraulic architecture to the fluvial system: transalpine preeminence The geography of transalpine scientific production is a reflection of the political division in the Italian peninsula, where powerful princely cities were sustained by a flourishing exchange economy [BOU 98] with a shared culture. Starting in the 16th Century, the capitals of hydraulic knowledge were the Great Duchy of Tuscany, the Papal States and the Duchies of Emilia-Romagna (Bologna, Ferrara and Modena), not to forget Venice; in a more discontinuous way over time, Lombardy and, to a lesser degree, Piedmont also belonged to this category. Areas with secondary importance but excellent technical know-how enriched the administrative, scientific and technical networks of water territories, for example Modena Ferrara, Pisa and Padua. Hydraulic scientists shared a strong mathematic culture, diverse curiosity for other sciences, such as medicine, and a common attraction to the naturalist approach. In short, the scientific fabric was of unequaled value. We turned to the Italian literature from the era, of which two works are essential: the naturalist and theoretical treatise written in 1762 by Milanese mathematician Paolo Frisi [FRI 62] and a recent presentation of the work of Italian hydraulic engineers by M. Di Fidio and C. Gandolfi [DIF 14], but we also found points of interest in older texts. It was M. Di Fidio and C. Gandolfi’s hope to fill the gaps in modern Italian science, which is little known. Thus, they participated in an indispensible cultural work, drawing on “Italy’s immense and complex historic-cultural wealth” that characterized the period from the 17th Century to the early 19th Century. They were therefore able to show the considerable importance that it had in Europe and that their country was proud of [DIF 14, p. 15]. Italian dominance started in the late 16th Century, after a long interruption due to the cult of secrecy, manifested by Leonardo da Vinci, and the Italians’ lack of curiosity at the time regarding their intellectual heritage. How did transalpine scientific knowledge, located paradoxically in a period of slow political decadence, emerge? In the first place, if the scientific culture of the peninsula was built according to a polycentric pattern, it was nevertheless unitary, with intellectuals who rose to the top of the elite in the principalities and who acquired a European audience thanks to the exchanges in which they participated with their peers in other countries. Before the nation was even united, Italian intellectuals patriotically defended a science that was not local, but above all else Italian, against its transalpine neighbor. What were the reasons for the peninsula’s scientific progress, when the geographic dispersion of the effects could have presented an insurmountable handicap? Paradoxically, Italian eminence

42

Sedimentary Crisis at the Global Scale 1

would be born of quarrels between cities, notably between Bologna and Ferrara, concerning the course of a tributary of the Po originating in the Apeninnes, the Reno (see Chapter 1). The Reno’s alluvia filled the piedmont plain and “paludified” the agricultural plain such that the Reno course was extended by the residents of Bologna towards the Great Po to the north of Ferrara. This wealthy city was isolated in its lands, given that the branch of the Po that irrigated it, the Po of Ferrara, was being blocked. Resisting Bologna and the sediment contributions of the Reno was considered vital for this ancient city hoping to find access to navigation on the Paduan River and commercial wealth. This is the reason for the emergence of two large scientific poles: Bologna and Ferrara. 2.1.1. At the roots of European science The first great architect, hydraulic engineer and military man was Giambattista Aleotti (1546–1636) from Ferrara, who, at the request of Pope Clement VIII, suzerain of the Duke of Ferrara, multiplied land reclamations in Polesine and the navigational channels [ALE 00]. He was the first engineer to get involved in the Reno controversy, which would last from the early 16th Century until the late 18th Century (Chapter 1). G. Aleotti was an ardent defender of the interests of Ferrara against those of Bologna. More generally, even beyond the case of Bologna and Ferrara, the cities exhausted one another with pointless contestation, but by demanding constant high-level expertise, they built up science with exceptional value, based on innovative instrumentation and on observations allowing innovative interpretations. Such is the paradox of an Italy that, in these exhausting conflicts, “revealed its genius and inimitable lifestyle” [ALE 00]. Three other major hydraulic engineers worked during the second half of the 17th Century: Roman intellectual Famiano Michelini (1604–1665), Tuscan Vincenzo Viviani and Bolognese Domenico Guglielmini [GUG 97, GUG 39]. Their works stand out from the other publications of the time due to the reputation that they acquired and the use of their works made by their successors, long after their original publication. Famiano Michelini (1604–1655) wrote the Trattato della direzione de’ fiumi, a treatise on river management published in 1664 in Florence [MIC 64]. The author, originally from Rome, studied mathematics before being called to teach in Florence and Pisa and to become the protégé of the Medici family. His work certainly mentions the basic shapes of rivers, including their mouths, but differentiated (and brief) terminology is not enough to found a modern vision of what we call “an integrated fluvial system” today. Nevertheless, longitudinal variations are presented by F. Michelini, essentially through the effect of the long-profile slope, but it is for the purpose of better studying the effect of oblique currents on dykes, in order to

Continuity in European Hydraulic Science (16th–18th Centuries)

43

improve their efficiency, which is reflected in the title of the work. The great Tuscan hydraulic engineer only worked from a perspective of improving local protection, so the floodplain dimension was missing from his treatise.

a)

b)

Figure 2.1. Two great Italian scholars. a) Vincenzo Viviani (1622–1703); b) Paolo Frisi (1728–1784) (source: Wikipedia)

Another great Tuscan scholar, Vincenzo Viviani (1622–1703), was innovative in his way of dealing with the question of the aggradation of the Arno and its tributary, the Ombrone, above their sediment [VIV 88] (Figure 2.1a); the issue concerned river beds that sit atop plains, which subsequently undergo progressive paludification. In doing this, V. Viviani challenged the weight of deforestation and of the erosion of farmed soils in the process of increasing solid transport, notably because pebbles were “furiously carried down the river”. Viviani proposed the first policy to defend soil with very precise solutions, including the construction of river check dams with their convex side facing upstream and a central channel; he also advocated a reforestation policy, using olive trees if possible, which would be facilitated by prior scarification. These are all measures that clearly anticipated France’s work restoring mountain lands, which came more than a century later, at least if we refer to Fabre’s treatise (see above). The measures recommended by Viviani were implemented sporadically in the 17th Century and integrated into public programs in a more general way in the 18th Century. It was all very well for Di Fidio and Gandolfi to believe that Surell’s treatise (Chapter 1) was not as innovative as French foresters claimed it was, but they were also aware of Italy’s weaknesses. Viviani did

44

Sedimentary Crisis at the Global Scale 1

not produce a treatise to save him from oblivion, his speech to La Serenissima only giving a partial vision of his work. That fact that Viviani was forgotten is due to the Italians themselves, which indicates in and of itself a waste of cultural resources for a country “that does not have too many”, according to Di Fidio and Gandolfi. Finally, the work that overshadowed Famiano Michelini [MIC 64] was that of Bolognese Domenico Guglielmini (1655–1710), who ended his career as a professor in Padua. His notoriety, however, was international in his day [GUG 97, GUG 39]. His great treatise on hydraulics, published in 1697, ensured his lasting fame. The subject of the book comes from the experiences of its author as the intendant of waters for the province of Bologna. D. Guglielmini, considered the founder of the naturalist approach to rivers, showed that the study of the “hydraulic body” and the “human body” must draw on similar methods. His geomorphological approach to the riverbed proposed that the balance between the forces applied on the floor and the resistance to flow (with the roughness* of the banks) was upset by floods; the author thus opened the path for dynamic approaches. One interesting aspect of his work is the description and interpretation of the material composition of rivers. For D. Guglielmini, it is obvious that “coarse and fine” gravel and sand are made of “pulverized pieces of rock”, which the particles pushed by the current wear down, round and polish through friction and impact. Unfortunately, he would be contested 65 years later by the remarkable hydraulic engineer Paolo Frisi [FRI 62, p. 11], who, while admitting some degree of erosion, considered rocks, gravel and sand to be bodies “prepared by nature”. 2.1.2. A great Italian scholar, Paolo Frisi Paolo Frisi (1728–1784) was a famous Milanese mathematician and hydraulic engineer [FRI 62]. Frisi is the archetype of the European scholar, for he taught in Milan and Bologna while also being a member of academies in Saint Petersburg, Berlin and Stockholm and a correspondent for that of Paris (Figure 2.1b). His famous treatise on methods of regulating rivers and torrents, published in Tuscany in 1762, was translated into French in 1774 [FRI 74]. The translation was published in Paris under the auspices of Daniel Trudaine (1703–1769), intendant of finance and director of the service of bridges and roadways, who died five years before the release of the text. We propose an analysis of this version of the text, dating back to 1774, insofar as it is the last great Italian treatise from the 18th Century. P. Frisi was fully aware of the history and reputation of hydraulic science in the large cities of Padua and Florence. He was in no way ignorant of the genealogy of theorems in hydraulics, whether it concerns the resistance of particles to flow

Continuity in European Hydraulic Science (16th–18th Centuries)

45

(described by Viviani, disciple of Galileo in Tuscany), the effect of pressure on speed from Guglielmini and Grandi according to abbot Castelli, or even the rules of slopes, from summits to the sea. Changes in scale, from the grains on the channel bottom to the broadest spatial perspective, are surprisingly flexible. The introduction of the treatise, in its French version, proudly underlines the importance of transalpine scientific contributions to the knowledge of rivers, even though other European countries have made their contribution, for “it is in Italy that the first seeds were sown”: “The architecture of the waters was born, expanded, and almost entirely perfected in Italy, where we have dealt with everything concerning the theory of torrents and rivers, the flow of clear and cloudy waters, slopes, the directions and variations in beds, and all of Hydrometry and Hydraulics… M. d’Alembert, in his Dictionnaire encyclopédique, attributes (to the Ultramontanes) the primary progress that has been made in this field” [FRI 74, pp. viii–ix]. The beginning of Chapter 1 offers this joyful news, if the reader loves cutting academic remarks: Paolo Frisi seems to lash out personally at Buffon, who published his Histoire naturelle in 1749. The allusion is quite transparent: “A solitary philosopher can, in the silence of his library, form doubts and question whether he knows if rivers draw their origin from the sea rather than the rain and melted snow. A traveling philosopher can have no doubt on this matter, when he turns his eyes to the bed of any river, and he wishes to make the effort to follow it to its initial source” [FRI 74, pp. 1–2]. Paolo Frisi classified imagining “underground channels leading a river to the top of mountains” as “physical revery” [FRI 74, p. 3]. To the contrary, he boasts about having truly observed rivers, for example, those in the Apennines between Modena and Pistoia, because he was responsible for tracing the new route between the two cities; he discovered the Panaro on the Lombard side and the Lima on the Tuscan side; another time, he said, and for his pleasure, he traveled up the Magra from its mouth near La Spezia beyond Pontremoli before passing the crest of the Ligurian Apennines and discovering the twelve sources of the Taro, which flows to the west of Parma before converging with the Po. Besides, Paolo Frisi “had many other occasions to walk alongside other rivers…”. He bragged in what we can imagine as a jovial way about keeping his science of naturalist observation true, with no private speculation. However, the profession of faith comes from his second book, the more

46

Sedimentary Crisis at the Global Scale 1

theoretical. He expresses his preference not for the rigor of mathematical demonstrations, but for the rigor of conjectures based on “the analogy of Nature” [FRI 74, p. 74], when they are more believable or physically certain; he thus proposes: “[…] regarding hydraulics and hydrometry as part of Physics rather than Mathematics, or as part of Mathematics in which the progress that has been made until now and that will be made in the future are purely hypothetical and limited to certain cases that may never exist in Nature”. This is the reason, Frisi tells us, why he: “[…] expressly renounced all demonstrations and hypothetical calculations… and instead of this, I gathered all the experiments, observations, and reflections that can shed some light on the cases with the greatest importance” [FRI 74, p. 64]. The reader today can only admit that he was right. Let us look at his primary contributions. In the least “consistent” mountains, those “in the most uncohesive land that, sometimes softened by underground water and rain”, hold unstable sectors. These are the “lavine”, numerous between the sources of the Panaro and the Reno, in the province of Frignano, or on Monte Cimone. The French expression “lave torrentielle” (“debris flow” in English) could derive from this term. Frisi avoided them as much as possible when he traced a route, but he saw their effects further downstream and had a passion for “rivers that flow on gravel”. In torrential rivers, he noted that the bed is formed by gravel transported as bedload, which comes and goes laterally, accumulating on one side and sweeping the bed to the other, such that the channel is sinuous. However, it is when the torrent arrives to the plain that it poses the most difficulties to human installations, to the stability of roads and cities. P. Frisi simply evokes the means to combat the alternating corrosion that drills into the concave banks, all the more so when the rivers are broad and powerful: inclined banks, large stones, piles and spurs of varying sizes, and stacks of gabions. However, as a scientist more than an engineer, P. Frisi was interested less in the means of defending the banks of torrential rivers than in the principles that govern the functioning of river channels. In the first place, the quantity of material present in the bed increases as the slope decreases: the bed divides into branches and “very small veins”, a beautiful way to describe a braided bed. P. Frisi then describes what

Continuity in European Hydraulic Science (16th–18th Centuries)

47

we call “particle fining*”, particles that detach from mountains: large irregular stones, then round stones, gravel and pebbles, “large sand beyond the extreme limits of gravel”, and finally “fine sand lifted from the floor by the violence of movement” and pure soil that leads to “suspended matter” [FRI 74, p. 5]. Today, we would call this gradient a “grain size continuum*”. P. Frisi is divided regarding the role of sorting (he is convinced of its importance) and that of erosion or the fragmentation of particles. On the one hand, P. Frisi ardently contests the principle of “formation of particle size” during their transportation, i.e. through fragmentation and erosion; instead, he favors the original character of sand, which is greatly abundant on the Earth’s surface. In this, he does not set himself apart from Buffon, Leibniz and Réaumur, and he even references their works. He is easy to follow when he states that sand can come from very rare “sandy” rocks, but not from limestone pebbles, he notes (Frisi even used surprising laboratory experiments to study the reduction in particle size in a drum filled with water)4. The consequence of this position is that Frisi favors the very modern principle of classifying particles downstream: “[…] the constant degradation of this material in rivers stems from the reduction in the fall and impetuosity of running water that, abandoning the largest and more irregular stones in the upper portions, can only transport round stones and smaller gravel over longer distances” [FRI 74, pp. 11–12]. However, following D. Guglielmini and E. Manfredi, like many authors who have measured the primary slopes of the Po basin, P. Frisi also admits the principle according to which the reduction in the size of sand downstream of the channel is purely the result of its wearing, this factor being more sensitive for sand than for gravel. Another remarkable point: Frisi explains the reduction of river slopes through that of particle size. One of the major themes of Italian hydraulic science applied to river channels is the rising of torrents and rivers atop their alluvia. Contrary to Guglielmini, who believed that rivers could not rise because wearing compensated for the solid contributions, Frisi innovated by justifying the aggradation of beds through sediment

4 We are summarizing Paolo Frisi’s treatise and the originality of his methods compared to French research, but Frisi is not an isolated researcher in that he scrupulously cites other Italian researchers, such as Father Belgrado (who contributes the notion of rolling and sliding of stones over the bed) and Father Grandi (rising of coarse particles in floodwaters, which reduces friction).

48

Sedimentary Crisis at the Global Scale 1

deposition, a reality that he observed in the Apennines and on its piedmont plains near Padua (Reno, Crossolo and Secchia) and in Tuscany (Arno). “All the bridges in Lombardy, whose arches are narrow and partially blocked, help all passers-by see the masses of gravel that accumulate there. In Pontremoli, where the Magra welcomes a large tributary, I have seen the vestiges of old houses below the level of currently existing houses” [FRI 74, p. 27]. If rivers do not rise as much as one might think, P. Frisi states, it is because gravel is only transported during the first spates of a flood and in rather limited quantities; furthermore, in the period 1750–1760, 125,000 ft3 of gravel (more than 3,500 m3) were extracted annually from rivers in his region simply to repair the routes. Sometimes, however, if transport was “extraordinary”, extraction was not enough, so the lateral roads had to be raised, such as along the Ombrone. A rising braided river, though, is very unstable, as the transalpine authors emphasize, and it can change beds, as happened with the Reno, the Panaro and the Taro (in Emilia); this also occurred with the upper tributaries of the Po in the Piedmont, where the dykes caused greater deposition (“dykes” is one of the terms used in the 18th Century in Northern Italy for the actual weirs placed through the river); nevertheless, dykes do not stop stones from moving past the sills because “the stones are lifted by the impetuosity of the water and transported at a certain distance above the bottom” by turbulence. The author also provides very concrete analyses of longitudinal profiles, observed in the case of transverse flow disturbance. P. Frisi is very reticent concerning the principle of “enclosing” gravel rivers in a straight direction, for they then deposit material both upstream and downstream, and the acceleration of the current projects gravel downstream. It also means that it is preferable not to transmit gravel and its associated morphology towards the plain using longitudinal dykes; he is highly aware of the need to think of rivers in their upstream–downstream continuity. Thus, the Lombard Po no longer welcomes tributaries loaded with gravel once it has acquired its sandy, silty floor. For P. Frisi, it is not about modifying this state of affairs through rash dyking; on this subject, he recalls the axioms of D. Guglielmini: “The best match will always be to leave (rivers) as they are, divided and sinuous… It would even be better to cross them (to build transverse sills), as Viviani suggested, in order to catch gravel in the upper trunk so long as this remains possible, and he recommended never introducing any river that transports gravel in the bed of a large river whose floor is composed of sand and silt; […] to never cut off the route of those that transport stones rather close to their own mouth” [FRI 74, p. 45].

Continuity in European Hydraulic Science (16th–18th Centuries)

49

To what end? Because a river receiving gravel would adopt the functioning of gravel tributaries and become unstable, even though we would rightfully hope to conserve a fixed bed on a plain. Thus, the Reno downstream of its last gravel reach, the Arno below the Empoli and the Tiber under the Capanaccia, upstream of Rome, have no gravel in their bed, which is, for this reason, fixed5. The conclusion of Lombard hydraulic engineers is thus indeed to straighten stony rivers as little as possible, because the risk of aggradation and overflow onto the adjacent fields is too high; we cannot “hold a river up in the air!” [FRI 74, p. 56]. It is safe to assume that Lombardy and the Apennines did not produce the first European river geomorphologist, but the one who imposed himself as the greatest. From the mountains to the sea, all the major themes were dealt with and placed into a historic perspective based on knowledge, validation and refutation. Modestly but scrupulously, P. Frisi founded his observations and interpretations on the work performed by many remarkable predecessors. P. Frisi confirmed, broadened and modulated the knowledge of the two centuries preceding his personal work; but he gave it a new impact, maybe because transalpine hydraulic science was so mature that it could move from mathematic formulation to the benefit of directly applicable results. The major question that has been asked for at least two centuries is actually the considerably excessive volumes of sediment that have descended from the Apennines to reach the Po and then the Adriatic, or the Arno and then the Mediterranean. These transfers are long-profile modification factors, involving the stagnation of water in damp, marshy basins; river science is coupled with an in-depth science of water drainage and, more generally, of land reclamation in the regions of Padua, Bologna and Ravenna, in Tuscany, or in the Pontine Marshes. There is no other region in Europe, aside from the Rhine outlet, where the drainage situation is as critical, for both natural and human reasons (the erosion of slopes and their negative consequences on the piedmont plains6. It is not surprising that transalpine river science has been called upon to differing degrees across Europe in the modern era. At the end of this development on Italian hydraulic science, it had reached such a high level at the start of the Sun King’s reign (starting in the early 1660s) that he sought to attract scholars, but the attractiveness of France was then too limited for most Italian intellectuals not to prefer the requests and debates making the relations between the cities on the Italian peninsula lively at the time. In the early 18th Century, the progress of French science would manifest itself as Franco–Italian disputes, and then the two communities would ignore one another, the French reading

5 See Chapter 1, dedicated to sediment crises of the modern era. 6 However, the climatic factor, namely consideration of a potential change in precipitation patterns, is out of concern, at least in P. Frisi’s treatise.

50

Sedimentary Crisis at the Global Scale 1

their Italian colleagues less and less at the end of the century – despite the translation of Frisi’s work – probably due to a sense of superiority that is in no way demonstrated, quite the contrary and even with harmful contempt for good relations between scholars. There remains at least one question at this stage: why did transalpine science, the first in Europe from the Renaissance to Italian unification, enter into progressive decline? According to Di Fidio and Gandolfi, the loss of Italian hydraulics’ reputation may have started in the mid-18th Century, the “glorious” scholar societies of their cities proving, despite certain efforts, to be incapable of resisting the power of the academies in Europe’s large nations, which more easily disseminated their work than in an otherwise divided, or at least fragmented, nation. We can add that the Napoleonic conquest and then domination were certainly welcomed at the beginning and allowed water administration to be organized on the Po floodplain, although, in attempting to impose the French model promoted by Gaspard de Prony on Italian hydraulics, this certainly did not lighten people’s concerns. The economy and civil society of the Po floodplain were greatly weakened during the Napoleonic Wars. As for Italian scholars, starting in the late 18th Century and after 1815, they turned to the Germanic countries, the Duchy of Milan and Veneto being under Habsburg control. The rupture of the Italian model, founded upon rivalries between cities and emulation among scholars, was consumed with the upsurge in new political and economic logics based on Italy’s loss of autonomy, as if the multiple forms of domination that Italy suffered in the second half of the 18th Century and its impoverishment in the early 19th Century were accompanied by a very difficult loss of part of its cultural resources; in particular, progressive eradication, loss of creativity and relative isolation in the field of hydraulics7. The beginnings of French science manifested themselves starting in the late 17th Century, before relative domination, if not real, then at least affirmed.

2.2. The first naturalist approaches to the water cycle in the Seine basin Since Aristotle, scholars have thought (admittedly, with numerous variants) that the water contained deep within the earth is vaporized up to the surface of mountains and returns as water. Bernard Palissy (1510–1589), on the other hand, explained that rivers are fed by rainwaters and melted snow that penetrate deep into the earth and return to the surface when they run into an underground obstacle; in

7 It is tempting to make a parallel with the arts, such as the decline of Paris in favor of New York after World War II.

Continuity in European Hydraulic Science (16th–18th Centuries)

51

1580, he came up with the concept of the water cycle* [CAP 44]8. Thanks to his observations, but without performing any measurements, Palissy had a good understanding of the basic principle of hydrostatics (underground waters never rise higher than the source from which they came), the link between the level of a river and a neighboring well; he associated gushing fountains or artesian aquifers* with this principle. In the 17th Century, two great scholars, some years apart, quantified scientific progress on a particular point, that of the water cycle. This notion concerns us primarily insofar as it rigorously introduces the contemporary notions of watersheds, branches of a network, analytical hydrological parameters and finally, the most important of all, water budget. 2.2.1. Pierre Perrault The first French hydraulic engineer, and the most famous, is Pierre Perrault, brother of writer Charles Perrault; Perrault was a contemporary of Colbert, and they were close for a time before the minister faced disgrace from Louis XIV. A hobbyist, because his career as the receiver general of finances in Paris ruined him and led to his exile in Dijon, Pierre Perrault (1611–1681) studied the origin of springs (or “fountains”) and created the notion of the hydrological cycle, addressing land with considerable surface area [PER 74]. Perrault did not hesitate to firmly oppose “common opinion”. He believed that the penetration of rainwater into the earth can only be limited and he intended to prove it. He chose to perform a large natural experiment in a small basin near the source of the Seine; he described the characteristics of the small river near the village of Aignay-le-Duc after a modest course of three leagues. His location in Dijon allowed him to easily reach the source of the Seine, likely making even more observations along the way. Comparing the volume of precipitated water in this small basin and the volume flowing through the river in the village, he calculated that 1/6 of the volume of precipitation was enough to feed the river continuously throughout the year, which makes Perrault quite close to the commonly accepted values today in the water balance for the Paris basin. Furthermore, Perrault noted that only storm rain and thawing weather (or rain falling on frozen soil) occasionally allow flow along slopes; this water (we would say “running water”) carries suspended matter that is, he said, much finer than suspended matter transported by mountain torrents. Perrault even understood hypodermic flow along slopes (without using these words, of course), when he 8 B. Palissy’s notions relative to natural water were published in 1850 in his Discours admirables de la nature des eaux et fontaines. All references here concern the editions published in 1844 by Cap.

52

Sedimentary Crisis at the Global Scale 1

described one- or two-foot penetration, the water flowing to the foot of the hill and the river “between two earth layers”, or gathered in a sandy vein. Overflowing water penetrates into the sandy sediment of the floodplain along the banks and through “the very surface of the flooded plains”, even more so as the flood persists; this water stored in the soil, in equilibrium with the level of the flooding river, returns “peacefully” to the river after the flood, often through sources (the underground water is discharged), such that this reserve keeps the river level in its “ordinary state”, until the next rain. As proof, Perrault made a list of the situations observed in Paris: sources flowing for three months towards the quay being built in the Tuileries, or even the rise of the water table in the cellars of the Royal Observatory in January 1761, a month suffering great rainfall, before taking four months to go back down. Finally, Perrault emphasized the role of evaporation, active in summer and even during the cold season, as he showed through his experiments. In fact, he wrote, evaporation is activated by the movement of the air. These observations imply great familiarity with nature on P. Perrault’s part (Table 2.1). Data measured 2

Perrault (1674)

Mariotte (1686)

Basin surface (km )

118.5 (Aignay-le-Duc)

59,280 (Paris)

Average precipitation (mm)

485

376

Flow (mm) and % of precipitations

78 (16%)

50.1 (13.3%)

Evapotranspiration

407

319.9

Table 2.1. Two water budgets in the Paris basin, that of P. Perrault in 1674 and that of E. Mariotte in 1686 (see section 2.2.2). Perrault’s numbers were recalculated by J. Sircoulon, Mariotte’s by the author, using the conversation table provided by J. Sircoulon [SIR 96]. The dimension of the Paris league changed slightly in 1674; this table does not explain whether this change was taken into account

2.2.2. Edme Mariotte Pierre Perrault’s junior for 10 years, abbot Edme Mariotte (1620–1684), chemist and botanist, wrote a series of discourses between 1679 and 1681 that would be recognized after the publication of his Essais de physique in 1686 [MAR 86]. In his Deuxième discours, Mariotte adopts Perrault’s theory, albeit without citing him, concerning the formation of sources and expands on it with numerous examples taken from the Seine basin. Without naming him, Mariotte mentions the “very skilled man” (here, we recognize the author of L’origine des fontaines) who built a rain gauge near Dijon and measured 15–19 inches of rain depending on the year; no reference is made to Pierre Perrault’s hydrological balance, which he appropriates. Mariotte performs calculations on the precipitated volume and the volume flowing in the Seine basin upstream of Paris, also finding an average precipitation value of 15 inches (as opposed to Perrault’s 19 inches).

Continuity in European Hydraulic Science (16th–18th Centuries)

53

According to J.C.I. Dooge [DOO 59], the controversy concerning the origin of the notion of the hydrological cycle, which was once attributed to Mariotte, should not have been; it was indeed Pierre Perrault who established the concept first. In reading Mariotte and his manner of referring to Perrault, the latter did indeed beat him to it. J.C.I. Dooge maintains that, for approximately a century after Perrault, Mariotte and then British scientist Edmond Halley (1687), there were no new advances in the study of the water cycle9. However, Dooge may have missed a piece of transalpine literature. It was likely Italian Benedetto Castelli who, starting in 1639, first linked a pluviometric episode occurring in the basin of Lake Trasimene with his emissary’s hydrology10. 2.2.3. French hydraulic science in the 18th Century Parallel to the hydrological advances of Pierre Perrault and Edme Mariotte, the art of hydraulics was a booming science that flourished in the second half of the 18th Century11. First, starting with the oldest, let us cite the treatise by engineer Bouillet [BOU 93]12, who essentially informs us about hydraulic architecture, with structures such as canals and locks. Bouillet remarks in his foreword that there is then “a scarcity of works that dealt with this matter in our language [French]”; that scholars, as he says, “indeed wish to share with the public the beautiful things that practice and meditation have taught them about this subject”. Bouillet even foresees objections from those who would think that his treatise was useless due to the very fact that “there is not a single river in France that cannot be navigated”; the author would like to cite the Bearn rivers that would enrich the country if work were done there and many others, “neglected”, that are ignored because merchants took land routes. Concerning these waterways, Bouillet recommended cutting down trees along the banks; placing vertical rollers at extremely sharp “turning points” to keep the boat away from the bank; bringing them together again when they divide into several channels by establishing low “jetties” (levees), made of piles lined with fascines ballasted with large stones (these are the ancestors of our gabions) and held

9 E. Halley positioned himself in a very new spatial level for his time. He showed the reality of evapotranspiration thanks to a comparison between evaporation in the Mediterranean area and the contributions of its tributaries. 10 A letter from B. Castelli to Galileo [DIF 14]. 11 A summary of the community of engineers and scientific progress of the day can be found in a book by Antoine Picon [PIC 92]. 12 Bouillet was accused of plagiarizing Dutch scientist Cornelis Meijer, who, in 1685, had written a book on the way to make the Tiber navigable where it crosses Rome (Arte di restituire a Roma la tralasciata navigatione del suo Tevere, Rome). C. Meijer reports on this in the foreword of his Nuovi ritrovamenti. Bouillet seems to have used (and translated) other Dutch works.

54

Sedimentary Crisis at the Global Scale 1

by piles of wood thrust into the bed; lifting boulders; crossing falls with the help of Dutch roller bridges, allowing boats to be fixed to them, or with the help of support locks; holding the banks in place with groynes; etc. Bouillet was not directly interested in solid transport mechanisms and the shape of rivers; in his work, these data are neutral and yield to the forms of water channeling, suitable for improving water commerce. In the next generation, but somewhat in line with Bouillet, we can find military engineer Bernard Forest de Bélidor (1698–1761), named artillery professor at the age of 22 based solely on precocious feats of arms. Bélidor was first and foremost a mathematician and he applied his science to artillery and military engineering, notably to the construction of fortifications, which he based on Vauban’s works. He was a provincial artillery captain when he finished the first book of his works on hydraulics. This great treatise, written in French [BEL 37, BEL 39, BEL 50, BEL 53], does not deal with the science of river channels, but rather with the hydraulic architecture used for war. If Bélidor’s writings are not very explicit concerning the behavior of rivers themselves, they nevertheless include precious information on the manner of placing groynes and reinforcing dykes, strengthening riverbeds to reduce the deposition of coarse particles and combat the tendency of riverbeds to rise, which threatens the land around the river: “In strengthened areas, such as between piles of bridges, the water never leaves deposits because it carries them downstream, where its bed is more open. From there, it continues on to give it greater activity and strength; it is not only necessary to correct its course, but also to reduce its bed, so that it can transport the matter it is filled with further, to the sea if possible” [BEL 53, p. 302]. This is what Bélidor had observed himself upstream of the confluence of the Ardèche and the Pont-Saint-Esprit Bridge constructed over the Rhône, at a broadening where, according to him, the pebbles accumulated dangerously. However, we will note on this point the delay in Bélidor’s conceptions with regard to those of Paolo Frisi. There is thus a connection between Bélidor’s work and that of abbot Charles Bossut (1730–1814), royal professor of hydrodynamics, whose works deal with canal projects and follow up on the second part of Bélidor’s architecture [BOS 64]. An original work is the Traité pratique des digues by engineer Bourdet [BOU 73]13, which he dedicates to a minister of the King of Prussia without explaining the 13 This work is dedicated to Othon Christophle, count of Podewils, Minister of War to His Majesty the King of Prussia.

Continuity in European Hydraulic Science (16th–18th Centuries)

55

reasons for his presence on Prussian territory although we infer that he could have emigrated there. Bourdet maintains, with some slight exaggeration, that “most authors, not to say all, have remained silent” regarding dykes, which is important given that his lands and province are surrounded by them (does this only concern Prussian authors and the levees in the Prussian provinces?).

a)

b) Figure 2.2. The correction of a sinuous, braiding plain river by (a) cutting it off and (b) lateral dyking (source: Archives nationales, Gallica)

56

Sedimentary Crisis at the Global Scale 1

Bourdet claims to have written an intelligible work, contrary to the work of Bossut and Viallet, which he considered “the flaw of being overly intellectual”; moreover, Bourdet very frequently cites “the illustrious” Bélidor (49 pages out of 164 in this treatise are free of citations of Bélidor’s treatise). Actually quite simple, Bourdet’s treatise tackles the questions of the height and space to be left between the dykes to allow floodwaters to flow; those of the orientation of the dykes in relation to the current and the shape to give them based on the land; and finally, the question of sandbank deposits encouraging the formation of jams (Figure 2.2). Bourdet’s work is skillfully addressed to the King of Prussia, as the author is concerned with the land and homes behind the dykes, which are indeed structures that aim to protect the king’s territory from flooding. Bourdet recommends “bench terracing” in front of of dykes on sinuous rivers (they are a characteristic of the current hydrological arrangement of Germanic countries); drainage-ditches beyond levees; the positioning of blocks and planting of willows to protect the dykes from shocks caused by ice floes; hunting moles and other rodents that dig up the ground; etc. Nothing very innovative! In 1775, a still highly regarded hydraulic engineer, Antoine de Chézy (1718– 1798), constructor of bridges and canals, perfected his “formula” for uniform flow*, which is still used today; in his equation, the so-called “Chézy’s coefficient” allows the resistance to friction* to be taken into account. It would be impossible for us not to mention the work of bridge and dyke engineer Gaspard de Prony, who stands out in the fields of fluid mechanics and hydraulic architecture, though without taking a great interest in the functioning of natural rivers [PRO 90]. The principal French work on hydraulic engineering from the late 18th Century, however, is that of Knight Pierre-Louis-Georges du Buat (1734–1809). An engineer at the young age of 16 thanks to an exception made by the École du génie de Mézières, du Buat was assigned as a border officer in the north; he was responsible for building the canal to connect the Aa and the Lys, as well as fortifications. Influenced, in his own words, by abbot Bossut, du Buat proposed a formula for calculating the flow of open canals using a parameter of bank roughness deduced from his works on fluid resistance to the sailing of ships14 in his Traité [BUA 79, BUA 86, BUA 16]. Following Chézy, he showed that the speed of a uniform flow in a river is due to the slope, but it is reduced by the friction of the water against the wall (the floor and the banks), friction that du Buat considered “resistance” in the edition of his work dating back to 178615; du Buat predicted, without demonstrating, the notion of viscosity*, which explains that the fluid communicates the friction to the water mass. He also showed that a riverbed with a circular section offers the least resistance to flow; this is not the case for torrents, which have a rectangular bed

14 Barré de Saint-Venant published the Notice sur la vie et les ouvrages de Pierre-LouisGeorges, comte du Buat. In particular, he summarized the Principes d’Hydraulique [BAR 66]. 15 We already saw above that these discoveries were actually Italian.

Continuity in European Hydraulic Science (16th–18th Centuries)

57

with a flat floor. In sandy beds, he described grooves perpendicular to the direction of the water (dunes), which migrate downstream due to the effect of the particles’ mobility (he thus measured a displacement of dunes at 2,400 toises in two years, i.e. nearly 4,800 m). In his emphatic style and rather vaguely, du Buat expressed the subject that is, in the perspective of this work, the continuity of the flow and of the transport of sediment from mountains to the ocean: “It is very certain that the surface of high lands, or those elevated above sea level, continuously changes, and that the ground on which we walk today is not that which our ancestors trod upon. The rain from the sky leads to valleys, or, in torrents, it precipitates part of the earth covering the heights and the hills; torrents transport this silt into rivers, rivers into larger rivers (fleuves), and those into the sea, where this grease of the earth, absorbed and swallowed into the water, is lost for vegetation. Thus, the hills shrink, valleys are filled, mountains unveil the rock of their depths; and the lowlands, lifted and fed for some time with the substance of the highlands, will in turn, though much later, sink into the ocean. The earth, then, reduced to a frightening level, will no longer present but an immense, uninhabitable marsh in the future…” [BUA 86, p. 106]. Du Buat thus anticipated the phases leading from its youth to the old age of relief brought about by the action of rivers. He conceived of this evolution towards the leveling of reliefs in the scope of a single cycle, the duration of which he estimated at 4,000 years, following the “sacred historians”. As for the fluvial system, river beds are formed of contiguous reaches whose slope progressively reduces towards the sea. The river incises its bed to the stability or “regime” when the resistance of the bottom presents an obstacle to velocity, which is responsible for corrosion towards the floor and the banks. The regime is modulated by flood periods considered “permanent, periodic, and extraordinary”, but a bed with the ability to return to the average regime (so-called “exact regime”) may be considered stable or a “permanent regime”. On both sides of the regime situation, the river incises its bed or deposits sediment, or even contracts. The riverbed also increases its resistance and reduces the speed of the waters thanks to its sinuosity. If the damage caused by the flood requires, in principle, improvements to the river, i.e. a reduction of its sinuosity, for du Buat, such work is not free of negative consequences with the projection of pebbles downstream, which Michelini was the first to demonstrate regarding the pebbles in the Arno. “It would be dangerous to correct the sinuosity of a river, in only one portion of its course, without doing so in the rest of the space that it covers up to the sea, or up to the primary river that receives its waters;

58

Sedimentary Crisis at the Global Scale 1

to act in this would be to relieve one region only to submerge another, for water, by quickly covering spaces where the first improvements would have been made, would flow abundantly, at least during floods, into those where the slope would not have received any increase, and there, it would cause much more considerable overflow than before… the speed of the current, having become faster through the increase in the slope, could be at the point of digging the bed and eroding the banks” [BUA 86, p. 185 and 187]. Du Buat had a strong understanding of the principles of solidarity that exist between a stretch of a river restructured through the reduction of its length and the hydrological and geomorphological effects that this type of work produces downstream; on the one hand, the floodwater flows more quickly, but on the other, the long profile of the floor is readjusted to reduce the slope through bed incision; finally, the lateral erosion in the concavities produces a lengthening of the route, which thus partially rediscovers its sinuosity prior to these works16. The principle of upstream–downstream solidarity in the functioning of a river was described for the first time in France, in a clear and justified manner, but the work came 20 years after that by Paolo Frisi, who was himself inspired by earlier transalpine authors. In short, the theoretical progress made in France over a century was considerable in the art of constructing canals, defending strongholds, creating ports and defending threatened banks. Priority was given by French royalty to hydraulics applied in two domains: on the one hand, the improvement of conditions of commerce on large waterways and canals and, on the other hand, the perfecting of hydraulics for military use. We can easily see that this work often concerned northern France. Mathematics led to progress in hydraulic science, while the objectives that we just listed led to progress in some types of applications on given river reaches, without engineers having admitted the importance of an integrated vision of river channels. The interest of considering more than a section of the river was not yet to be felt in the mid-18th Century. In the field of protecting lands against floods, similar practices had certainly existed for at least two centuries alongside torrential rivers. The geographic segmentation of empirical intervention consisting of protecting lands by using groynes and dykes is actually an old practice; it might go back to medieval torrential crises. In other words, the integrated conception of rivers considered along with their basin did not yet have its rightful place in France. Theorizing on innovative practices came later. The exceptions are the positions held by Philippe Buache and the Count of Buffon, who were more ambitious and tackled questions on a broader scale. 16 These principles were rediscovered on the Mississippi, when the meander was cut off in the late 19th Century to improve sailing conditions.

Continuity in European Hydraulic Science (16th–18th Centuries)

59

2.2.4. Emergence of the natural state of rivers in the mid-18th Century It is best to start by mentioning the works of geographer and cartographer Philippe Buache (1700–1773), who, in 1744, presented a very new map of France, but one whose actual significance is somewhat anecdotal [BUA 70]. The theoretical foundations of this map were not revealed to the Academy until 1752 in his Essai de géographie physique, which explains that “large mountain chains… naturally divide the land into either elevated regions or riverlands inclined toward each sea”. It would be erroneous to claim that the hydrographic basin (and the hydrographic network that drains it) is the primary subject of Buache’s map of France, even though the drawing seductively highlights a partitioned image of the country that we are all familiar with today [LAG 87]. The structure of the country (which P. Buache calls the “frame”) is given through the mountain chains represented systematically, even when there is no identified ridge line but a plain (like in Languedoc to the west of the Rhône delta or in some places along the Gironde estuary). The geographic position of the ridges is deduced from the primary lines dividing the country’s waters; this leaves the “riverlands”, which he would then call “basins”. Thus, the hydrographic basin, a very legitimate notion, appeared in 1744, based on a false hypothesis. Buache’s basin was not based on flow, but a priori on the physical structure of the land. However, river science is not limited to hydraulics and small-scale cartography; it begins to deal with rivers in their natural environment. Some years before Bélidor’s work, volume I of the Histoire naturelle by Georges-Louis Leclerc, Count of Buffon (1707–1788), written in Montbard in 1744 and published in several parts in 1749 [BUF 49], clearly showed more theoretical ambition. Leclerc’s approach was rather geographic, as it combined biology (primarily) and geology (slightly). Buffon’s contribution that is of greatest interest to us deals with the formation of valleys and river processes. N. Broc believed that Buffon had followed Bourguet’s positions [BOU 29], which inventoried the mechanical actions of water, though without understanding the connection that exists with the shape of valleys [BRO 69]. It is the “waters from the sky, tributaries, rivers, and torrents” that produce the largest changes in the Earth’s surface. They have formed their bed from the mountains to the sea17, sometimes underground, but Buffon believes that valleys are “canals” (channels) that were dug prior to this by “sea currents”. Buffon introduced the innovative notion of bankfull discharge* before overflow18; the elevation of the land bordering rivers through the preferential deposition of silt; the multiple entries to river mouths; the increased height of the water’s surface in the middle of a flooding river, such as on the Aveyron (written as Aveiron in the 18th Century),

17 [BUF 49]: part entitled “Second discours de l’histoire et théorie de la terre”, p. 116. 18 [BUF 49]: part entitled “Preuves de la théorie de la terre. Des fleuves”, pp. 333–375.

60

Sedimentary Crisis at the Global Scale 1

whereas the opposite occurs near the mouth during low tide because the river level is “dug” in the center; the movements of the river floor are signs announcing sudden floods; finally, the maximum speed is generally at a mid-depth, etc. All of these notions were nothing new in Europe, but they accessibly revealed to educated readers of the era rather in-depth knowledge of natural processes. Buffon should also be credited for the fact that these notions have in no way become outdated, whereas the concepts of the formation of the Earth have not resisted the later progress of science. In Les époques de la nature, a work that came 30 years after his Histoire naturelle [BUF 79, BUF 80], Buffon, then intendant to the King’s Garden and Cabinet, returned to the digging of valleys. He starts with the presence of fossils in the mountains to deduce that these were covered by the “universal sea”, then “boiling” and yet populated by organisms. The waters then opened underground routes and overtook “greatly sinking” areas on the Equator, thus lowering to their current level. The waters that drew back from the continents “worked the surface of the earth” by flowing to the West. It is surprising that Buffon did not foresee the action of tectonics, which Buache, however, proposed nearly 30 years earlier. Curiously, Buffon revisited Darcet’s proposition according to which the valley floors are filled in proportion to the erosion of the mountains, though he did not cite this. Buffon admits the transportation of sediment towards the old ocean, but he does not justify the filling of valleys. Les époques de la nature has a certain charm, but only a small portion of this book made it to posterity as a foundation of valley and river science today, as if Buffon’s great taste for fossilized plant and animal species, as well as the nature of rocks, had led him to mask his understanding of geological processes and the work of rivers; in this sense, Les époques de la nature is a curiosity and nothing more. Reputable scholars have adopted the naturalist approach dear to Buffon, with a sense of observation and an understanding of the processes that place Buffon’s contribution far behind. The work we will discuss was performed by scholars rooted in their mountain territory, i.e. their observation site. We will go in chronological order, even though this is not necessarily sensible, given that these works all date back to the period 1775–1780. Doctor and chemist by training, originally from Gascony and Landes, Jean Darcet or d’Arcey (1724–1801) claimed a physical and naturalist approach in the speech that he made in 1775 upon being hired by the Collège de France [DAR 76]. Darcet adopted the approach of following rivers in the Pyrenees from high mountains to the seashore. On the one hand, he opposed the lowering and degradation of the mountains and, on the other hand, the “rising course of large

Continuity in European Hydraulic Science (16th–18th Centuries)

61

rivers” when, after the strong slope of torrents, they arrive to the plains. Darcet attributed erosion to several processes: the chemical decomposition that altered granite and schist in different ways; the destructive action of roots and the combined set of snow and water with the phenomenon of avalanches (these only being due to the snow, snow mixed with debris, rapid melting and water delivered by “heavenly deluges”). Other processes: “liquid mud torrents”, which he called lavanches (a term that came before torrential lava or debris flow); the digging action of “fountains” (sources) in ravines and the action of torrents that dug and “still dig their bed every day”19. Not to mention earthquakes, violent winds and so on. All around the Pyrenees region, rivers (“gaves” or torrents) are covered in debris, the size of which decreases towards the plain, to finally be reduced to the size of pebbles and gravel in Bayonne20. With this text, the reader is in the presence of a vision not of river basins, but of an upstream–downstream succession of landforms, processes and deposits described from the summits of the Pyrenees up to and including the northern piedmont plain. This is the first French work to propose this vision, already complex in nature, and, as such, Darcet’s name deserves to be passed down to posterity. Les voyages dans les Alpes, written by Genevan intellectual Horace-Bénédict de Saussure (1740–1799), a trained physicist, is among the most remarkable naturalist research in Europe. Saussure visited the “Chamouni Glaciers” at the age of 20 and never again stopped climbing the mountains and traveling across Europe. His descriptions are primarily geological in nature, but some aspects of these descriptions concern us here. The first volume of Voyages is important for his notations on the effect of sediment contributions by the Rhône in Valais into Lake Geneva [SAU 79]; Saussure describes the growth of depositions at the mouth in the form of an “earthen edge” made of fine sand that had accumulated, according to a local witness, to over half a league (more than 2 km) over 50 years, under the effect of the Rhône and the wind. Saussure extended the process to the backfill on the Rhône valley floor upstream of the lake: “This valley is perfectly horizontal, made up of parallel beds of sand and silt, slightly elevated above the level of the river, and even still soaked with its waters, which make it marshy” [SAU 86, p. 3]. Saussure had the idea to measure the volume of sediment in a mass of water during different seasons of the year to determine annual deposition and the number of years that it would take to fill Lake Geneva. As he put it, if he chose to perform the experiment, he would have to take into account the suspended sediment 19 Darcet remarked that deposits like those from current torrents are stacked several hundred toises above the valley floor today, similar to the summit of Mount Irati. 20 Even in the Médoc, Chalosse and Armagnac, the Quaternary piedmont plain is thus recognized, even with a mention of the red earth that packs the rolled pebbles.

62

Sedimentary Crisis at the Global Scale 1

removed from the lake due to the effect of the strong cold wind on the floor; all while admitting that this loss was compensated for by the contributions of small tributaries. Here, we are thus in the presence of a very innovative project involving a lake’s sediment balance with input and output. Saussure also provided the seasonal pattern of the lake’s waters, rising from April to August and dropping from September to December. Saussure only had a minimal interest in rivers unless the site was picturesque and remarkable. He presented the blockage of water to the emissary of the department of Léman through the flooding Arve and lingered over the Perte-du-Rhône in Bellegarde. On this point, he expressed the hypothesis that the stones and sand descending from the Crédoz21 to the muddy slopes and rolling quickly under 20 m deep water increased the corrosive force of the Rhône on its limestone floor. During his journeys in the French Alps, Saussure also had an opportunity to describe the rupture of a natural dam located between Sallanches and “Chamouny” (Chamonix), which had been responsible for a debacle22. These observations, as educational as they are, cannot be organized into a general explanatory theory, even though he integrated Lake Geneva into the Rhône’s longitudinal profile. 2.2.5. Jean-Antoine Fabre, the great naturalist engineer of Southern Alpine torrents Until the late 18th Century, river science had remained a discipline for military and hydraulic engineers in France, with the exception of a few naturalists, like Buffon. One great name, however, blazed the trail to more “integrated” conceptions, from mountain torrents to river mouths. If posterity has remembered the name Alexandre Surell, then Jean-Antoine Fabre is the father of French forest science (see Chapter 1). Jean-Antoine Fabre (1748–1834), born in Saint-André-des-Alpes (HautesAlpes), a trained civil engineer, was named “hydraulic engineer of the States of Provence” at the age of 32. It seems that he achieved a successful career entirely or at least largely in Provence and the Southern Alps, i.e. his native land. In 1780, the year of his nomination, he presented a well-received essay to the Académie des sciences de Paris, his Essai sur la théorie des torrens et des rivières des pays de montagne. Fabre restructured and published this first study in 1797 while serving as the head civil engineer for the department of Var [FAB 97].

21 The Crédoz or Crêt d’Eau is the southern part of the easternmost mountain in the Jura chain. Its lower portion is formed of a pebbly moraine from Würm glaciation, which explains the descent of the material noted by Saussure. 22 According to [GRA 57].

Continuity in European Hydraulic Science (16th–18th Centuries)

63

The book is very likely the first on the matter written by a French hydraulic engineer. In his introduction, Fabre emphasizes the fact that “nothing has ever been said on torrents”, at least as he describes them, and that “this matter is completely new”23. His critique regarding his predecessors is quite interesting: “they wanted to apply algebraic calculations and to express through general equations the laws that the waters follow in their course”. Fabre approves of the method of works being performed by human hands, but condemns theory “on rivers, which have an infinite number of variations and operate differently at each stage based on the volume of the water and the nature and position of obstacles”, a purely hypothetical theory (“if we abuse calculations”), which “relates to rivers as we have imagined them, but which will be completely foreign to rivers existing in nature”. This text is a manifesto in favor of the naturalist approach to rivers, more than 50 years before Alexandre Surell’s work [SUR 41], nevertheless considered as pioneering by the corps of foresters. What are the primary contributions of J.-A. Fabre’s work to the science of torrents, “river-torrents” and rivers? By summarizing them, let us revisit the different points that Fabre deals with, reclassifying them from upstream to downstream of the hydrographic network in relation to the presentation made by the book. 2.2.5.1. (Alpine) torrents Mountains devastated by torrents are left with nothing but bare rock, without any topsoil, trees or bushes. “The destruction of the woods that covered our mountains is the first cause of the formation of torrents”, because the foliage “intercepted” the rainwaters and because the layer of vegetated land absorbed a considerable amount of water. Clearings have also favored the formation of torrents; the absence of terraces on slopes due to the occasional nature of harvests (terraces or walls nevertheless made necessary by law) has reduced the “tenacity” of the earth. Forests and pastures have been ruined. Rainwaters flow “superficially” on mountain slopes. Torrents are hierarchized organizations, with “partial and primitive” ravines of the first order, organizing themselves into branches of increasing rank. Torrents form in small valleys, dig their beds and retreat their upper course headward to approach verticality, since the current is always stronger than the floor’s resistance24. This hollowing of the channel creates an embankment that will itself be cut through by secondary torrents. Due to

23 Intentionally or not, Fabre ignores Italian contributions, which came much earlier, as we have already seen. 24 Fabre formulated this as follows: “the origin of the torrent must continuously rise”, which we call today the retreat of valley heads due to regressive erosion.

64

Sedimentary Crisis at the Global Scale 1

this, torrents transport more rocks, for example, than rivers in sloping, forested regions. Even without mentioning torrential lava or debris flow, as Fabre describes it, the torrent is subject to great variation; it carries soil and stones “that create a sort of liquefaction” and are deposited in a mix. Floods are strongest when the mountain slopes are steep, scarcely forested and bare. They are long and strong when they are due to general rains, even the melting of snow; they are short if the rains are strong and tempestuous, because the rains damp down the soil, which facilitates superficial flow. In a torrential river, the channel is deformed during floods, resulting in a new balance. In the short cycle of the flood, “the bed will lower at the start of a flood and rise at the end”. Furthermore, the bed may then be higher than before the flood. Another observation: the slope of the bed is reduced at the foot of the mountain, but it will be “ever steeper as the coarseness of the materials increases”. These two observations are remarkable in the context of French hydraulic history. Fabre distinguishes between two interconnected torrential beds: the “lit mineur” (that of ordinary water) and the “lit majeur” (floodplain), “aggraded by gravel deposited by large floods”. Based on this, the areas located along rivers were ruined by deposition, and navigation or floating is made impossible by the “divisions of the river into several branches”. Another type of disaster results from the deposits formed at the river mouths. They often intercept navigation, as is the case on the Rhône. The fact that “these deposits can only come from the wearing away of cleared mountains” shows, if we even need to point it out, the very integrated conception that Fabre had of a basin and its hydrographic network. 2.2.5.2. River torrents The gravel of a tributary “river torrent” (our torrential river) is coarser and less rounded than that of the river that receives it (and the flow is more limited). It follows that the slope is stronger on the tributary than on the river. The rise and lowering of the bed take place as in torrents, but, throughout the flood, river torrents lower less slowly than they rise. Regarding the river forms of the bed, river torrents do not, in principle, have any pools because the duration of the flood and the discharge necessary for adjusting the channel bottom is shorter there; in fact, “the bed does not have time to balance itself with the force of the water”. On the

Continuity in European Hydraulic Science (16th–18th Centuries)

65

horizontal plane, diversion of the channel and changes in the river route at the expense of riparian areas are frequent. 2.2.5.3. Rivers When tributaries dilute the torrential contribution with strong volumes of water after several days of activity, the river separates the suspended matter and the layers of gravel on the bottom. The volume of “steady” water brought by the tributaries25, which “gives consistency to the channel bottom”, provides confirmation that the torrent has evolved into a “perfect” river. The “river torrent” is then an intermediate state between the torrent and the perfect river. If the beds do not rise above certain limits, it is because the pebbles are being worn down during transport. In principle, a river should experience acceleration in the speed of its waters downstream, with an increase in flow and depth. However, this acceleration is limited by obstacles and inequalities on the bottom, like pools which “the law of balance requires at intervals”, sinuosities and, finally, the resistance of downstream waters. These pools are dug during floods; their elevated downstream extremity (the riffle) is modified during the flood. Rivers with gravel and stone bottoms flow below mountains; the size of the matter on the floor reduces as they move away. Rolling and shocks experienced by the stones transform them from an angular shape to a more regular and rounded one: stones become pebbles. It is the roundness of the rolled pebbles that sets a “river” apart from a “river torrent”. The resistance of stones on the channel bottom is, on average, proportionate to their size, but they can support each other (through particle nesting), which modifies the “resulting resistance”. The strength of a steady bottom is proportionate to the size of the stones. With a constant volume (or discharge), the slope reduces or increases with the size of the matter on the floor; however, if the flow reduces, the slope increases. The result is that the slope of a bed is not a straight line, but an asymptote made up of a series of planes whose inclination varies at each stage. The excessive length of rivers and large floods cause their beds to rise, their currents to flow towards the banks and their channel to divide into several branches (bringing about the river style we call “braiding”). The corrections proposed by J.-A. Fabre can be extreme as shown in Figure 2.3.

25 The notion of balance is very new for the era, because it connects the action of the current with the “resistance of material on the floor”. It will be revisited or reinvented later.

66

Sedimentary Crisis at the Global Scale 1

Figure 2.3. Different ways of constraining rivers between groynes and dry wall dykes, with the aim of fixing the channel and bringing about deposition on its margins (source: Archives nationales, Paris)

2.2.5.4. Mouths: the delta of the Rhône Bars or islands form at river mouths, where the material pushed by rivers is also pushed back by the sea. J.-A. Fabre distinguishes oceans from seas without tides, in which the absence of flushing causes bars to form closer to the coast until they form islands. Based on this criterion, the author sets Atlantic rivers in France against the Rhône, where “an island manifests itself each day”. These islands “soon connect to the Continent and extend it into the sea”, because the waters separating islands from terra firma are predisposed to deposition. Since 1711, “the Rhône [had] brought its mouth approximately 3,000 toises above the Saint-Louis Tower” in Aigues-Mortes, i.e. approximately 5,400 m. As the bed is extended, the floor rises upstream: “it follows that if the adjacent areas do not rise proportionately, they will become marshes”. These marshes, fed by filtration from the river and the rain, are drained towards the sea. “Marshes will continue to increase as the sea moves away”. Fabre provides the example of the

Continuity in European Hydraulic Science (16th–18th Centuries)

67

Monestier (probably the Abbey of Montmajour), which had to be built “dry”, but which was “well in front of the marshes” in the late 18th Century. Jean-Antoine Fabre’s treatise, the foundations of which were set forth in 1780 and refounded in the work published in 1797, is exceptional for several reasons. First, it deals with an approach that could be called “essentially Italian”, insofar as its author is rooted in a territory that he knows perfectly and has the ability to generalize, allowing him to provide a greatly valuable treatise. In fact, there are similarities with certain Apennine approaches. All of the geomorphological components of what we call “the fluvial system” are analyzed from the mountain torrent to the mouth, precisely explaining the criteria, often visual in nature, used to move from one stretch to another. However, Fabre’s work lacks vernacular and technical terms, like sites and experiments presented with the help of demonstrations (unlike that of Surell). These deficiencies, possibly simply due to a concern to avoid localism to the benefit of generalization, nevertheless reveal a certain distance from vernacular knowledge; they also translate isolation vis-à-vis literature on the matter, whether voluntarily or not. If the specific subject of torrential erosion was too poorly dealt with to allow any reference to previously published scientific works, the fact remains that Italian literature was completely ignored. Second, in Fabre’s treatise, the main channel is described and interpreted in its functioning in three dimensions (long-profile slope, length and vertical modifications of the transverse profile); we could add the fourth dimension, as the author modulates his interpretations based on the time of the flood, that of the decrease in flood levels, and low waters. To our knowledge, this is the first time that this complexity was represented so completely in French engineering literature. The on-site observations are remarkably imbued with mobility. The concept of balance between water flow and the stability or movement of material on the channel bottom is presented with the very term of equilibrium, which will be used much later. We can claim that Jean-Antoine Fabre was a precursor in understanding and formalizing the notion of dynamic balance. 2.3. Conclusion Knowledge of rivers and the intervention methods that were implemented in France in an empirical and increasingly normative manner by engineers stems directly from the civil and military needs of the 17th and 18th Centuries. At that time, the civil needs concerned the development of navigation, corrected natural river channels being considered safer and connecting channels allowing partial

68

Sedimentary Crisis at the Global Scale 1

networking of the land; they also concerned hydraulic machines to raise the water. The military needs involved defending strongholds through the mastery of waters in their immediate vicinity. The late 18th Century, particularly with the great study by Jean-Antoine Fabre, shows that it is necessary to express a strong need, a new necessity, for a paradigm change or rather for a new paradigm that is complementary to previous conceptions to come about. It was a matter of mastering torrential crises that, as we saw above, struck France and Europe starting in the late Middle Ages, were of local concern in the 17th Century and culminated in the 19th Century.

3 Exploited Nature and the River’s Responses to the Globe’s Surface

The first two chapters of this work showed the way in which Europe has undergone a series of hydrosedimentary crises since the end of the Middle Ages and has known how to overcome them by associating empirical knowledge, scientific knowledge and their concrete implementation. Everything indicates that this knowledge and the mastery of the waters have been of native essence, vernacular knowledge being passed along by academic knowledge in France, Italy and Switzerland, countries in which we have looked at this question. A landscape in equilibrium (always relative) presents erosion phenomena in elevated river basins and on river banks. The sediment thus produced is transported downstream, but part of it is deposited at the surface of the floodplain; this is a normal, progressive geological process, capable of incorporating nutritional principles that generally benefit natural vegetation and agriculture into the floodplain soil. However, erosion and sedimentation can increase excessively following climatic fluctuations or practices that do not respect the environment, or at least the system’s equilibrium; a crisis situation can result from this, characterized by the congestion of channels and sedimentary deposits. Crises have two main causes: climatic causes or anthropic causes, though this distinction is of little importance insofar as they are usually intertwined. Erosive crises caused by human activity are abnormally intense manifestations of erosion, more or less sustainable, that affect portions of the land subject to high or exceptional pressure [NEB 83]. We will distinguish the accidental crisis, which is the brief, generally unsustainable response of the landscape to meteorological stress, for example. There are several types of erosive crises. We will distinguish crises related to the destabilization of the soil and those created by the exploitation of mineral resources. The general effect of these is disturbing sediment injection and river transit.

Sedimentary Crisis at the Global Scale 1: Large Rivers, from Abundance to Scarcity, First Edition. Jean-Paul Bravard. © ISTE Ltd 2019. Published by ISTE Ltd and John Wiley & Sons, Inc.

70

Sedimentary Crisis at the Global Scale 1

In the first category, we will include two large regions of the globe: on the one hand, regions of the Old World, exploited for millennia, and on the other hand, regions of the New World, subject to the contemporary exploitation of their soil. – The erosive crisis of the Loess Plateau in Northern China belongs to the contemporary era, connected to overpopulation and poor development practices; however, Chinese specialists place its origin back at least 2000 years. The emphasis placed on the anthropic cause may indeed mask climatic factors still unknown today. – Without excluding the role of climatic fluctuations, erosive and sedimentary crises in new countries are largely due to the destabilization of soil by the eruption of space management methods that break away from earlier practices. First, we will mention this matter in rural regions of the United States, where relief energy plays a moderate role; secondly, in New Zealand, where it only took 40 years to create a major crisis. To show the considerable magnitude of this issue, we will simply cite a study conducted in the United States showing that “modern uses” (in fact deforestation and overgrazing since the start of colonization) were responsible for the following consequences (Table 3.1). Nature of damage due to accelerated erosion in the United States

Cumulative area (km2)

Destroyed agricultural land

200,000

Damaged agricultural land

200,000

Loss of the arable layer

400,000

Threats to the value of the soil and the durability of agricultural production

400,000

Total

1,200,000

Table 3.1. Nature and extent of the damage to the soil in the United States since developing the land [BEN 38]

In the second category, dedicated to the brutal and massive exploitation of natural resources, we will successively present the impacts of the Gold Rush in the Sierra Nevada in California and those of coal extraction in the Appalachian Mountains. The extraction of mining resources is often accompanied by water pollution; if mineral waste or wastelands are generally better contained than before, chemical pollution can be a scourge on ecosystems downstream.

Exploited Nature and the River’s Responses to the Globe’s Surface

71

3.1. Mistreated soil and accelerated erosion 3.1.1. The Huang-He (Yellow River) basin: accelerated erosion in a highly fragile milieu 3.1.1.1. The land of yellow soil The Loess Plateau is located in the median part of the Huang-He basin. This is the region of the globe where the loess1 is most developed, in terms of both surface area (approximately 430,000 km2) and depth; on average, loess thickness is 100 m, but more than 300 m in the province of Gansu near Lanzhou [SHI 00]. In the presentation made by Jules Sion regarding the “land of yellow soil”, it is simply a question of arid and miserable plateaus, nearly devoid of people and dug out for troglodytic habitats, of difficult access, which made it a permanent sanctuary land, though one cut up by fertile valleys [SIO 28]. Nowhere is it a matter of erosion of the loess, even though this is certainly an active process.2 In fact, a work written by Jesuit Priest Émile Licent presents a completely different and remarkable vision [LIC 24]. Having the advantage of living on site, he witnessed the destruction of forests and observed wind erosion north of the Shanxi and torrential erosion to the south, which attacked the plateau’s fragile sediment (the loess resting upon the Villafranchian and the Pontian clay). Father É. Licent, who held a doctorate in natural science3, measured the tonnage of suspended material exported by a torrent during floods, described an overloaded flow and, thanks to irrigation canals, the sale and provision to downstream farmers of water loaded with mud and rich in sheep manure. Most importantly, in 1935, he predicted that 100,000 people in China would have to leave the region within 50 years at the rate things were going. The concerns expressed by Father Licent clearly emphasized the absence of sustainability in those methods of development. Following Father Licent’s steps, the reader descends the Huang-He, nicknamed the “heartbreak” or “sorrow” of China, and enters the Grand Plain, where rivers change courses and build inhospitable desert deltas. A particular place is made for the threat of floods on the Huang-He, which, downstream from its closed valley and on both sides of the Shandong peninsula,

1 The yellow soil, huangtu in Chinese, is actually a light brown color. 2 For J. Sion, the non-dynamic approach to purely describing forms is similar to that of M.L. Fuller [FUL 22] and G.B. Barbour [BAR 35]. Barbour, who made the journey with Father Teilhard de Chardin, also does not deal with the issue of soil loss. 3 Licent made his colleague, Jesuit Priest Teilhard de Chardin, come along to participate in prehistoric discoveries.

72

Sedimentary Crisis at the Global Scale 1

has changed its route several times throughout history. Leaving its path to the north-northeast, followed from 602 B.C. to 1194 A.D., the river flowed to the east until 1853, then going into the Sea of China to the south of Shandong (see Volume 2, Figure 2.1). The alluvia from the yellow soil implicitly explain the mobility of the river over its immense alluvial fan and the rise of its bed since it was dyked in: “Towards the bifurcation of the old and the new course, the floor of the thalweg is 5 m above the surrounding plains. Should the slightest fissure come about in the dyke, the water will precipitate over them and cover them with infertile sand. If the breach grows, the river will take on a new course… It is always a terrible danger to hope to contain this large river in a bed that is too narrow and too high” [LIC 24, p. 102]. The upper Huang-He, upstream from Hekhouzen (and the Loess Plateau), covers half of the river basin’s total surface area (which is 750,000 km2) and collects 54% of its water, but the load it produces is only 9% of what the river has at the mountain outlet. The high basin is thus the primary source of the water discharge, which will allow it to dilute the enormous suspended load from the Loess Plateau further downstream. The mid-Huang-He, on the other hand, between Hekhouzen and Longmen, only represents 17.5% of the basin’s surface and only provides 14% of the discharge, but it delivers 55% of the sediment load. At its outlet from the Plateau, on the course going from Longmen to Tongguan, the river receives another 22% of its discharge and 34% of its load. Finally, two downstream tributaries, the Yi-Luohe and the Qinhe, contribute to the dilution of the suspended material, which is highly concentrated in this sector, by contributing 11% of the discharge, but 2% of the load [ZHE 89]. 3.1.1.2. The long geological history of the Loess The loess was contributed to by the winds blowing from the elevated deserts in the west, the northwest and the north based on the climates that Central Asia has seen, and it has accumulated since the end of the tertiary era (i.e. for approximately 20 million years). Recent work, however, shows that a major proportion of the loess comes from the wind erosion of deposits present in the upper course of the Huang-He. Over millions of years, the rise of the plateau has led to the incision of rivers, and repeated earthquakes have made the very structure of the loess fragile, its mass being divided into vertical sections. Subsequent cuts to the mass of the loess reveal stratified palaeosoils eroded by old ravine systems that were dated, some of them as for back as 600,000–460,000 years BP, 240,000–180,000 years BP, 128,000–75,000 years BP and since 11,500 years BP. To explain the formation of the plateau’s mass,

Exploited Nature and the River’s Responses to the Globe’s Surface

73

we must understand that old reliefs were fossilized by multiple contributions of fresh loess, which alternated with periods of erosion. The current landscape was forged in a long period of crisis that corresponds to the shift to a more humid climate in the Holocene period, approximately 11,500 years BP, and then to agricultural settlement which became a major erosion factor from 600 AD, the plateau then being the historic heart of China [HUA 06]. The accumulation of sediment from the erosion of the high basin of the HuangHe was measured in the low valley, thanks to deep probes. Three large phases could be distinguished [XU 98]: – 13,000–5,000 years BP: the accumulation rate was low and constant. Erosion started purely naturally due to the climate, which was cold and dry during the late ice age. The period from 10,000 to 5,000 years BP, on the other hand, was warm and humid, as indicated by tree pollen, which protected the soil. The accumulation rate was between 0.1 and 0.2 cm/year, which made up for the average rate of regional subsidence; – 5,000–1,400 years BP: the accumulation rate slowly increased due to a climatic change to colder, drier conditions, favoring predominantly herbaceous vegetation (Artemisia). Precipitation rates went down, but natural erosion creased during a phase of so-called “rapid natural erosion”, during which the sedimentation rate was between 0.2 and 0.6 cm/year. Human activity was present, but its effect is considered negligible; – 1,400–150 years BP: the destruction of vegetation by inhabitants was rapid, particularly with the dawn of agriculture under the Sui (581–618 AD) and Tang Dynasties (618–907 AD). The sedimentation rate in the low valley then increased to 0.6–2.8 cm/year. “Accelerated” anthropic erosion started and became a major cause of soil loss on the Plateau. Accumulation in the bed of the Huang-He was exaggerated by dyking, which started during the era of the Warring States (475–221 BC); – for 150 years, the destruction of natural vegetation, the erosion of soil on the plateau by the expansion of agricultural clearings and the perfection of downstream dyking led to extreme accumulation rates (2.8–8 cm/year). The subsidence of the low valley and the rise of the sea level had a secondary importance vis-à-vis that of sedimentation. The recording of the Huang-He’s flooding is another source of information that allowed the flooding periods to be quantified according to the large periods in Chinese history (Table 3.2).

74

Sedimentary Crisis at the Global Scale 1

Periods and number of floods

Number of floodings

Before the Sui Dynasty (581–618 AD)

1.1 per century

Ming Dynasty (1368–1644 sequence) (155)

2.04 per year

1912–1936 sequence (103)

4.29 per year

Table 3.2. Frequency of Yellow River floodings [SHI 00]

3.1.1.3. The erosion of loess soils The increase in floods is due to clearings and poor agricultural practices on the slopes of the Loess Plateau. The conquest of new arable lands, involving 70% of the loess’ slopes, is itself correlated with the acceleration in population growth, which exploded in the 20th Century, notably under Mao Zedong, who made this elevated, isolated region a refuge for his army [WAN 06] (Table 3.3). Moreover, the development of communications, industries and cities weakened the region; the projections concerning it are pessimistic. Time line

Population (millions)

End of the western Han Dynasty (2nd Century AD)

10

280 AD

3.5

Sui Dynasty (581–618) and after

Fluctuations

591 (Ming Dynasty)

11.5

1749 (Qianlong Dynasty)

23.5

1860

> 40

1953

38

1985

81

2000

104

Table 3.3. Population growth on the Loess Plateau over the past 2000 years [WAN 06]

The current climate of the Loess Plateau is not homogeneous. The summer monsoon drenches the plateau more in the south (sub-humid: 650 mm) than in the north (semi-arid: 200–250 mm). Natural vegetation (the forest to the south and the steppe to the north) protected the soil from erosion by streams on the surface until approximately 2,500 years BP,

Exploited Nature and the River’s Responses to the Globe’s Surface

75

but the forest covers no more than 6% of the area. Great demands have been made upon it, because it had traditional uses added to the massive deforestation from the period of the Great Leap Forward to feed the furnaces of the popular communes producing pig iron.

Figure 3.1. Loess Plateaus eroded and partially transformed into terraces in the Shanxi, a province of Taiyuan (late 1920s) (source: W.D. Jones, Univ. of Chicago Press)

Erosion today is highly correlated with the intensity and volume of the tempestuous precipitation occurring during the monsoon. The result of millennia of erosion and the recent crisis is a characteristic landscape made up of plateaus surrounded by steep escarpments, ridges and craggy hills in the loess with a yellowish tint that can be found in flooding rivers (Figure 3.1). The erosive activity is due to the semi-arid climate, which is too dry for vegetation to protect the soil; it is also due to a rise in the albedo* (due to clearing and cultivating), which led to reduced precipitation. The Loess Plateau is very sensitive to subtle climatic changes. Once the season has come, the monsoon rains are violent enough to erode the unprotected soil. Craggy surfaces are estimated to cover 60% of the area, which is a considerable amount. Hills with a 45º slope are cultivated even when agriculture should not be risked on slopes greater than 25º. This is the reason why erosion was

76

Sedimentary Crisis at the Global Scale 1

“accelerated” by human activities (Table 3.4). The density of the ravine network continues to extend through regressive erosion (the headward erosion of ravines can reach a value of 100–200 m during a single storm). On average, there are more than 3 km of ravines per km2 of surface area. On large ravines 5–10 km in length, low-order ravines branch out, 5–20 m deep, dominated by hill slopes that are, on average, quite steep (50–60º). The Loess Plateau is one of the most eroded regions of the globe today, with an enormous average material loss, erosion values standing between 8,000 and 25,000 tons/km2/year in the provinces of Shanxi and Shaanxi (the maximum loss measured is 60,000 t/km2/year). The quantity of sediment transported by the Huang-He is in the ballpark of 1.6 billion t/year, which represents 90% of the total load provided to the fluvial system (the difference, 10%, corresponds to trapping in ravines and deposition in river valleys). Other studies emphasize that the sediment delivery ratio* is approaching 1 (or 100%) in the floors of many narrow valleys devoid of trapping spaces. Period

Erosion (109 t/year)

742 AD

1.16

1820

1.33

1949

1.63

1980

2.23

Table 3.4. The acceleration in soil loss on the Loess Plateau since the early 19th Century (source: various authors)

In summer, when the monsoon rains come, the rivers in the region are heavily loaded with silt, with concentrations reaching up to 400 kg/m3, with a maximum4 of nearly 1,570 kg/m3. The Loess Plateau is undoubtedly the region of the globe with the most significant erosion, as much for its effects on the Huang-He as for the impact it has on the Plateau itself. This is because it impacts a considerable mass of the loess, which is quite fertile in terms of its whole thickness. Holocene soil, stricto sensu, has disappeared, but the substrate is fertile in and of itself. This case is unique for a region feeding enormous concentrations of suspended sediment in so large a river and indirectly responsible for large catastrophes through the defluviations that sediment accumulation imposes downstream.

4 As a point of comparison, the Rhône transports between 1 and 2 g/l or kg/m3 during floods.

Exploited Nature and the River’s Responses to the Globe’s Surface

77

3.1.2. Soil erosion in North America Now, let us deal with the question of accelerated erosion* in North American soil, more precisely in the United States. The essential reference source here is by George Perkins Marsh [MAR 74], the first American author to handle the impact of human activity on the Earth’s surface. His work was published during a transition period, when the effects of occupying territory were starting to be felt. 3.1.2.1. G.P. Marsh’s lessons and observations, from Europe to the Southern Appalachians The first chapter of Marsh’s work presents a reflection on the natural state of the New World; for him, it is not unjust to consider it in comparison to the state of erosion before European colonization. The author believes that, in a situation where nature is undisturbed (forested or not), a balance has been reached at the end of an expression of conflicting forces. This balance, certainly characterized by “small internal fluctuations”, would remain unchanged without human intervention. Until the early 17th Century in lands colonized by the British, the geographic elements were “strongly balanced and compensated for one another”. G.P. Marsh tells us that it only took one generation for abandoned Native American territory to return to its original state. He believes that the rapid recovery of nature after the Great Miramichi Fire5, which destroyed 15,000 km2 of land, and “seemed to have consumed the soil itself”, is proof of this. G.P. Marsh goes on to state that the death of trees is generally a reparable accident in forests where slow “successions” of “harvests” take place; the seeds of deciduous trees wait, sometimes for centuries, for fire to destroy the dense crowns of adult pines; the leafy trees then enter the succession and ensure the continuity of forest cover. In mountainous regions, on the other hand, characterized by dry seasons and struck by violent rains, vegetable cover opens up, soil is less resistant and ravines form: “Every considerable shower lays bare its roots of rock, and the torrents sent down by the thaws of spring, and by occasional heavy discharges of the summer and autumnal rains, are seas of mud and rolling stones that sometimes lay waste and bury beneath them acres, and even miles, of pasture and field and vineyard” [MAR 74]. According to G.P. Marsh, the massive exploitation of forests and fire can have disastrous effects; the scouring of soil does not stop regrowth, but the loss of soil prevents trees from reaching “maturity”. G.P. Marsh clarifies that, following 5 These fires spread in New Brunswick (Canada) during the very dry year 1825, in a region where colonists practiced slash-and-burn agriculture. Hundreds of people lost their lives, and the expanse of the forest destroyed was between 10,000 and 20,000 km2 (source: Wikipedia).

78

Sedimentary Crisis at the Global Scale 1

deforestation, it is geology that reveals the importance of the erosion caused by moving waters. He begins with examples from the French Alps and the smaller mountain crests in Tuscany and the Northern Apennines that he had visited6: “The declivities of the northern Apennines, as well as of many minor mountain ridges in Tuscany and other parts or Italy, are covered with earth which becomes itself almost a fluid when saturated with water. Hence the erosion of such surfaces is vastly greater than on many other mountains of equal steepness of inclination. The traveller who passes over the route between Bologna and Florence, and the Perugia and the Siena roads from the latter city to Rome, will have many opportunities of observing such localities” [MAR 74]. In these conditions, the old Mediterranean world was transformed into a state of desolation such that even its restoration had, in Marsh’s opinion, become impossible. The surface of Mediterranean regions managed to assume a lunar appearance in a brief historic period. G.P. Marsh praises the success of “physical restoration”, however, brought about in small territories, particularly southern France starting in 1863: “On narrow theatres, new forests have been planted; inundations of flowing streams restrained by heavy walls of masonry and other constructions; torrents compelled to aid, by depositing the slime with which they are charged, in filling up lowlands, and raising the level of morasses which their own overflows had created” [MAR 74]. In search of equivalent situations in North America, G.P. Marsh takes examples from New England. He claims that if the forests of the Adirondacks were cut down, this would bring about similar effects to those in the Southern Alps for northern and central New York State, so greatly would the absence of forests exacerbate a contrasted climate, if we wish to consider the length and cold of the winter season and the extreme heat in the summer in this state. The hills are certainly not the same there; though steep, the friability of the soil is the cause of fragility. G.P. Marsh observed previously untouched sectors where the effects of clearing could be felt. Thus, rivers in the Hudson basin have markedly low water levels, have greater water fluxes after strong rains and carry more sediment. Obstructions of the navigable channel extend downstream in proportion with the clearing of the forests and arouse fears that irreparable damage is being done to commerce. The solution, Marsh tells us, is for measures to be taken to prevent the extension of the “improvements” implemented beyond what wise economics require [MAR 74]. 6 In Marsh’s work, we find the still catastrophic state of the regions of Europe most harshly affected by the crises of the Little Ice Age, regions that he had visited.

Exploited Nature and the River’s Responses to the Globe’s Surface

79

G.P. Marsh thus seems to believe that, with the exception of small regions (in his words, “narrow theaters”) such as the Adirondacks, the nature on the American continent is lucky enough to lend itself to proper scarring in case of disturbance, even when this is violent in nature. The land of desolation is not his nor the Loess Plateau, but rather the mountains that border the Mediterranean. 3.1.2.2. Erosion in the Mississippi basin (1820–1940) The first mentions of accelerated erosion are localized and difficult for the reader to access, unless the documents are reviewed on site. Luckily, G.P. Marsh mentions the issue of accelerated erosion under English geologist C. Lyell’s pen. Traveling in Georgia and Alabama, Lyell observed the initial formation of hundreds of valleys in locations where the primitive forest had been cut down. In Georgia, he precisely described a ravine 15 m deep, 275 m long, and between 6 and 55 m wide dug in 20 years in soil composed of clay and sand, produced by the decomposition of the underlying gneiss. In these states, the brutal flow over the land, over territories deprived of their tree cover, has dug ravines 70–80 ft deep; in the northeastern states, similar processes have been observed on the friable sandy soil that has lost its pine coverage ([LYE 66] cited by [MAR 74]). Thus, we are in the presence of sediment production areas most likely situated in upstream watersheds. We could probably multiply the rapid case studies without approaching the real significance of this phenomenon. This is the reason why we will prefer to present more recent research that incorporates the entire duration of agricultural occupation, thanks to the use of diachronic methods. The first large study was performed by the Soil Conservation Service. It presents the interest in considering the issue from a perspective other than erosion, that of “accelerated sedimentation”, erosion appearing implicitly. The authors’ goal is to establish a program to stop or reduce the damage caused to the valleys’ “resources” through sedimentation. The historic approach emphasizes that the issue of erosion and river overloading is attested to, for example, in Virginia, and has been since the early 19th Century, with no anticipation of the subsequent effects. The rich land of the Appalachian piedmont plains was invaded by gravel and left infertile. This issue was declared to be of national interest in 1908, once studies had multiplied. One precious study appeared in 1935 regarding two small river basins on Lafayette County, in the north of Mississippi State [HAP 40]. The “severe” erosion of fine-textured soil (silty-sandy substrate with a loess covering) has had considerable effects on the accumulation of deposits along the coast of the Gulf of Mexico. These basins, tributaries of the Little Tallahatchie, itself a tributary of the Yazoo and the Mississippi, receive 1250 mm of precipitation per year on average. In the mid-19th Century, these regions practiced the monoculture of cotton (a cash crop) on hills, as well as the monoculture of corn in the valley floors. These were

80

Sedimentary Crisis at the Global Scale 1

opened to agricultural colonization in 1835, following the demographic boom from Mississippi Fever. The problems connected to the erosion and displacement of sand were proven in 1860 and became worse in the following decades. In the upstream– downstream process chain, the phenomena observed were: 1) the clearing of forests for agricultural use; 2) an increase in flow, accelerated erosion on hills through gully erosion and surface flow (with the local loss of a soil layer at an average rate of 15 cm in 10 years), with a spectacular landscape dissection; 3) down the hills, the digging of valleys with flat bottoms and steep slopes, which efficiently transferred sediment, notably sand; 4) the aggradation of rivers due to coarse material and the formation of river plugging due to the inability of fluvial systems to adjust and maintain their balance; 5) the generalization of the frequency and intensity of floods accompanied by the spreading of infertile sand over the floodplains; they fossilized the black “prairie” soil, characteristic of the period before the establishment of agriculture; 6) the formation of deltas at confluences; 7) the rise in groundwaters, the development of marshes on bottom land and the death of trees due to root asphyxia; 8) the development of malaria, a deadly disease at the time; 9) crop abandonment in favor of low forests and plant formations in wetlands; 10) finally, considerable expenses to replace buried equipment and to drain waterlogged land. The volume of soil redeposited on the floodplains was measured, and it allowed estimation, for the elevated areas of the two basins studied, of the layer of topsoil lost: on average over a period of 100 years, it was 14 cm. If the considerable mass of material eroded from the hills that is deposited downslope as colluvia without reaching the fluvial system had been taken into account, the total sediment production of the hills would have been in the neighborhood of 18,000 tons/ha in 100 years (180 t/ha/year) over an area of 10,000 ha. The work performed by the agents of the Soil Conservation Service in the late 1930s provided the foundation for modern hydromorphological science, particularly that of sedimentary budgets in areas suffering stripping and areas prone to deposition; they have certainly influenced those of their successors whose scientific contributions will be examined. It was the crisis of the 1930s, caused by deflation west of the Mississippi (the Dust Bowl) and crises caused by heavy precipitation falling on exposed soil with no restraint, that relaunched American science, following the pioneering works of G.K. Gilbert (see above, this chapter).

Exploited Nature and the River’s Responses to the Globe’s Surface

81

Figure 3.2. Current gully erosion of loose formations in the Valley and Ridge province (Washington County, Tennessee, Appalachians). Result of 30 years of uncontrolled erosion on schists profoundly weathered into silt and clay, in a humid subtropical climate (1,000–1,100 mm) and on hills lower than 1,000 m (source: I.E. Luffman [LUF 15])

A few years later, Luna B. Leopold [LEO 56], a renowned hydraulic engineer employed by the US Geological Survey, very rightly confirmed that it is extremely difficult to evaluate the sediment load of the hydrographic basins of the West during the presettlement period. The density of forest cover (trees and bushes specific to semi-desert regions) was significant, but nothing allows us to confirm that the load was limited in these conditions; moreover, we know nothing of the impact of two centuries of Spanish occupation and the description provided by Lewis and Clark after their journey across the West remains limited. Citing a study by Brune [BRU 48], L.B. Leopold described a 50-times increase in suspended load in relation to the “geologic” norm, when more than 50% of the land was cultivated or fallow in the Great Lake and Ohio River basin; this growth may increase to 75 times when such demands are placed on 40% of the land in the upper Mississippi basin. While recognizing that any generalization is a delicate matter, L.B. Leopold admitted that the increase between the conditions before and after settlement went from a factor of 2 to 50. L.B. Leopold also reported on advances he actively participated in concerning the “hydraulic geometry*” of rivers. Their transverse profile reflects the quasi-equilibrium between the water discharge and the sediment load; the floods shape the channel and control the relative elevation of the alluvial plain, such that overbank flooding takes place every 6 months on average.

82

Sedimentary Crisis at the Global Scale 1

L.B. Leopold posed the hypothesis that if the balance between water discharge and sediment load changes to the load’s benefit (following the erosion of the basin), the channel will rise over a “long period” of time and the level of the alluvial plain will aggrade (by accumulation) until restoration of the previous flood’s return period. As a practical consequence of this, a land restoration program will have to consider the reverse process, unless other impacts (like the construction of storage dams), even more serious, arise. As we can see, modern systemic conceptions of breakthroughs and research on the balance in watersheds were already well established in the late 1950s. In the upper valley of the Mississippi, an erosive crisis affected the thin loess soil in Wisconsin between 1820 and 1940. The southwestern part of this state, which looks down on the river, is called the Lead-Zinc District. The absence of quaternary moraine on the rocks of the old substrate, as well as the erosion of the loess deposited by the cold winds of the Pleistocene era, were considered to encourage the presence of plant species subject to the presence of metal-bearing veins in the rocks under the surface of this mineral land. This is the reason why these low hills, covered in a mosaic of forests and prairies, were saved from clearings and mining exploration by the federal authorities until the 1840s [KNO 87]7. The soil of the Native Indian or “presettlement” era, rich in dark and highly fertile humus, absorbed the precipitation and reduced water discharge to the rivers, preventing runoff and erosion. Another element to consider is the strong resistance of Native Americans to the occupation of their territory during this period. After more than 150 years of cultivation, today we can see a new level of fine alluvia in the valley floors, superimposed upon the dark gray primitive soil: these are “historical overbank deposits”. These alluvia, 30–80 cm thick based on their position in the fluvial system (they are thicker downstream, where the slopes are more gentle), are brown to yellowish, massive and rich in clay. The annual precipitation volume, which is, on average, 1,000 mm, allowed the alluvia to leave the hill region towards the Mississippi, where they mixed with those from all across the river basin, such that their route gradually becomes unclear. Lead, the production of which peaked in 1845–1847, and zinc, the peak of which was more recent (1916), are the markers used today to date the historic alluvia in their depth, since these peaks appear in their mass. The techniques used by geomorphologists attest to two periods of heavy sediment transport and vast spreading, the first in the 1870s and the second in the late 1920s and the 1930s. These crises are explained by the conjunction of several factors: a natural increase in precipitation, a strong runoff over soil used for corn cultivation with ill-adapted techniques and gully erosion on the slopes of the cultivated hills (Figure 3.3). The current situation has returned to a level near the one that prevailed starting in the 7 It would be advantageous to turn to a recent work on the same subject by S.W. Trimble [TRI 13].

Exploited Nature and the River’s Responses to the Globe’s Surface

83

1820s. Finally, for J. Knox, there is no doubt that the destabilization of the landscape by human action is the first source of the erosive crisis.

Figure 3.3. The North Platte, a sub-tributary of the Mississippi, located on the Wyoming–Nebraska border. The flat bed, 800 m wide, is congested with sand; the low water flow, which is pronounced, reveals the braided style, a response to significant sedimentary feed in the early 1930s (source: N.H. Darton, US Geological Survey)

3.1.3. Accelerated erosion on the Great Russian Plains, from Belarus to the Urals The plains of Russia and Belarus are one of the regions of the globe most affected by accelerated erosion, covering an area of 4 million km2. This immense area, which extends from the Barents Sea in the north to the Caspian Sea and the Black Sea in the south, includes several kinds of deposits (glacial, loessic) on a relief of low, slightly winding hills. The zonation of soil changes with latitude, moving from podzol in the taiga to chernozem* in the forest-steppe belt and gray, semi-arid soil, in climates where erosion is controlled by the melting of the snow to the north and by the stormy summer rains further to the south. The protective role of vegetation has been greatly reduced in regions suitable for agriculture. Clearing began in the 14th Century, became widespread in the 19th Century on steeper slopes and caused the onset of gully erosion [TOU 17]. The crisis truly began in the late 19th Century, with the suppression of triennial crop rotation in favor of the monoculture of grains in a period of rapid demographic

84

Sedimentary Crisis at the Global Scale 1

growth, this work reaching even the steep slopes of the dry valleys. Mechanization in the 1950s had harmful effects before land protection measures were taken (notably the introduction of grain feed). Russian specialists emphasize the extreme complexity of the combination of erosion factors and provide remarkable erosion values between 25 and 220 t/ha/year for localized catastrophic episodes. In the long term, the soil loss was 0.5 or 1 t/ha/year on gentle slopes, but extreme values of 15–20 t/ha/year have been recorded on steeper slopes and the fragile soil of the Stravopol region. The average value of soil loss is 4.8 t/ha/year in Russia and 3.6 t/ha/year in Belarus. The dynamics of soil erosion in the territory in question show a more or less long existence of erosion processes. – First, the sod-podzolic soil of the Smolensk–Moscow region and the mid-Volga; the layer of soil lost measured more than 10 cm on 40% of agricultural surfaces in the Moscow region and more than 60% of surfaces in the mid-Volga region in the late 19th Century. The chernozem belt remained mostly unaffected, barely exceeding the soil formation rate, which is 4–4.5 cm/100 years (as opposed to 2–3 cm/100 years in the previous region). – In the 20th Century, the erosion rate dropped greatly in the sod-podzolic soil belt, because humans stopped laboring the land on the steep slopes, which were already highly eroded. Erosion increased, on the other hand, in the chernozem belt and developed in the Stravopol region, which had been newly cleared. Accumulation by periods studied in Russia show increasing loss over time, a loss that reflects the spread of cultivated surfaces (Table 3.5). Period

Russia Gm3

Belarus Gm3

1696–1796

5.9

0.74

1796–1887

30.8

2.02

1887–1980

33.8

1.51

Total 18th–20th Centuries

70.5

4.3

Table 3.5. Evolution of the volume of earth lost by 100-year periods in Russia and Belarus (source: [SID 06])

Negative effects on the fertility of soil add to the mere loss of earth. These are changes in the microflora and the soil chemistry, alterations in water capacities, and “dehumidification”, which reduces productivity and the ability to resist erosion. Ravines are the most spectacular form of erosion, particularly in the belts of

Exploited Nature and the River’s Responses to the Globe’s Surface

85

deciduous forests* and the steppe. In Russia, the total volume of ravines formed in 300–400 years is estimated at 3.5 billion cubic meters (Figure 3.4). The period of maximum ravine growth is dated between 1860 and 1910, the highest rates being located today in the cleared sectors of virgin territory. The aggradation of river bottoms went from an average of 0.6 mm/year in the period 2,500–200 years BP to 6–6.5 mm/year for the last 200 years; this takes place in the floor of small valleys, such that only 7% of the sediment produced by erosion is transported to the sea. In other words, the sediment delivery ratio is particularly low, and it can be said that the continuity between the upper basins and the sea is much more reduced than in the following example [SID 06].

Figure 3.4. “Ovraghi*” gully erosion in the Saratov oblast (Budanova Gora) (source: Wikimedia Commons)

Another Russian region has experienced significant agricultural clearing and the phenomenon of accelerated erosion: the agricultural belt of the Ural–Tobol plateau in western Siberia, at the border between the forest and the steppe. It was first cultivated in the mid-19th Century; the intensification of agriculture did not lead to sensitive impacts until the mid-1950s, once a threshold was exceeded, it seems (and this independently of all climatic change). The process chain includes the compaction of soil and the reduction in the resulting hypodermic flow, not to mention an increase in sediment transfer to rivers. The shaping of badlands* began with an erosion rate of 15 t/ha/year, causing bed aggradation in low-order rivers [GOL 11].

86

Sedimentary Crisis at the Global Scale 1

3.1.4. New Zealand, “destruction on the pretext of development” [WYN 02] New Zealand is one of the most remarkable regions of the globe regarding the severity of the erosion that can be seen there; it is considered to have been caused by the relatively recent colonization of the land by Europeans. The Maori, who had constituted the indigenous population over the 10th–13th Centuries, were hunters and cleared areas before the arrival of European colonists, but their traditional practices, founded on burning, would have encouraged coverage by dense ferns, ensuring effective protection against erosion (Gisborne region). The Maori population dropped to 40,000 inhabitants in the early 19th Century due to the epidemics brought by Westerners. After the Treaty of Waitangi, signed in 1840, New Zealand officially became a British colony, connected to Australia by strong economic ties, but armed conflicts with the Maori continued until the 1860s. These conflicts revolved around an essential point for our understanding of the landscape today: land ownership [BOA 08]. From 1841 to 1860, The Crown’s right of pre-emption provided that Maori land would be reserved for the British Crown and exclusively purchased by the same, at a low price; they would be sold with enough profit to finance, through the New Zealand Company, an emigration fund meant for British colonization, including the establishment of colonists. First, this was the case for the South Island (fully surrendered in 1860) and for part of the North Island. This procedure was replaced by the Native Land Acts between 1862 and 1869. These allowed the sale of lands by the private sector (pioneers and speculators); this procedure was the cause of wars on the North Island, in what was called the “bush frontier”, an agricultural border. This story is that of very skilled confiscation, with the exception of a few indigenous reserves. The Maori’s rights were only recognized on effectively cultivated lands, based on the highly practical criteria set forth by William Locke; the rest was relegated to the status of a wasteland destined to be burned and converted into pastures. London’s role has been compared to that of Chicago vis-à-vis the Midwest: supplying English cities with wool, cheese, butter and frozen meat in the scope of an imperial economy, but in this case, at the price of the organized dispossession of the indigenous people. Economic development first materialized in the form of wood exports, notably Kauri wood, to New South Wales in Australia; these were transported overseas by enormous wood trains. The colonists built sawmills to produce boards, produced varnish from trees and used burning techniques to clear lands, thereby maintaining their right to occupy the land [WYN 02]. The Europeans initially used natural clearings to raise sheep for wool, then started deforesting the bush frontier of the North Island. Significant British and Scandinavian immigration was organized to develop cattle and sheep farming (the archipelago was home to more than 20 million sheep in 1900). Butter and meat exports to Great Britain were possible thanks to Bell-Coleman’s invention, which was placed on sailboats at the time and broke

Exploited Nature and the River’s Responses to the Globe’s Surface

87

down all barriers regarding cold storage. The invention preluded the boom of refrigerated transportation starting in 1882, and the purchase of Maori land by the Crown became large scale. That being said, erosion in New Zealand took place in an environment whose extreme fragility was long ignored. The construction of New Zealand’s mountain range is due to the subduction* of the Pacific Plate under the Australian Plate, from which the chain rose at a speed of 3–4 mm/year in the southern area of the recently recognized Zealandia continent. The South Island culminated at more than 3,700 m, while the North Island rose to nearly 2,800 m. The fragility of the ranges is explained by the steepness of the slopes, frequent seismic activity and the lack of cohesion in sedimentary and volcanic rocks forming the skeleton of this relief. Another natural factor explaining the sensitivity to erosion lies in the precipitation of between 500 and 10,000 mm/year, with a steep altitude gradient, given the exposure of the western range and violent storms; these are due to active frontal depressions in a temperate climate and extratropical cyclones, like the Bola cyclone that struck New Zealand in March 1988 (300–900 mm of rain in 5 days). The studies performed in the Alps of the Canterbury region (South Island) showed that erosion may be greater under the saturated elevated forest than in the dryer eastern plain [MCS 89]. Deforestation and the creation of prairies, made profitable thanks to the development perspectives of pasture livestock, were made responsible for very active erosion phenomena in an exceptionally short timespan (Figure 3.5). It was the storms of the 1930s in particular that revealed the gravity of the situation, and the first erosion map was created in 1944. Awareness grew quite brutally, because the plains were directly threatened, even destroyed by floods and river erosion. The most harmful processes were groundwater erosion, wind derosion on bare soil, screes, mass movement and gully erosion, which took a complex form in this space [MOS 80, BAS 13]. A study was conducted on more than 200 small basins with forest or extensive semi-natural pasture cover. Specific erosion* extends from 30 to 100 t/km2/year in dry regions between mountains, with records of 20,000 t/km2/year in the dissected tertiary marls of the Waiapu basin (East Cape), and even nearly 30,000 t/km2/year in a chain of highly saturated, friable schists on the South Island (Westland). It is the geology and the precipitation that control the spatial variability of the erosion rates in the natural basins of New Zealand. We are far from the records of the Loess Plateau, but these values are considerable [HIC 96]. On the North Island, the region of Gisborne-East has been metaphorically called the “large farm”; this reflects the extent of clearing and transformation to raise livestock. The landscape has been reduced to an astounding degree by the deforestation that took place between 1880 and 1920, and this basin is considered the most affected in New Zealand [PAG 00]. Gully erosion produces more than half of the eroded material, ahead of landslides (10–20%) and surface erosion (10%). The primary local river, the Waipaoa, contributes a portion of the sediment mobilized in the basin

88

Sedimentary Crisis at the Global Scale 1

to the ocean, i.e. a contribution of 68 t/ha/year, only 1% of which is bed load. In the Waipaoa basin, modeling has concluded that a 140% increase in the load was due to the arrival of the Polynesians around 1400 AD, 350% upon the arrival of European colonists and 660% once the high basins were deforested. Furthermore, since the 1880s and breaking away from the previous period, floodwater from the Waipaoa is highly concentrated, which leads to the creation of density currents once they reach the coast (these currents dive into the ocean mass and, due to this, do not provide sediment to the delta) [KET 07].

Figure 3.5. Manifestations of gully erosion on pastures after strong rains in February 2004 in the Manuwatu basin. These rains were responsible for 100-year floods in nearly all of New Zealand. Note the reforestation underway [FUL 05] (source: New Zealand Farm Forestry Association [KNO 06])

Across the country, the erosion of the highlands, all processes combined, was rapid and led to the rise in river levels starting in the early 20th Century. The phenomenon reached its height in the 1950s, then storage in the river channels and their aggradation declined, particularly near reforested areas, the country actually entering a forest restoration period (Figure 3.6). This tendency, however, is challenged by cyclones, which, while infrequent, demonstrate that erosion will not stop without an active, constantly monitored policy.

Exploited Nature and the River’s Responses to the Globe’s Surface

89

Figure 3.6. Mouth of the Rakaia (North Canterbury, South Island). An excellent example of a braiding river rich in salmon, with an average flow of 200 m3/s (source: North Canterbury Fish & Game)

The Soil Conservation and River Control Act was passed in 1941 at the same time as catchment boards were created to propose solutions. The first reforestation undertaken by the New Zealand Forest Service did not come until 1960, however, and ended in 1987 with the elimination of this administration in favor of forest concessions primarily granted to foreign investors. Cyclone Bola (1988) raised awareness that it was not possible to avoid public forest policy, at least in the district of Gisborne. In 1992, the choice was made to promote reforestation with dense exotic pines (Pinus radiata, Corsican pine) and Douglas firs, all exploitable species, on the threatened agricultural lands (Figure 3.7). Overly dependent on market principles and the farmers’ goodwill, this program was adjusted in 1988 to subsidize farmers oriented towards forestation based more on indigenous species and natural regeneration. On the 835 km2 of the East Coast region subject to erosion, i.e. 90% of its territory, 26% was returned to native plant and secondary plant formations and 20% into reforestation made up of exotic species. The effects of this policy are perceptible in the activity of forms of erosion, and low-order rivers began to incise, a sign of the reduction in contribution. Recent policy, which has become more coercive, turned not only towards the fight against erosion but also biodiversity and carbon sequestration were implemented and generalized across the country [PHI 13].

90

Sedimentary Crisis at the Global Scale 1

Figure 3.7. Efficiency of the reforestation measures taken to combat erosion in New Zealand (Kairakau, Hawkes Bay Coast, South Waimarama) (source: Peter Scott, www.abovehawkesbay.co.nz)

In the table he outlines of the state of the New Zealand landscape, Don Garden laments its dominantly exotic character, far from the original. On the North Island, imported forest species take over, like Lombardy poplars, considered efficient in the fight against pasture erosion; he also notes the absence of indigenous birds in an agricultural landscape drawn in perfectly straight lines and lined with hedges on the Waikato plain. The South Island’s landscape resembles that of Switzerland with its sheep (not Swiss, it is true), its prairies, its conifers and its snowy mountain summits. Furthermore, the archipelago imported fauna “for improvement”, with a foundation of fish and mammals imported from every continent; the latter became so numerous for hunting and touristic purpose that they damaged the soil. In short, a “hybrid” landscape developed, far from the kind expected in the Pacific, even though the archipelago has maintained 35% of its original landscape [GAR 05].

Exploited Nature and the River’s Responses to the Globe’s Surface

91

Figure 3.8. Recent forestation with Pinus radiata on a pasture hill ravaged by landslides in the Manawatu–Wanganui region (North Island); slope transformed with terrasses, scars from stripping and bulging zones (source: Robin Gruel)

The examples provided were chosen from various regions of the globe for their remarkable character, but this choice is naturally highly limited. It would be possible to present elaborate studies, like accelerated erosion in Northern Africa at the end of French colonization, or examples observed on site, like the erosion of the Bantustan in South Africa and the erosion of the new soil of the Parana in Brazil. 3.2. Mineral predation and river bursts Excessive deforestation and agriculture are capable of destabilizing considerable areas. Mining exploitation has more limited effects, but they are no less real, as we will see. The predatory exploitation of a mineral resource can cause a considerable volume of material to be displaced. This may be the case for underground resources extracted in galleries. Waste is disposed of in tailings on slopes in the case of metallic minerals or carbon; they are even piled up on structures on the surface of the soil. In some cases, often specific to new countries, but not exclusively, considerable amounts of rubble and waste from extraction are placed, after sorting,

92

Sedimentary Crisis at the Global Scale 1

in valleys in immense dry landfills or behind dams. The range of situations is wide, from debris from hydraulic extraction of gold-bearing alluvia, performed in the torrential valley of the Sierra Nevada (California), to coal waste filling Appalachian valleys. 3.2.1. Lead and zinc in the Pennines: mines threatening dairy livestock One of the reasons for the study of the hydrosystems impacted by former mining activity is to follow up on the effects of waste produced by former extraction on the downstream environment. The fluvial system is a vector of the pollution that can extend from the mining heaps or piles disposed of from the slopes of the valley to the ocean. The old mountain chains in the United Kingdom, rich in minerals, were the seat of extraction in the Middle Ages and possibly under the Roman Empire. The extraction of lead, zinc and silver in the Northern Pennines Orefield (Cumberland and Northumberland), the most significant in the nation, took place in the hydrothermal veins present in the thick layers of limestone (the “great limestones”) and sandstone of Carboniferous age from Alston Moor. While the extraction of lead (galena, a lead sulfide) is attested to as early as the 11th or 12th Century, the primary discoveries were made between the 15th and 19th Centuries; the extraction of lead developed around 1650, thanks to new techniques of water pumping and fusion, and peaked around 1850; zinc mining, on the other hand (sphalerite, a zinc sulfide), peaked around 1900, before the halt in 1920 due to the drop in international prices. One of the most studied fluvial systems is that of the South Tyne, a branch of the Tyne, a coastal river that drains the Pennines mountain chain in England [MAC 89b]. The South Tyne meets the North Tyne and flows into the North Sea near Newcastle after a distance of 120 km. The alluvial plain of the South Tyne demonstrates inactive river forms bearing witness to strong flooding in the 19th Century. Banks and islands of pebbles and gravel are formed of particles from glaciers and rivers eroded by floods; they incorporate fine mining debris from heaps eroded by the basin’s rivers. The banks of the South Tyne were raised by flooding and then perched by the hollowing of the river. These fossil landforms are very rich in fine metal particles. The concentrations of lead and zinc measured there reach values of 8–15 g/kg in the banks formed between 1860 and 1900; the concentration drops in river forms dated back to 1900–1915; it then continues to drop after this. The effect of dilution by the unpolluted upstream contributions is perceptible. One major question posed by this observation is the long-term future of the pollutants that are eroded, transported in suspension by strong floods (like the one caused by Hurricane Charley in August 1986) and redeposited at the surface of

Exploited Nature and the River’s Responses to the Globe’s Surface

93

the floodplain further downstream. In the plains downstream of the basin, the lead and zinc concentrations lie in the range of 0.25–0.5 g/kg; we must add to this trace metals, such as cadmium, copper and silver. Lead is present in its dissolved form, fixed onto grains of quartz and calcite; the mobilization of heavy metals is caused by the processes of adsorption* and desorption associated with the presence of iron oxides, manganese and organic matter [MAC 89a]. Furthermore, pollution affects the plants consumed by bovines, and heavy metals can enter the food chain through dairy consumption. This example shows the close connections between mines, the injection of metal from heaps into transported sediment, and finally their incorporation into the soil of the alluvial plain, bringing them finally into the food chain. The sea is the final receptacle of this waste. 3.2.2. The “debris” from the gold-bearing alluvia of the Sierra Nevada (California) The most remarkable environmental destabilization in the United States was caused by mining activities on the western slope of the Sierra Nevada. The founding study was conducted by Grove Karl Gilbert (1843–1918), a renowned geologist, at that time working for the US Geological Survey. This study, published in 1917, demonstrates factual intelligence and a quantitative foundation that far outranks previous studies and pre-empts the works to come on river geomorphology by several decades [GIL 17]. It may have involuntarily masked the reality of accelerated erosion due to Californian agriculture, which was a more extensive phenomenon on the surface, but less intense, particularly lacking scientific attention of this quality, before the 1930s [JAM 99].8 If the discovery of gold dates back to January 1848 in California, its exploitation in the Sierra Nevada began in 1853, first manually, then through hydraulic mining financed by rich investors in San Francisco a few years later. The gold-bearing river deposits from the Tertiary age, present in valleys, formed more or less coarse and consolidated placers; hydraulic mining of increasing intensity used a technique known as “jet-bore mining” (Figure 3.9). The coarse particles were eliminated in dry sieves, known as a trommel screens; the gold was then extracted from the sediment through cleaning and hydraulic sorting using sluices made of wood, equipped with ramps where the gold particles were deposited according to their density. The need for large quantities of water taken from torrents motivated the creation of new water rights, known as the senior rights, which stated that the first to take possession of a river resource in the West became the sole owner, which has since excluded any use upstream of the first water diversion.

8 According to Happ et al. [HAP 40, pp. 9–10].

94

Sedimentary Crisis at the Global Scale 1

Figure 3.9. Miners using hydraulic means to excavate an old stony terrace with gold-bearing placers, Malakoff Diggins, South Yuba River watershed, Nevada County, California (source: USGS, Historical Ph. by Carleton E. Watkins, Hearst Mining Collection)

The Sierra Nevada is the seat of flash floods due to the steepness of the slopes, which is exacerbated by tectonic uplift, the rise of air carried by depressions coming from the Pacific in a Mediterranean climate and finally the great variability of precipitation values each year, which was underestimated by the pioneers. The catastrophe that took place in winter 1861–1862, striking the mining region, would turn out to be a “mega-flood” with a return period of 100–200 years. This would be due to the phenomenon presently known as an “atmospheric river*”; the precipitation fell continuously for 43 days, causing thousands of deaths and the loss of 800,000 cattle in the Central Valley. The city of Sacramento was fossilized under 3 m of earth as a result of the landslides that took place in the adjacent mountains. The mining debris, trapped locally at the bottom of hills or in the canyons, was brought to the Central Valley of California. The 1861–1862 flood and the following eroded debris, whose total volume exceeded 1 billion m3 across the Sierra Nevada region between 1853 and 1884, and its effects from this period in time were recently reevaluated [DET 13]. In the Sacramento Valley, the following floods, raised by the sediment deposited in the channel, threatened to bury the lands of the lateral areas

Exploited Nature and the River’s Responses to the Globe’s Surface

95

under the debris, the river’s natural levees being easier for floods to cross since the channel had aggraded. The deposits also affected navigation, both on the low coastal rivers and in the San Francisco Bay, where the marshy delta shared by Sacramento and San Joaquin was suffering an accelerated backfill. Around 1880, 16,000 ha of rich agricultural soil and orchards were completely destroyed and another 108,000 ha seriously damaged. Yoruba City and Marysville were also ravaged. The farmers were so enraged that, in 1878, they created the Anti-Debris Association, which was quickly contested by the Hydraulic Miners Association; their legal complaint led to the pivotal ruling by a San Francisco court judge in the Woodruff versus North Bloomfield case: jet-bore mining was henceforth forbidden by the 1884 law, such that mining production nearly ceased, the figures dropping to approximately 2% of what they had been in 1870 for decades. – From mines to the delta, around 1910, Gilbert distinguished several types of river reaches more or less affected by “waves” or moving masses of debris; these were significant accumulations, the longitudinal profile of which was similar to that of the flood wave. – On the mountain, only the large accumulations of hydraulic mining debris, leaning against the hills, were in direct contact with torrential rivers and eroded at their downstream extremity. The material filled the valley floors and, at Gilbert’s times, became the primary source of material for rivers, their evacuation having begun around 1890. The backfill from the late 19th Century remained present at the edges of the valleys in the form of terraces sheltered from future erosion. – On the piedmont plain, the river channels rose during the gold extraction phase and in the decades to follow, during the canyon-draining phase. The thickness of the debris had reached its peak of 27 m in the high piedmont plain of the Sierra Nevada, drained by the Yuba and the Bear, two other tributaries of the Sacramento. The size of the deposits shrank downstream as the energy reduced. The upstream contributions were reduced after mining activity stopped; since that time, coarser material has been torn from the canyons’ thalweg, then moving downstream. – In the valleys, transportation took place because the material on the piedmont plain gave the riverbed a stronger transport slope than it previously had. For several decades, channel bottom aggradation (up to 4 m) prevented the tides from entering the lower reach of the Sacramento. The end of extraction caused long profiles to return to their 1850s state. – Gilbert estimates that 880 million m3 of sediment were deposited in the bays between 1849 and 1914, including a fraction of the soil eroded in the Central Valley; this involved a series of bays within Suisun, San Pablo and San Francisco, opening onto the Pacific Ocean at the Golden Gate bottleneck, which cuts up the coastal chain. The mud tidal flats* spread while also being reduced in favor of salt marshes*.

96

Sedimentary Crisis at the Global Scale 1

G.K. Gilbert reevaluated and synthesized the sediment budgets, from all sources, created since the 1880s for the Sacramento watershed; he updated these budgets in 1914 (Table 3.6). Budget elements Mining debris

Volumes (106 m3) 1,282

% 70.4%

Non-mining debris Total

538 1,820

29.6% 100

Table 3.6. Nature of the debris mobilized in the Sacramento watershed over a 65-year period (1849–1914), in millions of 3 m and as a percentage of the total (source: [GIL 17])

Recent calculations based on a mixing index evaluated the percentage of sediment from a source other than gold extraction at 17%. The redistribution of mining debris by morphological mega-unit from the sierra to the ocean is presented in Table 3.7. Material redistribution sites In the Sierra Nevada On the piedmont plain In river valleys On flooded valley lands and in delta marshes In bays In the ocean Total

106 m3 203 398 76 225 876 38 1,820

Table 3.7. Redistribution of material displaced towards the San Francisco Bay over a 65-year period (1849–1914) by mining operations 3 and rain (in millions of m ) (source: [GIL 17])

The concern over catching sediment in the mountain came to light in the 1880s, after hydraulic mining and the construction of transverse brush dams were banned by the US Army Corps of Engineers. The damage caused by sediment accumulation was so great, however, that in the late 19th Century, it led to the creation of the California Debris Commission. In 1904–1905, it commissioned the construction of the first “dam” taller than 4 m at the Yuba’s canyon outlet. This blocked the debris upstream until the flow dried up and caused channel incision further downstream due to the reduced load. The dam was completed in 1910 by another work, placed downstream from the former. The width of the channel was progressively reduced to 600 m, and the long profile evolved towards an equilibrium adjusted to the reduced load. To complement this, the watersheds inundated by the overflowing Sacramento

Exploited Nature and the River’s Responses to the Globe’s Surface

97

were protected by levees financed by the State of California and the federal government on the one hand and landowners on the other. The height of the floods was reduced using artificial channels that bypassed the loops of the Sacramento; extrapolation allowed Gilbert to propose a scenario in which the mining debris deposits would be progressively expulsed from the river network. The backfill of the San Francisco Bay was nevertheless harmful and, in his eyes, would remain so until the debris was evacuated. The first results were encouraging. In 1917, after a period of high tension, Gilbert was already able to observe improvements in the agriculture and navigation of the coastal river, though he continued to be concerned about the San Francisco port. The construction of the underwater bank beyond the Golden Gate bottleneck, the redistribution of the sand on the coast by northward long-short drift and the tidal fluxes, on the other hand, were concerning at the time. Gilbert predicted a complete offset of the “sedimentary wave” within 50 years, along with the incision of the river’s channel. This meant a return to the original conditions, at a speed identical to that of the aggradation of the channel following the impact of hydraulic mining. The hypothesis posed by Gilbert was the reversibility of function and river landforms once the sediment flow had dried up. In the end, Gilbert calculated a remarkable budget of benefits related to the different uses of the Sacramento watershed’s resources, which were gold extraction, agriculture (the leading economic sector at the time) and maritime commerce. In his opinion, the negative effects of extraction had already disappeared again by the time he wrote his report, but the erosion of agricultural soil had increased, in a conflict which henceforth set the agricultural sector against commerce, frontier farming working with “imprudent [methods] and sources of waste” according to the river’s users. The question posed by mining was the delivery of coarse debris, responsible for the aggradation of the channel and accumulated damage caused by floods; the agricultural issue then concerned the excessive suspended material that it produced in favor of feeding the salt marshes, but to the detriment of the integrity of the inner bays, of the positive effects of the sediment flushes ensured by the tidal fluxes, and finally, of the maritime access to the San Francisco port. The economic logic explained the rise in the costs of gold extraction due to the decreasing quality of gold reserves, the obligation to trap sediment on the mountain in a permanent way and the increased cost of hydraulic mining in a context where the economic value of the water meant for energy production was growing; in the future, agriculture will demand its share, Gilbert tells us, before cities in turn become competitors of agriculture, buying water rights. From this, Gilbert deduces the impossibility of extraction reconquering its economic position in the future due to the drop in its profitability. The future, he says, will inevitably involve cooperation between water users, notably through multiple uses of the resource; this was in 1917! Furthermore, he suggests building a 430,000-m3 dam-reservoir on the Yuba (near Hammonton);

98

Sedimentary Crisis at the Global Scale 1

this dam could store 190,000 m3 of mining waste without causing harm to electric production or irrigation. Gilbert surprisingly anticipates the management techniques that prevail today. Geomorphological studies were revisited by Allan James in the 1980s to test the validity of the sedimentary wave model proposed by Gilbert. The author shows that Gilbert’s model does not consider that the decrease in storage over time follows an asymptotic curve given that the marginal storage sectors are less and less accessible, at least in broad valleys; a portion of stored debris could remain on site for several millennia [JAM 89]. In fact, one century later, transport was still clearly greater than the values in 1850, and 90% of the sediment stored there (more than 100 hm3*) was still present in the low valley of the Bear, a river that had aggraded by 5 m between 1853 and 1884 (Figure 3.10). It once again achieved its original long profile in the channel, but not at the valley edges. Furthermore, mining started again until 1953, mobilizing more modest volumes of debris than in the previous period, but the geomorphological and hydrological effect thereof on the flood levels was nevertheless quite significant. The construction of reservoirs interrupted the sediment flux, and the sediment deficit they caused increased the erosive capacity of the downstream channels [JAM 99].

Figure 3.10. A terrace uncovered by the recent erosion of the Greenhorn Creek in a secondary river deposit of the Red Dog Ford. The material of the terrace stems from the exploitation of gold veins in the Sierra Nevada. The red vehicle at the summit of the terrace shows the scale (source: Allan James, diagonal photo taken by a drone on December 28, 2016). For full color image see: www.iste.co.uk/bravard/ sedimentary1.zip

Exploited Nature and the River’s Responses to the Globe’s Surface

99

The Californian model for managing rivers draining the Sierra Nevada to the Pacific Ocean is not innovative with regard to its dimensions and the technical choices made. It is quite classical in terms of moving sediment volumes in the fluvial system. In Leon, Spain, near the mines of Las Medulas, the Romans extracted gold from Quaternary gravel by diverting rivers for mining the alluvia and washing them to extract gold. This is not the erosion of hills feeding the hydrographic network, but secondary erosion, i.e. erosion affecting debris produced by mining activity, whether this is from coarse tailings after sieving or fine particles from hydraulic cleaning. This aspect of things is classical in the ancient world, as in the new. What is a priori surprising is the apparent absence of any anticipation of the consequences of the management method chosen, a flaw that could be due to lacking knowledge of the torrential functioning of the Californian mountain, or much more mundanely, a result of the search of the maximum profit from frenzied mining activity, as if extractions, local by nature, excluded any vision of the problem concerning the downstream valley. The devastation must have been intense, but it only flowed for 20 years (1863–1884) between the start of jet-bore mining and its prohibition, most likely a time that the State’s administration and/or the federal government measured with precision. In fact, the sierra-ocean continuum is visible in the chain of constraints undergone by the upstream residents and in the actions implemented. These actions dealt with the cause of the disturbances (stopping jet-bore mining in 1884), a policy to block sediment on the mountain, at least insofar as this was possible. The solution to the problems found downstream came about at the turn of the century, based on the anticipation of the effect of the sediment deficit in the channel, on the river storage present at the outlet of the mountain and in the Central Valley. The positive consequences will be a drop in flood levels and the reduction of constrained flooding. However, the inability to truly deal with the matter of fines circulating and being deposited in the fluvial system remains. In California, every compartment of the continental fluvial system and the San Francisco Bay was severely affected by the differing effects of using inorganic mercury in gold extraction in the 19th Century and the early 20th Century. Mercury was extracted from mines on the Coast Range and used by miners to facilitate gold separation. It is estimated that 7,600 t were then thrown back into the river of the central Sierra Nevada. Mercury was present in the 250 million m3 deposited into the San Francisco Bay. Inorganic mercury transforms into methylmercury that is highly toxic to fish populations (even in low concentrations) as well as to humans9.

9 For more on this subject, see the website of the association Gold, Greed and Genocide, which campaigns for recognition of the Native American genocide caused by the Gold Rush: www.1849.org/ggg/legacy.html.

100

Sedimentary Crisis at the Global Scale 1

The levels measured per watershed were correlated with the volumes of sediment produced; the strongest were in the Bear and Yuba basins. We saw above that the period of sediment deposition from extraction areas would be longer than predicted by G.K. Gilbert, but too many actors in this section still make references to this great geologist’s work. The destruction of the Englebright dam would allow salmon spawning beds to be restored, bringing about the repopulation of the species in the Yuba through the Central Valley. However, the transit of the sediment fraction that would come from the heaps also poses the question of their pollution by the galleries dug in the rock, by the erosion of the old, forgotten catchment structures and by the sediment that would be eroded in the reservoir itself [JAM 05]. There is another model of change, that of semi-humid mountainous Colorado. The Central District is located more than 2,000 m above sea level, in the Front Range Mountains, 40 km northwest of Denver. Gold and silver were exploited between their discovery in 1859 and the 1880s; the pine forest was cleared between 1860 and 1863, during a bonanza period. The loss of forest coverage caused a strong runoff, as well as damage to the buildings on the valley floor and, very quickly, the incision of rivers several meters deep, as witnessed by photos from the time. In these mountains, the imbalance was not the result of excessive mining debris, but of a disturbance in the water discharge, which increased the energy in the arroyos such that instability is still present today [GRA 79]. 3.2.3. The coal mines of the Loess Plateau, the Huang-He watershed In the mid-Huang-He watershed, coal extraction produces debris adding to the loss of land to superficial erosion. The mines in the coal district of Shenfu Dongsheng (2,750 km2) are located on the plain or at the edge of ravines. Extraction has a significant influence on the concentrations of the suspended material measured in rivers. Fines come from craggy tailings or direct waste in the hydrographic network. The increase in the “normal” load (i.e. with no mining impact) is between 25 and 30%, whereas the specific loss of material (assimilated to the specific erosion) is between 35 and 60 t/ha/year. These contributions to the river are polluted by heavy metals and deposited in the channel; the waters have harmful effects on agricultural production. Moreover, the increase in runoff increases the level of the floods and sometimes is responsible for the flooding of the mines themselves since their sites are exposed. Among the measures adopted, we can find the protection of riverbanks and dam-reservoirs against debris [FAN 92].

Exploited Nature and the River’s Responses to the Globe’s Surface

101

3.2.4. Mountaintop mining in the Appalachians at the risk of downstream reaches In the United States, open-air mining, organized in contour mining or in excavations, has been practiced for decades. Extreme mechanization, specific to these methods, explains their low cost, but they have the downside of producing very large quantities of wastelands in the form of debris from mountain rocks. In the Appalachians, the first coal region in the United States, the method of extracting coal considered most economic is using explosives to level wastelands forming the summit of ridges to directly access the thick layers of coal without requiring the digging of well-supported tunnels. Mechanized scouring is known as mountaintop removal mining (MTR) or more simply mountaintop mining (MTM). The coal basin of the central Appalachians covers an area of 48,000 km2; a 5,700 km2 area has already been affected by mining in the states of Kentucky, West Virginia, Virginia and Tennessee since the 1960s (i.e. more than 7% of the area). The profitability of this method is ensured by a strong reduction in mining jobs and the sale of coal to produce thermal electricity; the cost per ton is 15 dollars using MTM as opposed to 27 dollars with underground tunnels. This production method makes up 40% of American coal production, and the potential coal makes up 25% of global reserves (2 billion tons have been extracted since 1970). Scouring is completed through the deposition of the debris as close to their removal site as possible: this takes place through a gravitational backfill of the adjacent valleys (the hollers), generally drained by low-order rivers (upper reaches with intermittent flow, valley heads), but not solely, for flowing rivers are intercepted and covered by backfills. Despite the Clean Water Act (1972), the Endangered Species Act (1973) and the Surface Mine Control and Reclamation Act (1977), the federal legislation, reinterpreted through contradictory judgments, in practice tolerated direct backfill in rivers. Approximately 450 km of Appalachian valleys have already been backfilled with 6.4 km3* of schist debris and 1,600 km of drains have disappeared; another 300 km of valleys are currently being backfilled. In recent years, the extraction and backfill procedure has assumed the official name MTM-VF (VF for valley fills). The operations authorized just from 1992 to 2002 would, once completed, bring about the loss of another 1,944 km of high-basin rivers (Figure 3.11). With scouring-backfilling, we are in the presence of massive operations, but their defenders claim that, except for accidents leading to mudslides, the stored debris does not migrate into the fluvial system or only in relatively small quantities; it advances contouring and revegetation methods. It is true that means are implemented to catch the debris on these sites, but the Sierra Nevada catastrophe,

102

Sedimentary Crisis at the Global Scale 1

due to extraordinary precipitation, could take place again elsewhere, notably as an effect of climatic change.

Figure 3.11. Scouring-backfill landscape on the Appalachian ridges (mountaintop removal) (source: Vivian Stockman, ohvec.org, flyover courtesy of SouthWings.org)

The watershed most affected by extraction-backfill is that of the Coal River in West Virginia; 320 km of its upstream hydrographic network have been fossilized. The immediate questions are thus those of landscapes, of the ravages to species destroyed on extraction sites and on backfill sites, and finally, the effects of backfill leaching on the quality of the groundwater and river water. The controversy, which is often expressed violently, sets distribution companies and miners’ families against defenders of the environment. An official report by the Environmental Protection Agency [EPA 11]10 recently listed the following impacts: – the definitive loss of springs and all the small intermittent and permanent fluxes;

10 See also the legal analysis by Congress [COP 15].

Exploited Nature and the River’s Responses to the Globe’s Surface

103

– an increase in the average discharge due to reduced evapotranspiration and an increase in peak floods downstream due to channel incision [MIL 14]; a regularization of the seasonal regime as a result of winter water retention in the porous rubble and the accumulated flow of saltwater in the dry season; – elevated concentrations of major chemical ions, more than 10 times the natural rates, even reaching lethal levels; – concentrations of selenium that are toxic to fish and birds, with selenium being responsible for malformations. The acidification of the surface water through contact with rocks rich in sulfur generally encourages the leaching of toxic heavy metals into the environment; – constant shrinking of fish and bird communities in the Appalachians, which demonstrate the greatest biodiversity in the nation alongside California; – partial trapping of fine sediment from the superficial erosion of backfill, percolation into porous backfill and leaching (the river substrates actually reveal a moderate increase in the presence of fines). The authors of the EPA’s synthesis recognize that the upstream–downstream spread of undesirable chemical effects has not yet been studied, particularly for concentrations of major toxic ions. A recent scientific synthesis requires studies to be conducted in three dimensions, integrating the vertical dimension, because impacts are much better correlated with the volume of debris than to the area affected by extraction. The new topography is flatter and more elevated on average than the original topography, despite the volume of extraction performed as a result of the creation of a very high volume of pores in a non-compact milieu (1.3 km3, capable of storing the equivalent of one year of precipitation in the area in question). The volumes of extractionbackfill in the Appalachians are so high that they can be compared to those stemming from very large volcanic eruptions like that of Mount Pinatubo [ROS 16]. If the authorities that manage the Appalachian mining basin are not as worried as the inhabitants, and maybe rightly so, the fact remains that these practices have real effects felt at the regional scale. There are numerous situations where serious, brutal impacts affect basins for long distances, particularly when dams holding back wet or liquid waste burst, or even because groundwater accumulated in old mining tunnels destabilizes hills. This question was asked in the Appalachian coal basin. In 1977, as part of the Clean Water Act, a regulatory body was instated, the Surface Mining Control and Reclamation Act (SMCRA). This body was reinforced in July 2015 by the Stream Protection Rule created by the Office of Surface Mining Reclamation and Enforcement (OSM), a service regulated by the Department of the Interior [CRS 17].

104

Sedimentary Crisis at the Global Scale 1

To protect the community’s health, the Stream Protection Rule stipulated the prohibition of mining activities fewer than 30 m from permanent or intermittent rivers, including the deposition of mining waste, the collection of data to measure impacts and measurements to protect flora and fauna; the cost of these measures would be charged to the mining industry. This river protection group was a major stake in the years between the federal administration and courts; the Clean Water Act forbids the dumping of pollutants directly into rivers, but the exemptions accepted or tolerated by the Obama administration made it functionally inoperative. The Stream Protection Rule, a goal of the Obama administration, but revised in a restrictive sense in December 2016, was the subject of a congressional resolution to block the implementation of the regulation; this was immediately repealed by President D. Trump (March 2017), which pleased the mining industry and its workers. The notion of damage to the hydrological equilibrium outside the mining concession area was involved here, i.e. legal recognition of the simple downstream impact on a river subject to dumping or pollutant deposition near riverbanks and likely to affect the water quality. 3.3. Conclusion The meteorological conditions may or may not allow the expression of a rupture of human origin, for example, in the case of soil fragilization following clearing or overexploitation. Erosive crises are significant sources of sediment. The products of erosion may be deposited downstream of land plots, in which case the storage is local (e.g. in the form of colluvia at the foot of a hill). The sediment released often feeds rivers, but the overflowing floods carry it and release part of the suspended load over the floodplain. The remaining material, if it is a coarse bed load, may be deposited in the channel, which undergoes aggradation, at least temporarily; if it is a suspended load, it may instead reach larger rivers in the overall system. If production is particularly high and/or the coastal river is short, the sediment may reach the sea. One way to ask this question is to ask why hydrological and geomorphological science, which ensures the necessary understanding of the processes in their upstream–downstream continuity and over time, has remained so much in the background compared with engineering science for more than 2,000 years; in reality, the understanding of nature is quite recent compared to the historic success of engineering. According to A. James, “the pursuit of practical solutions to immediate problems takes precedence over the search for natural causes” [JAM 99]. The crisis experienced by the Sacramento River and its tributaries between 1843 and 1884 affected the geomorphological compartment as well as the hydrological compartment of the Central Valley’s fluvial system in the decades that followed. It brought about the ever-greater need among the Californian engineering corporation

Exploited Nature and the River’s Responses to the Globe’s Surface

105

to concentrate on the complex hydraulic aspects of controlling floods along rivers, dams and lateral dykes. In doing so, this corporation neglected the full understanding of long time scales and adjustments that were at work in the fluvial system following jet-boring mining. This remark is a bit severe when we think of G.K. Gilbert’s contribution to the subject in the early 20th Century, but the argument holds for the period to follow. A. James believes that the long-term scale – which provides the right perception of geological and historical time scales – is necessary, the former being based on field research and the second on field research and on documentation. The objective is not to replace one science with another, but to combine them in a multi-methodological approach meant to bring two very different time scales closer to one another. It is one thing to construct shared knowledge; it is another altogether to spread acquired knowledge, for this transmission is often inadequate. The facts of nature have been considered in Europe in crisis situations affecting urban property. There was a strong interaction between hydraulic science, advanced local knowledge and a strong political drive. This principle worked regionally in Tuscany, in Bologna and in the Southern Alps, for the reputation of written works or technical successes makes the transmission chain successful, whether it is oral or written in nature. The example of a small alluvial plain in the upper French Rhône, the Chautagne, is highly informative. In the Ancien Régime, a peat marsh was submerged each summer by the flooding Rhône; on the grounds that floods deposited fertilizing silt, the peasant community strongly opposed the construction of dykes meant to “improve” or reclaim the marsh. Modest inhabitants of these commons empirically knew that their way of life would be irrevocably doomed if the silt, capable of providing green fertilizer carried away by the flood, no longer entered the communal pastures. The memory of vernacular heritage is transmitted around the rural world as know-how, just as intellectual memory spreads in educated society. The counterexample could be that when Alexandre Surell, who came from the Vosges to the High Alps to work as an engineer, discovered Fabre and quoted him, but did not really use his work since his predecessor’s work on torrents had not been highly valorized since its publication in 1797. Surell unveiled his own vision of things in 1841. We would have to wait for Joseph Scipion Gras [GRA 50, GRA 57] for torrents to once again find their place in the chain of technical tradition and for the French “treatment” of torrents to really be reintegrated into the mountain land restoration procedure. The similar impermeability of knowledge probably existed in Europe and the United States in the second half of the 19th Century as well. It is curious to note that G.P. Marsh [MAR 74] dedicated many pages to erosion and reforestation in the Alps, yet without these lessons later being used across the Atlantic. As for gold extraction in the Sierra Nevada, it may be best to consider this the effect of scale, although its brutality was easily assimilated to the effects of Alpine debris flow, and

106

Sedimentary Crisis at the Global Scale 1

the gigantism of sedimentary injection due to water under industrial management was not of the same order as extended erosive processes from former times and partly controlled by nature. The Californian mining sector also may have become autonomous within the mountains for decades before being controlled by the public authorities. What is happening today in the Appalachians shows the power of the mining sector in the United States and its ability to free itself from environmental rules considered to be given in Europe.

4 From Hills to the Ocean: Production, Transfer and Trapping

Studies on river contributions to oceans have multiplied since the 1960s, with increasingly sophisticated means, but without the methods or results truly being comparable. The responsibility therefore falls primarily on the scarceness of quality measurements. One of the major observations is the fact that the results obtained at river mouths stand in great contrast to the erosion measurements taken at river basin heads. This point will lead us to inventory natural trapping processes along rivers before comparing the recent statistics that distinguish geological (or “pre-human”) fluxes, fluxes increased by agricultural erosion and, finally, fluxes reduced by trapping in artificial reservoirs. Contributions to oceans fell during the 20th Century and this tendency has continued to steadily decline; we will attempt to understand why. The hydrological and sediment impact of storage dams comes into harmony with the success of soil protection policies and river extraction, which concerns massive amounts of sediment deposited in river channels or transported towards oceans. 4.1. Global continental contributions to oceans Many studies have been conducted since those by French pedologist Frédéric Fournier, who attempted to calculate the first continental erosion statistics by seeking to interpret the sediment balances of 96 rivers chosen primarily from the northern hemisphere using factors like relief and climate [FOU 60]. Today, Fournier is seen as a precursor in the field of scientific study that we are concerned with in this chapter, even though the overall values he produced were distorted by an overrepresentation of heavily loaded rivers, like those in China, and an overrepresentation of highly productive small and medium basins, such as those on

Sedimentary Crisis at the Global Scale 1: Large Rivers, from Abundance to Scarcity, First Edition. Jean-Paul Bravard. © ISTE Ltd 2019. Published by ISTE Ltd and John Wiley & Sons, Inc.

108

Sedimentary Crisis at the Global Scale 1

the Italian peninsula [WAL 96]. Since first being asked, the question has been revisited and refined countless time. The most recent numbers are once again approximations, as measurements are often lacking and discontinuous over time, and sometimes highly unreliable. The data available are much more often those concerning suspended load transportation due to a lack of measurements of bed load; moreover, the fraction of sand may belong to the suspended load when flooding brings about major turbulence; it may also stem from the bed load when the water discharge is lower (in this case, the sand rolls on the channel bottom and is transported in saltation). Since the distinction between bed load and suspended load is rather theoretical, specialists settle on approximations. Finally, the data are not fixed, because they depend on changes in soil use, storage dam construction, etc. The fact that data from different periods are used and averaged is thus often dangerous [WAL 06, WAL 09a]. Whatever the case may be, the numbers are produced and an estimate considers the fact that river transportation is responsible for 95% of the fluxes leaving the continents for oceans [MIL 13]. 4.1.1. Continental denudation and sediment flux to river mouths It is rather rare to measure sediment flow, since only 10% of river basins see measurements taken at their outlets. This is because measurements taken in basins are even more rare and because measurements become less numerous over time. If we look at the Rhône, for example, the only measurement station is the one that was only recently established in Arles, in 2009. It measures the suspended load and a certain number of data, particularly the river’s radionuclide load connected to the nuclear industry, notably at the Marcoule Nuclear Site [EYR 15]. The existing measurements, moreover, are local and discontinuous, generally taken at monthly intervals at French surveillance stations, whereas measurements should clearly be better, particularly on small rivers. This state of affairs prevents flows from being precisely evaluated [MOA 06]. This means that measurements are deficient, even in a country like France (EDF, however, follows alpine rivers), and this flaw does nothing to assist in decision-making with regard to river management. Digital models are used to attempt (though with difficulty) to make up for these deficits [COH 14]. Numerous summarizing studies must face these difficulties; they give suspended sediment transfer values from continents to oceans in a wide range, from 13.5 to 21.0 Gt/year, regardless of origin (Table 4.1). J.P.M. Sivitsky et al. reached a value of 16.2 Gt/year* (without considering trapping by reservoirs at this stage), adding 1.6 Gt for bed load and 2.9 Gt for the dissolved load, which comes to a total of approximately 21 Gt leaving the continents for the globe’s oceans [SYV 05]. As a point of comparison, Ludwig and Probst list a total of 16 Gt/year [LUD 98].

From Hills to the Ocean: Production, Transfer and Trapping

109

Recent values thus substantially lower the 51.1 Gt estimate provided by Fournier in 1960 [FOU 60]. Studies

Gt/year

Fournier (1960)

51.1 (MES) total: 58

Holeman (1968)

20

Schumm (1963)

20.5

Sundborg (1973)

15

Meybeck (1982)

17.5

Milliman and Meade (1983)

13.5

Walling and Webb (1983)

15

Milliman and Sivitsky (1992)

20

Ludwig and Probst (1998)

16

Syvitsky et al. (2005)

17.8

Wilkinson (2007)

21

Table 4.1. Tons of eroded sediment on the Earth’s surface measured at river outlets. These values use different methods and consider an increasing number of rivers around the world

Denudation* is the measurement of the average lowering in the Earth’s surface; it is deduced using measurements of matter exported at a basin’s outlet. These are averages at the continental level or in large river basins. The geography of “chemical” denudation rates is correlated with the climate and nature of the rocks. “Mechanical” denudation, on the other hand, is correlated with the difference in basin elevation, their relief, their slope, the speed at which they are raised by tectonic activity, or even the volume of water available to cause erosion, not to mention the ambient temperature. One important aspect of denudation, which has not been given its rightful place in studies, is the fact that river deposits, stored for millennia, even longer as particles, have undergone chemical processes in the long term. They are weathered in place and lose their cohesion such that part of them feeds the rivers’ dissolved load. This is the case, for example, of sediment deposited millions of years ago by the Orinoco River in the Llanos of the eastern Andean

110

Sedimentary Crisis at the Global Scale 1

piedmont plain, where they are picked up again by the lateral shifting of rivers. American researchers have shown that the unstable fraction of material that is originally from rivers has undergone chemical dissolution and that at 90% they only present a small fraction of sand composed of quartz; these materials have thus lost a quarter of their original mass since settling in place [MEA 07]. The measurement of suspended material at river basin outlets is generally considered to estimate what is known as “river denudation” (wind processes and glaciers make their own contribution, but this is limited). If we maintain a flux value of 21 Gt, this would correspond to an average denudation of the Earth’s surface of 71 mm/1,000 years. Wilkinson et al. [WIL 07] prefer a value of 62 mm/1,000 years, which is close. This average does not reflect the variability of the specific fluxes on the Earth’s surface, which range from 1 to 10,000 t/km2/year! The majority of sediment comes from mountains and, on rare occasions, from orogenic mountain chains subject to strong tectonic activity (Milliman and Meade, 1983). This includes all fractured and brecciated areas, steep slopes, or areas with seismic or volcanic activity [MIL 92]. Approximately a third of the globe’s river sediment is transported by the family of rivers that drain the mountain chains formed by tectonic collisions between the Indian subcontinent and the rest of Asia: these are the Indus, the Ganges, the Brahmaputra, the Irrawaddy, the Salween, the Mekong, the Red River and the Yangzi. Elsewhere, we could add the Fraser (British Colombia), the Orinoco and the Amazon, as well as many rivers further south in the Andes. Some islands like Taiwan have erosion rates between 10 and 100 times the continental average. The most important examples in the Alps are the Rhine and the Rhône. The most relevant explanation for the intensity of transport is not the energy of the relief, but the “degree of tectonism”. This includes catastrophic events such as volcanic eruptions, landslides and earthquakes; after the Kobe earthquake on January 17, 1995, the open tears on the slopes around the city were striking. Milliman and Meade (1983) cite the measurements taken on the Cowlitz, a small tributary of the Columbia that drains the western floodplain of Mount Saint Helens in Washington State (USA). In the four months following the volcano’s eruption, the river’s sediment load rose to 140•106 tons, compared with the Columbia’s average sediment flux of only 10•106 tons/year. In the years following the eruption, however, this flux rose to 35•106 tons/year at the river outlet due to the delayed contributions caused by the eruption of Mount Saint Helens (Figure 4.1). This is a good example of the sediment production that a subduction chain can bring about, at least locally and for a limited time. A sophisticated model, capable of focusing on the regional scale, was used to simulate the water discharge and sediment load in the period 1960–2010 [COH 14].

From Hills to the Ocean: Production, Transfer and Trapping

111

The authors showed that the primary determiner of suspended load variation is the spatial and temporal variability of precipitation, which is, in turn, modulated by the relief. For example, in 1965, Europe’s rivers transported a sediment load that was more than 80% of the annual average; it was transported by a disproportionate water discharge (5% above the average). This discrepancy has been explained by heavy rains on the alpine reliefs and the valleys’ loess soils. We can assume that the nature of soil occupation in Europe also plays an important role.

Figure 4.1. The Toutle River upstream from Camp Baker, invaded by alluvia and dead trees (debris flow). These came from lahars resulting from the eruption of Mount Saint Helens. The Toutle is a tributary of the Cowlitz, which, in turn, flows into the Columbia. The sediment arrived to the Pacific (source: J.-P. Bravard)

The spatial variability of denudation across the globe is impressive, given that the range of global values of mechanical and chemical denudation, based on data measured at coastal river outlets, goes from 4 mm/1,000 years for the Kolyma in Russia (difference in height: 560 m) to 688 mm/1,000 years in the Brahmaputra watershed (difference in height: 2,700 m). Nearly three quarters of the material carried to the oceans has its source in mountains taller than 3,250 m [SUM 94]. With the exception of the Huang-He, the global values for natural denudation pale in comparison with those attributed to soil erosion to the extent that this

112

Sedimentary Crisis at the Global Scale 1

could be measured [WIL 07]. The following data concerns the largest floodplains (Table 4.2).

River

Catchment area (106 km2)

Annual discharge volume (km3/year)

Solid flow rate (106 t/year)

Specific contribution rate to the oceans (t/km2/year)

Amazon

6.15

6,300

1,200

195

Colorado

0.64

18 (A-B)

100 (B)

0.02

Columbia

0.67

251

10 (B)

15

Congo

3.72

1,250

43

12

Danube

0.81

206

67

83

Ganges-Brahmaputra

1.48

971

1,060

716

Huang-He

0.75

49

1,050 (B)

1,400

Indus

0.97

238 (A)

59 (B)

61

Mackenzie

1.81

306

42

23

Mekong

0.79

470

160

202

Mississippi

3.27

580

210

64

Niger

1.21

192

40

33

Nile

3.03

30 (A)

0 (B)

0

Orinoco

0.99

1,100

150

152

Saint Lawrence

1.03

447

4

4

Fraser

0.23

36

17

68

Table 4.2. Current water discharge and sediment load in some of the world’s largest rivers (various authors, see also [MEY 09]). A: current value reduced by irrigation; B: current value reduced by trapping in reservoirs

4.1.2. Natural sediment interception on the way to oceans The erosion of natural milieus and agricultural land is frequently measured over very small areas or experimental “land plots”. Large-scale measurement generally provides high specific erosion values, even very high ones. Values of between 100 and more than 500 tons/ha are thus reported for small Chinese watersheds [WAL 96]. It is true that these rivers descending from the Loess Plateau transport materials coming directly from the slopes of small, sloped basins are that are

From Hills to the Ocean: Production, Transfer and Trapping

113

overexploited and often ill-equipped with trapping structures. Rates of 180 t/ha/year have been observed in the Appalachians (Chapter 3). The extreme values are disproportionate to those of flows measured in rivers and, a fortiori, with the values measured at the stations closest to the outlets of river basins (Figure 4.3). It would not be wrong to predict that the issue of geographic scale is an essential explanatory factor. The question to be asked is thus as follows: how does the material torn from the continents by human activities, particularly those with an agricultural origin, not all reach the oceans? Several factors come together in a particular way in each floodplain, making this question highly complex and making generalization nearly impossible. We will briefly present the major factor, which is natural trapping exercised by the surface of the Earth itself. Transfer and deposition processes act through a series of other processes of a geomorphological nature, integrating the complexity of the Earth’s landforms on which these fluxes circulate.

Figure 4.2. Fine sediment eroded at the surface of immense fields after a storm is partially stored at the bottom of the slope as colluvia and partially exported towards the hydrographic network; the network extends to the upstream interfluve (source: US Department of Agriculture)

114

Sedimentary Crisis at the Global Scale 1

The explanation for this enormous difference between the tonnage eroded on land subject to agricultural practices and that which actually reaches the oceans lies less in storage by artificial reservoirs (recent and small scale in relation to the tonnage in question) than in the following fact: the material eroded along fields do not all go to rivers and oceans, far from it. They are deposited at intermediate sites, for example, at the bottom of slopes (as colluvia), in certain aggraded channels and at the surface of the floodplains or the river margins, the aggradation of which they contribute to (Figure 4.2). They are also deposited in natural lakes, such as the glacial lakes sprinkled across certain regions of North America, or in mountain ranges. The majority of these deposits are formed on the floors of rivers ranking 3–6 (medium-sized rivers)1. Storage is not definitive, as phases of erosion remove a portion (but slowly, as we have seen). We can deduct from this that it is a question of net storage. An expression has been coined to consider the fact that the continent maintains a portion of the sediment eroded in cultivated land parcels: sediment delivery ratio (or SDR)2. The average SDR value is below 10% of the global tonnage mobilized by agricultural practices on the Earth’s surface, estimated at 50–70 Gt/year [WAL 83b]. This rate approaches Fournier’s estimates [FOU 60]. To illustrate this general principle, let us look at the classic example of the Coon Creek basin (360 km2), located on a plateau of the so-called Driftless area in southwestern Wisconsin in the United States [TRI 83]. Spared by Quaternary glaciation, the plateau on which Coon Creek is located is no longer covered by a fragile loess coating, which was ravaged by agricultural erosion during the colonization period. In 1934, it became the first basin to welcome a measurement station set up by the US Department of Agriculture. An estimate of its sedimentary budget for the period 1853–1938 reveals that only 6% of the material eroded in the basin by superficial erosion (and some through gully erosion), leaving it to contribute to the Mississippi’s load, while 38% remained stored in the colluvia at the bottom of the slope; the rest, which formed the majority, was deposited in the floodplains and channels situated at intermediate geographic positions (56%). The estimate for the period 1938–1975, on the other hand, shows a ratio of 7.5% between the eroded tonnage and at the basin exit, and 63% for the part stored as colluvia; the portion stored on the floodplains’ surface having dropped greatly (Figure 4.3).

1 This issue was addressed in Chapter 3 with regard to the erosion of land in Wisconsin and the Loess Plateau. Wilkinson et al. [WIL 07] consider these deposits as post-settlement alluvium. This does not apply to deposition sites in the Old World, where the history of agricultural erosion is infinitely more complex. 2 The notion of SDR goes back to Walling [WAL 83a].

From Hills to the Ocean: Production, Transfer and Trapping

115

Figure 4.3. Sedimentary budget of Coon Creek (Wisconsin) for the period 1853–1993 (source: Science, AAAS [TRI 99], modified)

The portion of eroded material contributing to the river load thus remained very low. Rivers, as emphasized by S.W. Trimble, behave as if they were incapable of exceeding a certain transport capacity, the adjustment to excessive water and load injected into the network taking place through deposition on floodplains, due to the simple fact that the rivers overflow more; the floodplains and basins act as filters upstream from bottle necks, such that the load is reduced downstream. When the input into the fluvial system is reduced, the transfer should theoretically return to equilibrium and the total deposition on the alluvial plains should reduce as an effect of removal by the river. The numbers provided above show that the erosion at the expense of alluvial plain deposits, which must logically accompany a decrease in contributions from upstream, is actually a slow process to implement. For decades, the channel and the banks react by losing sediment, but the plain’s storage remains intact, even for strong flooding in milieus stabilized by silt and vegetation like that in Wisconsin. The erosion of a sinuous riparian strip allowing the disappearance of river deposits whose edification took decades (an aggradation of 5 m took place

116

Sedimentary Crisis at the Global Scale 1

on the plains of the rivers draining the Driftless area) may take millennia. In other words: “Floodplains may be rapid sediment sinks but slower sediment sources, especially in humid areas where extreme events are less common so that vegetation has a greater opportunity to secure and armor the flood plain” [TRI 83]. S.W. Trimble thus challenged the notion of equilibrium (steady state) between the material eroded in an accelerated way in a basin and the material transported by the river at its outlet and forming the river yield. He also emphasizes the fact that pollutants previously deposited at the floodplain and adsorbed by the soil, such as PCB and pesticides, will pose very long-term problems (as in the low valley of the Tyne, Chapter 3); recent pollution, on the other hand, will be stored and not measured at the river unless there is close follow-up. The model that was just presented applies to sectors with steep slopes and limited vegetation coverage in the high basins that serve as the primary providers of load on the Earth’s surface. The load is deposited in stages and decreases downstream; the specific sediment flux likewise decreases downstream [DED 92]. Russian researchers have proposed an inverse model from the one previously presented, which they call “negative”. In basins where the relief’s energy is low and which have maintained dense forest cover, like that of the Yenisei River in Siberia, the material does not come from slope erosion, but rather from the erosion of river channels in proportion with the transported sediment, such that the specific load is a function of the watershed’s surface (“positive” model). A similar result was obtained in British Colombia, where the alpine and subalpine levels provide small quantities of material to rivers due to the poor efficiency of the erosive processes on the substratum and intense storage at the foot of the slope in relation to lower main valleys, where rivers rearrange sediment into a secondary position. This first point was an attempt to present the “global” continental fluxes destined to reach the oceans. Extensive development has shown the way in which sediment, often produced en masse in mountain regions, whether natural or strongly influenced by human action, are transported in the fluvial system. They are deposited, picked up and deposited again (their displacement has been compared with the halting motion created by a conveyor belt). The final conclusion is that the flow from the continents is not of the magnitude of those measured leaving small surfaces equipped with measurement instruments.

From Hills to the Ocean: Production, Transfer and Trapping

117

The following point will show us that sediment is not only largely stored en route, but that factors have played a powerful role in the 20th Century to reduce contributions to the ocean. Three major human factors will be presented. Among other things, they allow us to characterize the period of Earth’s history that has been called the “Anthropocene” for some years now. These are: 1) the effect of reforestation policies or the spontaneous recolonization of land abandoned by human societies; 2) trapping in artificial reservoirs; 3) the extraction of sand and gravel from rivers since they began to be considered resources. 4.1.3. Disturbances in “geological” fluxes during the Anthropocene3 4.1.3.1. A global assessment: geological flows and influenced flows Just 20 years ago, we knew very little about the natural flows of the Earth’s system [MEY 03]. Regarding natural basins, Michel Meybeck could only cite those in Alaska, the Amazon, the Congo and part of Siberia, i.e. 17% of the continents’ surface area; and he also incorporated hot and cold deserts. What is happening on the continents is actually in part a black box that is difficult to enter with certainty. Let us recall where we are at this stage in the analysis: 1) on the one hand, geological erosion has been largely surpassed by agricultural erosion in numerous regions of the humanized globe. In this regard, L.B. Leopold (Chapter 3) admits to a multiplying factor of agricultural erosion between 2 and 50 depending on the American hydrographic basin with a size of 1–10 ha as opposed to pre-settlement conditions; 2) since the low-order drains, materials have become involved in the fluvial system to finish their journey to the ocean. They are deposited along the way to a greater or lesser extent, with the majority settling on the floodplains adjacent to rivers and in deltas or estuaries. This behavior (the relative importance of which is recorded by the sediment delivery ratio) is rather well understood in basins, where quite precise sedimentary budgets have been calculated, but we are far from having a global image of the operations of natural trapping;

3 The Anthropocene is defined as a geological period during which the human race has impacted the surface of the Earth to the extent that it has created a worldwide signal recorded permanently in the geological strata. The starting dates depend on the discipline: between 1950 AD (increase in ocean temperatures) and 6,000 years BP (start of the Neolithic) [CRU 00, MEY 03].

118

Sedimentary Crisis at the Global Scale 1

3) a recent and very promising advance is the calculation of the volumes of sediment trapped in artificial reservoirs. However, before presenting trapping in those reservoirs, let us look at the calculations that have been made in an attempt to compare the current sediment flux and the “geological” or “pre-human” sediment flux (in other words, the flux before the Anthropocene, if we consider this to have begun around 6,000 years BP). One way to distinguish the role of natural erosion and that caused by human activities (regardless of their origin) has been to use a flux model on more than 200 basins capable of determining “pre-human” solid loads; it is based on the energy of the relief, the basin’s surface or its equivalent, the average load and temperatures [SYV 05]. Surprisingly, this study reveals that the modern sediment flux is also dropping with respect to geological flux (also called “pre-human” or “pre-Anthropocene”). In 2005, the highest specific loads (reported on the surface of the continental mass)4 were found in Indonesia and Taiwan, because these lands were severely waterlogged, extensively cleared and still poorly equipped in terms of storage dams; these loads were clearly increasing. The different continents reacted in different ways. By making large reservoirs operational, a strong reduction in fluxes reaching the oceans of Asia (31%) and Africa (25%) was recorded. Europe, Australasia, and the Americas had similar reductions: 12%, 13% and 13%, respectively. Around the globe in 2005, it was interesting to compare the geological flux recreated through modeling, the global flux increased by agricultural use of lands, trapping by artificial reservoirs and, finally, the modern sediment flux [SYV 05] (Figure 4.3): – the geological or pre-Anthropocene flux (also called pre-agriculture/ deforestation) was 14 Gt/year of suspended sediment load, but 15.5 Gt/year including bed load; – the flow without reservoirs, incorporating the agricultural load (and a portion connected more generally with human activities) is 16.2 Gt/year or 17.8 Gt/year including bed load; – large storage dams trap 20% of the global sediment flux and small ones another 6%, or 3.6 Gt/year of suspended material. The total tonnage of sediment stored on the Earth’s surface in reservoirs is 100 Gt; to this, we must add 1–3 Gt of carbon. The majority of this has been stored for around 50 years; – the modern sediment flux (1960–1995) is therefore no greater than 12.6 Gt/year; the reduction in this load is 1.4 Gt/year on the Earth’s surface.

4 It could be preferable to take a series of typical rivers rather than keeping the averages of very diverse continents.

From Hills to the Ocean: Production, Transfer and Trapping

119

Regions

Surface (106 km2)

Water flow rate (km3/year)

“Pre-human” suspended sediment load (Mt/year*)

“Modern” load 1960–1995 (Mt/year)

Africa

20

3,800

1,310 +/– 250

800 +/– 100

Asia

31

9,810

5,450 +/– 1,300

4,740 +/– 800

Australasia (Australia, New Zealand, New Guinea)

4

610

420 +/– 100

390 +/– 40

Europe

10

2,680

920 +/– 210

680 +/– 90

Indonesia

3

4,260

900 +/– 340

1,630 +/– 300

North America

21

5,820

2,350 +/– 110

1,910 +/– 250

Oceania

0.01

20

4 +/– 1

8 +/– 3

South America

17

11,540

2,680 +/– 690

2,450 +/– 310

38,540

14,000 (15,500 with bed load)

12,600

Globe

106

Table 4.3. Suspended sediment load from the continents to the oceans in “pre-human” and “modern” conditions (source: [SYV 05])

A more recent evaluation [SYV 11] (Table 4.4) revised these numbers by increasing geological erosion in Africa, Indonesia and South America, whilst reducing it in Asia, Europe and North America. Erosion under human control has once again risen in Africa and Indonesia (greatly), but it has dropped in Europe and North America. These changes consider climatic and new human parameters. Regions

“Pre-human” suspended load (Mt/year)

“Modern” load 1960–1995 (Mt/year)

Africa

1,600

1,100

Asia

5,300

4,800

Australasia

240

280

Europe

600

400

Indonesia

2,400

2,400

North America

1,700

1,500

Oceania

3

4

South America

3,300

2,400

Globe

15,100

12,800

Table 4.4. Suspended sediment load from the continents to the oceans in “pre-human” and “modern” conditions (source: [SYV 11])

120

Sedimentary Crisis at the Global Scale 1

4.1.3.2. Deforestation and agriculture, one of the origins of the Anthropocene Agricultural erosion is the primary contributor to suspended material transport. An attempt has been made to assess the material mobilized by human activity and particularly by agriculture. Returning to the specific erosion of 35 t/ha/year [PIM 95], the (rather bold) hypothesis was made that this rate can be considered constant in the first seven millennia of agricultural history, regardless of the techniques used throughout the centuries, then decreasing for 2,000 years with the relative development of pastures [HOO 00]; by keeping this rate and using it for the (estimated) population of the globe in different periods, multiplied by the cultivated and pasturage areas necessary for subsistence, a total of 20,000 Gt eroded would be reached5. Agriculture is by far the leading agent shaping the Earth’s surface. Wilkinson and McElroy [WIL 07] emphasize that this total volume is practically 1,000 times the global load that is actually leaving the continents. The erosion rates due to agriculture are by far the highest for which human activities are responsible, even though local actions can be extremely intense; this includes mining, road construction and all work requiring excavated material and backfill. Erosion affects cultivated lands and highly degraded pastures on steep slopes. Concerning this highly controversial topic, Wilkinson and McElroy [WIL 07], after compiling diverse sources, estimate that the average value of denudation of agricultural land across the globe could be reasonably estimated at 60 cm/1,000 years. This is far higher than the average denudation of the Earth’s surface as a whole, but this number does not apply to the entire surface area. If we measure the yield that agriculture might provide to estuaries, it may be as high as 63 Gt/year (or even 75 Gt/year according to Pimentel et al., [PIM 95]); we are way off with the yield from overall denudation (including agriculture), as deduced from the measurements at river outlets (let us recall that this is only 21 Gt/year). For continental lands, specific erosion values on agriculture land, i.e. erosion reported on a given surface (e.g. 1 ha), have often been proposed; they are based on measurements made with rain simulators. These values would be 17 t/ha/year in Europe and the United States (for wind and water erosion), as opposed to 30–40 t/ha/year in Asia, Africa and South America. Around the globe, specific erosion would be 35 t/ha/year according to Pimentel et al. [PIM 95]. To compare, the erosion rate in undisturbed forests is only 0.004–0.05 t/ha/year, and the average for pastures is 10 t/ha/year.

5 However, we would have to subtract the sediment redeposited as colluvia and alluvia, see above.

From Hills to the Ocean: Production, Transfer and Trapping

121

These numbers only have a concrete meaning when compared to the soil formation rate, with pedogenesis primarily taking place through the transformation of parent material into soil: this rate is 1 t/ha/year in temperate countries. We can thus see that erosion exceeds natural soil renewal conditions. To finish, let us note that processes “cascading” in time and space are so complex that we do not know the share of sediment caused by agricultural erosion out of the 21 Gt reaching the oceans. Studies conducted on formations and soil in the Rhône valley have shown that, over the past few millennia, the periods of erosive scouring in areas with heavy human settlement and aggressive climates alternated with periods of pedogenesis during depopulation, reforestation and hydroclimatic calm periods [BER 16]. 4.2. Selected case studies on the Earth’s surface Let us use a few examples of sedimentary budgets taken from regional floodplains. 4.2.1. The Yangzi basin In the 1980s, forest coverage fell to 18% of the Yangzi basin area. In anticipation of the construction of the Three Gorges Dam, erosion studies were conducted as early as 1980 in the upstream basin (over 1 million km2), then producing a sediment flux of 0.52 Gt/year. Erosion in this basin is connected to the subtropical climate with heavy monsoon rains falling on varied soil, locally weakened by agriculture. In the mountains of the Jinsha basin, a large tributary of the Yangzi, a third of the cultivated area is on slopes greater than 25º. The maximum specific erosion rates measured on land parcels are between 90 and 360 t/ha/year; the aggravating factors here being the loss of natural plant coverage and inefficient agricultural practices. Erosion can be seen in ravine dissection or sometimes-enormous mass movements occurring particularly in old, poorly consolidated Quaternary deposits. Chinese scientists use the expression “combined erosion” for the dominant gravitational processes associated with the action of precipitation; debris flows, which often directly feed rivers, are an example [DIN 96]. 4.2.2. The sediment load of rivers in mountain regions subject to tropical cyclones Locally, this can reach specific values around 100 t/ha/year, but this is generally in volcanic regions with lahars, debris flows and landslides. The usual values are clearly lower. Thus, in the Upper Chao Phraya (the basin of which covers 14,000 km2) the load is approximately 1 t/ha/year and did not increase between the late 1950s and the mid-1980s despite heavy deforestation and the expansion of shifting

122

Sedimentary Crisis at the Global Scale 1

or slash-and-burn agriculture* ([ALF 92] in [WAL 06]). The measurements taken in Borneo show very strong acceleration, notably in the traditional pepper plantations in Sarawak, where erosion is 200 times stronger than that in the natural forest. Perhaps paradoxically, slash-and-burn agriculture proves to be relatively unaggressive, the organic debris left on the soil controlling gully erosion on these plots. Tree trunk transportation in forests increases the sediment load by 10–20 times, most changes stemming from improper slope maintenance. This increase nevertheless remains lower than that from urban expansion [DOU 96]. 4.2.3. The effects of the recent protection of degraded continental milieus Soil conservation policies are becoming more popular in Asia. Owing to the very fact that measurements have been taken at the source on small experimental parcels of land, it is difficult to extrapolate for larger basins. Moreover, since the reduced load reaching outlets combines several factors that are difficult to separate, the numbers must be taken with a grain of salt. It is in the Loess Plateau region in the mid-Huang-He basin that measurements were most extensive and the positive effects were the most convincing [WAL 11]. 4.2.3.1. China: the Loess Plateau (Huang-He) Since the foundation of the People’s Republic of China, the government has undertaken extensive works on the plateau, building reservoirs on the tributaries to reduce the transfer down the Huang-He. This river was threatened by dyke breaching, and extremely expensive works with the aim of containing the sediment inside the dykes have been undertaken. The sediment in the channel and at its edges are sand and therefore it is easily mobilized and suspended, because the turbulence of the flooding river is reduced. One part of the solution can be found on the upper Loess Plateau, where more than 100,000 km2 of 430,000 are the source of erosion and exceptionally significant transfer. Three quarters of the river’s solid load originates there, of which 0.730 Gt/year is composed of sandy and silty sediment. The most extreme erosion phenomena can be found in the northeastern part of the Loess Plateau in the form of gully erosion and mass movement. Land protection has required the construction of stepped terraces, the “replacement of agricultural land by forests or grass”, the construction of tens of thousands of loess “traps” (such as check dams) and, most importantly, the construction of more than 2,000 reservoirs since 1970 (685 of which measure more than 1 million m3). The effects of these works were measured on the densest measurement network in the world: between the period 1950–1969 and that of 1970–1984, precipitation dropped by

From Hills to the Ocean: Production, Transfer and Trapping

123

15% and the sediment load by more than a third [GON 87]. The greatest load decrease is interesting in and of itself, but it is clear that erosion decreases with rain (18 to 20%), which would result in 16–19% for the positive effects of soil protection work6. More than the works concerning the direct protection of soil, it is clear that check dams (on average, 15 m tall in this region) and reservoirs have been the most efficient means of action. In the Wuling basin, the transfer was reduced by 62%. This breaks down as follows: 49% for check dams, 9% for reservoirs, and lastly terraces and soil relocation (4%). Check dams allow the partial backfill of ravines and reduce the relative height of their steep slopes. Even though heavy flooding causes their collapse, they are far from being able to evacuate all of the silty sand deposited and fixed thanks to the cohesion of sediment. Moreover, the fact that the Shaanxi check dams are more than 200 years old is rather reassuring with regard to the longevity of this policy. A recent study was conducted in the semi-arid loess Huangfuchuan basin (3,246 km2), where the average soil loss is 125 t/ha/year [ZHA 17]. This study accepts that erosion control efforts have been performed over large areas, but that it is still unclear if they will affect the river’s load in the Huang-He basin. Movement from one spatial scale to another has taken place through the use of models. From 1990 to 2006, bare soil was converted into pastures and reforested, whereas 502 dams were built, providing a total capacity of 571 Mm3. The results are promising, with a 31.4% load reduction without dams and 51.9% with them, though without soil conservation measures. The cumulative effects of the two practices reduced the flow by approximately 80%. The limits of the dam solution, these already storing 571 km3 in 1969, are emphasized, because they are reaching the end of their lifetimes, becoming fragile and needing to be maintained or to undergo sediment flushing or removal by mechanical means. This is an undeniable success, but will it last? The reversal of the tendency that leads to increasing degradation will, in principle, be difficult to obtain. The future depends on complex climatic variables and interactions with yet-to-be-seen human activity. A recent study concerning the Yan, a direct tributary of the Huang-He draining the Loess Plateau, revealed that the water discharge and sediment load are highly variable from one year to another due to interannual climate variation and the fact that the two rivers incised their bed significantly between 1950 and 2010, but in a nonlinear fashion. This would translate the positive effect of the soil and sediment conservation methods (afforestation, terrace farming, industrial crops, sediment retention behind check dams and in reservoirs) which are addressed in section 4.2.3.2. [WAN 13a]. According to some scientists’, the situation has lost its sustainable character due to the limited lifetime of check dams [WAN 06]. 6 The authors of this study accept the proportionality between the natural reduction in rain and that of erosion; it has not been proven.

124

Sedimentary Crisis at the Global Scale 1

In this uncertain context, how significant are the effects of reforestation, particularly those involved in the Green Great Wall in China (or “Three-North Shelter Forest Program”), a project launched in 1978 in the hope of reforesting 1.5 million km2, i.e. 15% of the surface, by 2050? Is not the effect of this gigantic reforestation effort, if it is positive vis-à-vis erosion control, cancelled out by excessive water consumption? The Loess Plateau forests, which covered 6% of the soil in 1950, had expanded to 26% in 2010. Is the extreme drop in the Huang-He’s output due only to climatic variation? It must be partially due to the evapotranspiration of this quickly growing forest made up of exotic species that consume great quantities of water, in which case the effect of afforestation could be “disastrous” for the future of irrigated agriculture on the Yellow River’s plain, after having significantly reduced the Plateau’s agricultural surface. The search for synergies between the environment, economy and societies is seen as desirable [SCH 17]. 4.2.3.2. China: the Yangzi basin The eroded area exploded in the Yangzi basin between the 1950s and the 1980s, increasing from approximately 360 to 560 km2. The soil conservation policy used starting in the late 1980s has since decreased this area. Owing to this, the river’s sediment load in Datong, the river’s last station before its outlet, has decreased from 2.3 Gt/year in the late 1980s to 0.9 Gt/year in the period 2003–2012. The soil conservation policy itself would explain 10% of the decrease in transit in Datong (i.e. a decrease of 0.018 Gt/year). Chinese scientists are well aware of the issue of spatial scale taken into account. When speaking of small, well-studied basins, the reduced erosion has been measured, with values of 30 to 70%, but the transport recorded at the largest basins’ outlets is quite different and cannot be “discernable”, at least not in the short term. In fact, artificial reservoirs play an important role in sediment trapping (see the following). The concrete effect of storage dams and small dams built to trap the load provides an example of this. A total of 12,000 reservoirs were built with a capacity of 18.9 km3 in 1995 – 7 years before finishing the Three Gorges Dam. These reservoirs trap 10.5% of the materials produced by soil erosion in the upper basin, but the reduced load recorded in Ychang (at the border between the upper and lower Yangzi) is only 3–4% of the average annual load; in other words, reservoirs only control a portion of the upper Yangzi basin [DIN 96]. It is also interesting to see how land protection has been dealt with in the United States since the serious crisis of the 1930s which affected both water discharge and sediment load. Two classical studies were conducted on the Southern Piedmont of the Appalachians and in the Coon Creek basin in Wisconsin (see Chapter 3).

From Hills to the Ocean: Production, Transfer and Trapping

125

The Southern Piedmont is one of the lands on which reforestation policy has been tested regarding its effects on water fluxes, knowing that previous studies performed in the Northern Appalachians had already provided promising results in the 1960s. The Southern Piedmont is a low, dissected and wet plateau (1,250–1,500 mm/year) located between the Blue Ridge Mountains and the Atlantic Coast. The area studied includes part of Georgia and South Carolina and encompasses equipped watersheds that have enabled diachronic comparisons regarding a cumulative area of nearly 55,000 km2 between 1920–1940 and 1955–1975 [TRI 87]. The results indicate a slight variation between basins as opposed to the results seen when comparing experimental parcels of land. It is an undeniable fact that reforestation, when covering 10–30% of the area, reduces water discharge significantly, and this is particularly true during droughts. The average drop in eroded amounts is even recorded at 0.3 m3/m2 of reforested areas and the water discharge is reduced by 4–21%, which also poses problems for water supply at the regional scale, hydroelectric production or pollution dilution. We could add to this list the decrease in ability to transport sediment, but it was not simply the object of the study to foresee this. The agricultural practices in the Driftless area (see section 4.1.2) were revolutionized less by a change in crops than by the adoption of new techniques: once again contour plowing and seeding, crop rotation, the use of protective plant waste, planting clover and controlled pasturage. Between 1934 and 1975, the average loss of land on elevated slopes, measured using an equation, was divided by four since it dropped from 30 to 7.20 t/ha/year. All things remaining the same in terms of climate, the backfill of small reservoirs downstream from planted parcels, converted into a specific erosion rate, dropped from 48 to 0.63 t/ha/year; the deposit rates on floodplains, on the other hand, fell from 37 to 0.30–0.70 t/ha/year, i.e. from 1 to 2% of the values from the 1930s [TRI 82]. Through these examples, we can see that erosion control policies can have spectacular effects. 4.2.4. Mining and the increase in river loads Mining activity is relatively temporary in terms of its occupation of space, but it has powerful effects, as we saw regarding gold extraction in the Sierra Nevada. Let us look at specific examples from regional watersheds. The basin of the Magdalena River in Colombia, covering an area of 250,000 km2, transports 9% of the load going to the Caribbean Sea by itself. The measured load increased by 40–45% during the period 1975–1995 as a result of clearing, more intense usage and gold extraction.

126

Sedimentary Crisis at the Global Scale 1

In Western Siberia, the Srednekansk station upstream of the Kolyma basin (99,400 km2) saw an uncontested increase in sediment load between 1941–1988, the values being multiplied by 1.5 between 1964–1988. With no precipitation variation in this period, the increased load is attributed to the impact of gold mining in the watershed7. An even more spectacular example is that of the Fly River watershed in Papua New Guinea. Until 1985, the basin was natural and its load was around 10•106 t/year. In early 1985, a new gold and copper mine was opened, the Ok Tedi, which bears the name of a tributary to the Fly. Each year, 90 million tons of mining debris and treatment waste were poured into the river; this material was transported by floods and 70% of this load was deposited, primarily on the Ok Tedi’s floodplain. 30% of the material reached the Fly, increasing its load to 35•106 tons/year. The Fly in turn flows into the Strickland, whose sediment load reaching the ocean increased from 85•106 t/year before 1985 to 120•106 t/year (+40%), a great example of the domino effect connected to mining extraction [WAL 06]. Extractions are thus highly likely to radically modify the budget of certain large rivers. Their relative contribution on a global scale remains yet to be seen. 4.3. Irreversible flux disturbances 4.3.1. The major role of artificial reservoirs A major breakthrough came about in the late 19th Century with the hydraulic development of large rivers, then smaller rivers as well. This breakthrough came in stages. Before the 1920s, hydroelectric power was produced under high waterfalls using the model developed in 1869 by Aristide Bergès in Lancey, at the foot of the Belledonne Mountains, near Grenoble; it was also produced through the technological adaptation of watermills. Late-19th Century factories combined energy production downstream of a diversion channel fed by a rise in the water level brought about by a river dam. This dam, equipped with mobile sluice gates, also served to allow floodwater and a significant sediment load. In a way that seems unusual to us today, Élisée Reclus noted that people had a much better idea of how to use stream waters than large rivers: “Barely one thousandth of its force is used for industry; its waters, far from flooding over the fields in fertilizing canals, are instead edged out by lateral dykes and pointlessly contained in their channel. Whereas the stream is already part of the history of humankind during 7 Cited by Walling [WAL 06].

From Hills to the Ocean: Production, Transfer and Trapping

127

the industrial era, the most advanced of all, the river represents little more than a very old era of societies, one where rivers only served to keep boats afloat” [REC 69, p. 208]. Reclus emphasized that the mastery of river waters implied that the technique would one day allow “the discharge to be regulated according to needs” and that associations would be formed to allow “the regularization of the still-brutal force of water mass”. At the time, dams were not yet conceived of, but things would change to the extreme in the late 19th Century. After World War I, the lack of energy was the motivating factor for the construction of storage dams. These would henceforth trap the bed load and part of the suspended load both upstream and downstream of the network, on torrents as well as on the Rhône (the Génissiat Dam). The advantage was to use the entirety of the river’s discharge, particularly flood discharge, wasted by the aforementioned technique, whereas storage dams trapped the floodwaters, at least partially. Extraction ahead of reservoirs was henceforth (and often) considered a need to slow down its backfill with bed load. The gravel flux became cumbersome. In the lower Isère and the Rhône on its floodplain, the choice of low impoundments and bypass channels, made necessary by the impossibility of building large reservoirs, posed the question of gravel in a new way: bed load was not allowed to enter the Rhône’s by-passed channels, which had low energy and were incapable of downstream transport. This regional situation illustrates the fact that the succession of technical choices, as well as the material demand, motivated the organized depletion of the bed load in many European rivers, not to mention other basins around the world [BRA 16]. One of the first works highlighting the responsibility of reservoirs in the generalized change to continental trapping (this switchover having taken place just twenty years ago) introduced the beautiful neologism, “neo-castorization” [VÖR 97]. The globe then had more than 36,000 “large dams”8 and the authors of the report hoped to record a strong human signal in the sediment cycle. The number of operating dams in the world is now more than 48,000. Before 1950, China had eight dams; today, it has more than 18,600, i.e. 55% of the worldwide total [SYV 11]. 4.3.2. Hydrological and sedimentary effects Vörösmarty et al. estimated that a water volume of 15,800 km3 had been trapped in the mid-1990s; in other words, more than 40% of the globe’s annual water flow was intercepted by more than 600 large reservoirs [VÖR 03]. The flow intercepted

8 A “large dam” is a structure measuring more than 15 m high and/or controlling a retention area of at least 23 km2.

128

Sedimentary Crisis at the Global Scale 1

in 2012 has increased again, corresponding to approximately half of the global flow [WAL 12].

Figure 4.4. Located just downstream of the Sesan and the Srepok Rivers, the Sesan II Dam was built by a Chinese group and made operational in 2017. It creates a 75 km2 reservoir in which water loaded with suspended sediment from the Srepok, descending from the Annamite Range (Vietnam) settles. Sesan II reduces the sediment contributions to the Mekong and fish productivity (source: NASA/Earth Observatory, 2018). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

One hydrological aspect of the effects of the more or less intense sedimentary backfilling of reservoirs is the loss of the water’s storage capacity in a number of artificial reservoirs (Figure 4.4). From this, we can understand the full importance of human uses that are threatened in the long term as well as when we consider the rarefaction of sites likely to retain water. There would be a total loss of 0.5–1% of the total volume per year around the globe (i.e. 48 km3 corresponding to 60 Gt/year of deposited sediment). At this rate, the average lifetime of reservoirs would be 100–200 years. Given that the best sites are already taken, what will we do a century from now? How can we reconcile this significant loss with the estimated reduction in continental transfer to oceans? The loss directly connected to reservoir backfilling would have taken place (at least in part) through deposition on the floodplains, as

From Hills to the Ocean: Production, Transfer and Trapping

129

stated by the principle of conveyance loss generally accepted by specialists (i.e. 40–50%). Moreover, increased anthropic erosion in certain regions of the globe could be compensated through trapping in reservoirs; some works have been dedicated to retaining products of erosion to better preserve strategic development schemes. The fact remains, however, that the difference between actual trapping in reservoirs and the lacking fluxes entering the ocean is still very high in favor of the latter and has not really been explained, which relativizes all the produced data with often incomparable methods [WAL 12]. Dams have many effects: these are of a hydrological, sedimentary, geomorphological and, finally, ecological nature – a point which is not addressed in this book [COO 96]9. 4.3.3. Trapping and effects on sediment transfer Dam efficiency in terms of sediment transfer interception depends on numerous factors. Above all else, the ability to intercept depends on the relationship between the reservoir’s volume and the river’s water flow: a high reservoir volume/annual discharge ratio is responsible for significant trapping; this ratio conditions the water’s residence time. The higher it is, the greater the sediment trapping [VÖR 97]. Interception also depends on dams’ positions in the basin and the intercepted area they control. The further downstream dams are, the more efficient they are, and the more they are aligned in a chain, the more difficult it is to make them transparent regarding sediment. The most remarkable example is likely that of the Ebro (Chapter 5), whereas the Rhône is much more complex and flexible; the Nile has a dam that was placed 1,200 km from the sea, but the almost complete absence of tributaries on this stretch makes it more like the Ebro. In reservoirs, the trapping efficiency exceeded 50% in 1997; in other words, half of the sediment from upstream of the river basin was trapped, but the trapping efficiency could actually be 85% in 2003, according to the same study [VÖR 03]. However, the average trapping in reservoirs is one thing (and are still rather poorly studied), and the load reduction in a partially equipped basin is quite another. In 1997, for the Earth’s entire system, reservoirs were meant to trap sediment amounting to 16% of the global flux towards oceans, but this number was likely higher, as it was not possible to actually consider all the globe’s reservoirs. In 2003, a GIS approach locating 633 large reservoirs and 40,000 small ones provided a water discharge interception percentage of 40% and a range of 25–30% for the

9 A simplified approach for the United States (particularly Colorado, Chapter 5) [COO 96].

130

Sedimentary Crisis at the Global Scale 1

sediment flux reduction, i.e. 4–5 Gt/year for a total of 15–20 Gt/year reaching the oceans. Syvitski et al. are not far from this estimate, with 20% or 26% depending on whether only large dams or the small ones are also considered [SYV 05]; in this case, trapping would amount to 4.4 Gt/year for a total annual flux to the oceans of 12.6 Gt/year. More recent estimates assuming a natural flux of 15.1 Gt/year for selected rivers showed a flux reduction of between 10–15%, but the corpus of dams is partial [SYV 11, WAL 12]. River

Country

Sediment load reduction (%)

Colorado Nile Cauvery Krishna Asi Rio Grande Indus Sebou Ebro Volta Chao Phraya Limpopo Zambezi Orange Narmada Godavari Red Total

Mexico Egypt India India Turkey United States Pakistan Morocco Spain Ghana Thailand Mozambique Mozambique South Africa India India Vietnam

100 100 99 98 98 97 96 95 93 92 90 82 81 81 79 72 60

Load reduction (Mt/year) 120 120 32 63 19 19 240 35 16 17 27 27 39 72 55 123 60 1,395

Table 4.5. Estimate of the load reduction experienced by the world’s largest rivers (sources: data from Milliman and Farnworth [MIL 13] and Walling [WAL 12])

The reduction of transport capacity downstream of a dam or a chain thereof contributes to sustainable trapping sediment in the channel, thus reducing their contributions to oceans. The load of the Danube entering the Black Sea dropped by 30% from its values in the 1950s at the station in Ceatal Izmail in Romania; the cause for this was a multitude of small dams built in the 800,000 km2 of the watershed and particularly the effect of the Iron Gates Dam (1972). Inversely, limited change in the hydrological regime throws off the balance of water discharge/sediment load in favor of increased net energy downstream of the

From Hills to the Ocean: Production, Transfer and Trapping

131

dam, thus triggering a sediment release to the channel’s detriment (water deprived of sediment displays increased energy and applies it to eroding the riverbed, on the floor and edges when possible). In fact, there are many scenarios based on the one observed; this view of matters goes beyond our aim [BRA 00, PET 84]. On a regional scale, other examples are significant. Let us consider the example of the São Francisco River, which drains 8% of Brazil, or 645,000 km2. Hydroelectric development consisted of building a chain of storage dams and a series of other dams on tributaries. The Sobradinho reservoir alone, completed in 1978, covers 4,220 km2. In total, the load reduction on the São Francisco is 95% at its outlet. The sediment discharge of the Chao Phraya in Thailand has declined by nearly 80%, from 28•106 t/year around 1970 to 6•106 t/year in the 1990s. The water discharge being little reduced, contrary to that of the large Chinese rivers, the responsibility for this imbalance doubtlessly falls upon the Bhumibol and Sirikit reservoirs, which were built to produce electricity and provide irrigation water [WAL 06]. 4.3.4. River diversion, loss of transport capacity and trapping One aspect of the problems found in sedimentary blocking due to human impacts is connected to the diversion of agricultural or industrial water from rivers to sectors where it is used. Let us also mention the voluntary alteration in the hydrological regime, which aims to store floodwater for economic uses and protection against large floods (these two functions often complement one another). Two scenarios can present themselves: – water intake can be conditioned to reduce sediment input into canals in order to extend their lifetime. However, the diversion of relatively clear water reduces transport capacity on the drained river; the logical effect of this water–sediment imbalance is deposition in the canal downstream of intake. Such is the case of the downstream stretch of the Indus in Pakistan; – if intake does not decant the water removed from the canal, the sediment is removed to the detriment of the flux towards the ocean. This can be seen in the Huang-He in Northern China. 4.3.4.1. The Indus and irrigation The Indus, measuring 3,180 km, and its large Indian tributaries find their source in the Hindu Kush, the Karakoram and the Himalaya Mountains; the river flows into the Arabian Sea.

132

Sedimentary Crisis at the Global Scale 1

In its long, narrow basin, the Indus has changed routes several times throughout the millennia, sweeping over an alluvial fan 50–100 km long. Downstream of the Thar Desert, the Indus has migrated to the west, abandoning all of its more easterly routes after 900 AD (the Eastern Nara and the Khaipur–Shahdadpur, which function seasonally during flooding); it would be helpful to see, in these avulsions, the effect of neotectonics, which would have slowly bent the Indian Plateau east of the basin, the effect of competition between broad alluvial fans, the aggradation of channels on their alluvia, and their overflow, not to mention earthquakes causing breaches in the natural levees. Damming the primary channel for 150 years has placed the channel at such a level that a strong flood can produce spillage onto the plain and catastrophic flooding, a situation comparable to that of the Huang-He around 1855 (see Volume 2, Chapter 2). The Indus receives adequate feeding from melting snow and summer monsoon waters on mountains (the heavy loads and floods are from May–October), then crosses the dry regions before building its delta at its outlet to the Indian Ocean (Figure 4.5). The glaciers make significant contributions to load genesis. The Indus’ natural flow into its downstream course was 207 km3/year with maximums of 30,000 m3/s. The load may have varied between 300 and 1,100 Mt/year in the Holocene and sediment feeds were connected to the rupture of dams formed by enormous landslides and erosion on terraces that were formed by temporary natural reservoirs [CLI 13, SYV 14].

Figure 4.5. Simplified schema of the Indus’ basin and hydrographic network. The regions in orange do not provide water to the Indus, which depends on contributions from the mountain ranges (source: Keenan Pepper, Wikimedia Commons). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

From Hills to the Ocean: Production, Transfer and Trapping

133

On the floodplains, which occupy a quarter of the country’s area, agriculture has been and remains an essential sector of Pakistan’s economy, if we consider the fact that the rural population exceeds 90 million and represents approximately 60% of the country’s total. Dams and canals were built at the instigation of the British for more than 150 years in order to reduce flooding from summer monsoons and to contain irrigation water form the largest irrigation system in the world, the Indus Basin Irrigation System. The upstream basin was equipped with large dams like the Mangla on the Jhelum (1967) and the Tarbela on the Indus (1971) – dams capable of containing floods. It is therefore the upstream sector that is capable of holding water. A second series of dams was built further downstream, not to store, but to divert water towards the channels in the downstream part of the alluvial plain, the Sindh, where precipitation is only 100–200 mm/year. The British built the Sukkur (1923– 1932) and Guddu (1957–1962) Dams. The Kotri Dam (1955–1961) has played a significant role, because it is the furthest downstream. Part of the upstream basin of the Indus has been located in India since the division in 1947, such that the international status of the water there is uncertain, even though the 1960 Indus Water Treaty regulates the division of water between the two countries in principle.

Figure 4.6. The Guddu Dam (1962), located on the border of the provinces Punjab and Sindh, has a dual function: containing floodwater and allowing for irrigation. The low waters from early June 2009 go through the dam and there is limited containment before the monsoon floods (source: NASA Earth Observatory, Landsat 5)

134

Sedimentary Crisis at the Global Scale 1

Reservoirs strongly influence the flow diverted in favor of a network of 43 large irrigation canals measuring 57,000 km, developed by the British since the mid-19th Century (Figure 4.6). This network feeds 80% of the agricultural land in Pakistan and diverts 42 km3/year towards India. The large storage dams (Sukkur, 1932; Kotri, 1961; Mangla and Tarbela, 1976) have reduced the water discharge to the ocean. From 220 km3/year in the mid-19th Century, it has dropped to less than 10 km3/year at the outlet. The number of days without a flow downstream of the Kotri Dam has increased from 0 in 1955 to 100 between 1962–1967 and 250 after the Mangla storage dam was made operational, allowing irrigation to other agricultural areas [INA 07]. The sediment flux was very significant before large hydraulic works were undertaken, with approximately 250 Mt/year reaching the ocean of the 600 Mt/year leaving the mountains, which highlights the importance of alluvial trapping during the Holocene. The delta, due to heavy deposits on the alluvial plain, ensured by the meandering channels, now only receives 0.25•106 t/year (essentially silt, sand, and clay) and the sediment contribution to the Indian Ocean has dropped by 96%. The systemic diversion of waters prevents all transport for two-thirds of the year [GIO 06, WAL 12]. The effects of reservoirs have been positive for several sectors of Pakistan’s economy, but the social effects have been catastrophic in terms of displaced persons, notably after the beginning of operations of the Tarbela Dam, which also resulted in the great inequalities in the division of benefits. Heavy water consumption and continental trapping of sediment, both in reservoirs, in channels and irrigated perimeters, have desertified the Indus delta, with very little hope of improvement, regardless of the policies implemented in the future (see Volume 2, Chapter 3). 4.3.4.2. The Huang-He and water diversion The loss of sediment along the downstream stretch of the Huang-He between the station at Huayuankou (at the entry onto the plain and downstream of the reservoirs; altitude: 110 m) and that of Lijin (station located in the delta 100 km from the outlet) has been the subject of more or less competing interpretations. Over nearly 800 km, the water discharge decreases by nearly 20% (from 30 to 32 km3/year). This drop is explained primarily by the diversion of river water towards the adjacent regions; it is also caused by the drop in precipitation and by water consumption in the reservoirs, not to mention increasing consumption by forests planted in the tributary basins.

From Hills to the Ocean: Production, Transfer and Trapping

135

As for the sediment load, it decreases from 0.97 Gt/year in Huayuankou to 0.8 Gt/year in Lijin (1950–2005). If the loss of sediment between the two stations has been attributed to the set of deposition and erosion processes in this long stretch of floodplain that experiences inexorable aggradation, another explanation deserves to be mentioned, that of sediment impact from water diversion. A recent study, after having presented previously the data, innovates in terms of interpretation [WAN 07]. This study calculates that there is a 20% loss of sediment between Huayuankou and Lijin following the diversion of 10 km3/year since the 1970s (a stable figure for 30 years now). 4.3.5. Predation of river resources: sand and gravel The predation of natural river resources, gravel and sand means that they are removed from their environment. This is not a form of trapping at an intermediate, natural or anthropic site, but a subtraction performed to the benefit of basin sites (backfill and construction) and sometimes sites outside the basin harmed by maritime transport. What matters here is that the “removal” is matter that will not reach the ocean. It is thus worth tackling this subject. The invention of concrete is correctly celebrated as one of the great innovations of the modern world, allowing architectural prowess and accompanying the visual modernity of contemporary architecture. Nevertheless, the environmental effects of the construction and public works sector that provides the aggregate, not to mention the binder that is cement, are aspects that are rarely cosidered. The production and transport of the components as well as the preparation of the concrete has an economic cost incorporated into the final price. A recent book on the glory of concrete [GEN 15] only mentions the environmental issue in a few lines with regard to human health in the implementation of concrete (which is a legitimate concern) but it is insufficient. It is paradoxical to boast the vernacular origin of traditional materials (soil, wood) without sand and gravel being mentioned as well; moreover, these two materials “borrowed” from the site when exploited (which are not always restored and, mostly created in the Holocene) are not renewable. The impacts of extraction are the incision of rivers, the reduction of their load, downstream impacts and, finally, the effects on the rivers’ free space due to the fact that it is generally necessary to isolate the river from extraction quarries dug into the floodplain. The profession of a dredger, the excess of which has been called into question, is more and more constrained by restrictive regulations. Their treatment of extraction has also become much better. However, the construction branch, as well as urban planners and architects, are reluctant to imagine the origin of their favorite material,

136

Sedimentary Crisis at the Global Scale 1

as if its origin was neutral or did not concern them. A round-table discussion held in 2016 at the Gallo-Roman Museum in Lyon, rightly or wrongly claimed that concrete is a French particularity and advocated for the French construction and public works sector (which is a global leader, built on ideas of the national glory of architecture and demonstrating mastery of remarkably complex techniques). However, the damage to the environment is covered by a chaste veil. This form of “cognitive dissonance” is even more remarkable when considering that the extraction sector always lies at the heart of complexes of powerful economic and environmental interests, both local and regional in nature, which are fundamental to the equilibrium of an economic sector conceived as integrated. 4.3.5.1. General data The excessive extraction of sand and gravel (called “aggregate”) has become a global issue. The tonnage of all sorts of materials extracted annually around the globe, in rivers and on the coasts, is estimated at between 47–59 Gt (billions of tons); aggregates represent between 68–85% of the mass of resources extracted (stone and cement); however, it is also the mediocre quality of the statistics that has blinded many from the gravity of the situation, with dredgers demonstrating great modesty in their statements [UNE 14]. Aggregates are used in a variety of fields: to manufacture concrete (this number is slightly more precise: between 25.9–29.6 Gt/year worldwide); for backfill and road surfacing; and for diverse industrial purposes. In total, sand and gravel extraction exceeds 40 Gt/year worldwide, double the rivers’ contributions to oceans (but of the 20 Gt/year arriving to the oceans, only 2 Gt is sand and gravel [UNE 14]). This approximation is somewhat questionable, as the extractions certainly take place in river channels and valleys, but also, to a certain extent, in river terraces inherited from the Quaternary Period. On the other hand, the contributions to oceans include all suspended material (sand, silt and clay). The portion of sand is generally in the minority (gravel and/or sand bed load is between 1–10% of the total load); it is thus much more likely that sand and gravel extraction exceeds at least 10 times river contributions of the same nature to oceans. There are unbelievable amounts of sand available around the globe. The cost of transport could be overcome in some cases, but in a country like Algeria, the prices of “classical” aggregate (i.e. that adapted to construction) have exploded due to the scarcity and cost of transport. Why, then, do not we use dune sand to mix concrete, to obtain “sand concrete”? Sand covers 60% of Algeria’s territory and is very abundant in the Middle East. In the United Arab Emirates, sand is in demand to create artificial islands and beaches. To create the small islands of Palm Jumeirah in Dubai, 186.5 Mm3 of sand was required, whereas Palm Jebel Ali, followed by the World Island Project, required 225 Mm3. The issue is mixing concrete for the

From Hills to the Ocean: Production, Transfer and Trapping

137

construction sector, which is booming. Sand had to be brought from Australia to build the Burj Khalifa. The response is that desert sand is not ideal for mixing concrete, despite its purity. The grains of a sand dune are fine and homometric10, which prevents cohesion of material used for concrete and its stability when compressed (the weight of a building, for example, or that of the load supported by a viaduct), even after mixing it with cement. It is, however, possible to create good “sand concrete” for low buildings and those with limited constraints, so long as it skillfully worked in throughout the process, for example, by improving its texture through the addition of crushed limestone to expand the granulometric range or by adding sawmill waste treated with hydration and drying cycles [BED 12]. 4.3.5.2. The situation in France and Europe The significant quantity of bed load extraction in rivers increased in the 19th Century, an era when Europe was rebuilding its roads and creating railway embankments. Roadwork required gravel from rivers to fill highways. Massive and still scarcely documented extraction took place to build roads and railways on floodplains; this was the case for hundreds of kilometers of railways in the 19th Century, along the Rhône, for example. The use of “aggregate” to mix concrete for construction became a general practice in the second half of the 19th Century and then, in the 1950s, it became important in river beds near cities undergoing massive demographic expansion. This process was forbidden in France in the 1990s. The geography of extraction has changed in Europe since the 1950s. Great demands were placed upon riverbeds, as gravel was cheaply washed by the flow of water and the sand present in the matrix of materials on its floor was easily separated from the gravel. Extraction policies became more restrictive when it was proven that river extraction and incision brought costly negative impacts: geotechnical impacts (threats towards bank defenses and bridges), impacts on groundwater and potable water feeds, and ecological impacts [BRA 97]. Extractions were then carried out on the pebbly-sandy layer that forms the substrate of floodplains. Inherited from former channel migration, this resource is practically non-renewable through the centuries. The landscape produced, even though it is claimed by lakes and some ecologists (from continental regions) who praise their virtues for the avifauna, does not fail to arouse opposition. Finally, the surveillance of extraction is easier in these cases than in new sites on coasts. The situation is now growing worse on beaches, foreshores and even deeper (even though sand enriched in sea salt is not recommended for mixing concrete).

10 The smooth grains are all the same size, so gaps form a significant part of the volume.

138

Sedimentary Crisis at the Global Scale 1

The low-quality uses of materials (e.g. backfill) have remained majoritarian until recently. Great demands were placed on the rivers of Northern Italy starting in the 1920s, the year 1924 being the date the first highway in the world, the autostrada dei Laghi connecting Milan to Lake Como and Lake Maggiore, was made operational. The year 1935 saw the opening of the Camionale, connecting the port of Genoa to Serravalle through the Apennines, before later connecting Genoa to Milan and Torino. If the universities of the Po floodplain were pioneers concerning the study of mining impacts, this is because the torrential rivers there were overexploited. In Germany, the rivers of the Bavarian piedmont plain were strongly affected starting in the 1930s under Hitler’s regime. In the Lower Rhône, the Arve and the Fier Rivers, as well as the Rhône’s tributaries downstream of Lyon, were overexploited after World War II for similar purposes (backfill for highways and nuclear power plants, construction, etc.). In the Seine basin, river extraction of aggregate amounted to 50 Mt in the 1950s, 60 times more than the sediment flux to the English Channel; the landscape of open borrow pits on the floodplain is the result thereof [MEY 98]. The alluvial plain of the Gave de Pau River in the department of the PyrénéesAtlantiques provides a perfect example of the institutional organization that presided over mining in river milieus in the Béarn, a small region of Southwestern France. Mining and river change are not the result of spontaneous and disorganized intervention, but of well-reasoned policy. The reasons for mining are directly connected to the search for ways to reduce flood risks. In the late 1960s, the administration of the department of the Pyrénées-Atlantiques, under the authority of the prefect, planned the extraction of aggregates in the river bed to benefit the region with the maximum of positive economic effects. The method chosen was to intercept the significant sediment load descending from the Pyrenees and extract much more than the natural contribution so as to lower the river channel by several meters; it would then be controlled by transverse weirs and bank defenses. A management scheme was designed by SOGREAH (Grenoble), an engineering consultant experienced in the subject, with no sensitivity regarding river ecology (which, we must note, was the norm in the early 1970s). The drop in floodwater levels and the narrowing of the braiding river would have positive consequences: reduced flooding would allow for more agricultural land and space meant for thoroughfares, companies and urban and periurban growth of the Pau community. If some economic sectors benefited from these administrative choices, notably public works, the negative effects are no less real. First, no one knows today what the floodplain’s reference flooding conditions are for the hundred-year flood; in the absence of modeling, which has not been decided yet; the 1952 flood has been kept as the reference. In the mid-1980s, the French Ministry of the Environment criticized the ecological impacts of reducing groundwaters, particularly the decline in the floodplain forest, the “saligue” (from the Latin salix, willow), without the bridge and road service of the Pyrénées-Atlantiques bothering to take any

From Hills to the Ocean: Production, Transfer and Trapping

139

other measures. Moreover, the loss of a portion of the groundwater resource accompanying the Gave is a non-negligible economic loss that will one day be made passive in this method of development. The economic interests chosen in the late 1960s still predominate today and the water pits remain the official policy for providing the region with aggregate, despite the vague hope for an integrated policy of the saligue. Many other Pyrenees and Alpine valleys have the same history. 4.3.5.3. The Asian boom Western practices have exploded in the developing world, notably for creating industrial platforms and artificial islands in India and Southeast Asia, like with Singapore. The market value alone of aggregate often justifies the considerable and excessive volumes that have been and are still extracted today. Coastal rivers pay a large contribution, particularly with the modification of the transverse profiles and the rather generalized reduction of flooding on the floodplain. How can we evaluate the weight of mining, largely unknown to this point, through methods allowing cross-referencing? One is to use cement consumption, which is highly correlated with that of sand and gravel, used in mixing concrete, for which 1 t of cement is mixed with 6–7 t of aggregate. Global cement consumption went from 1.4 Gt in 1994 to 3.7 Gt in 2012. In 20 years, China has become the number one consumer of aggregates worldwide, using 58% of the world’s cement (2.15 Gt), far ahead of India with 6.75% [USG 13]. This “proxy”, cement, is quite useful as it demonstrates, if necessary, that the tonnage of extracted aggregates is underestimated, as dredgers’ statements are notoriously false. Next, we turn to the Lower Mekong. On a completely other geographic scale, that of a very large river, the Mekong provides an exemplary case study. Like practically all of the coastal rivers in Asia today, the Lower Mekong has been considered a resource that, if not inexhaustible, is at least exploitable by riverside nations. Sand and, to a lesser extent, gravel are demanded by Singapore, for example, which is expanding its polders, not to mention by Vietnam for construction. Mining has been justified locally by its effect, judged to be positive in improving the conditions for navigation and flooding conditions; in the capital of Cambodia, Phnom Penh, the city is built on the right bank of the Mekong, near the river diffluence that empties and fills Tonlé Sap during each monsoon cycle. A portion of the Mekong’s average annual flow (350 km3) leads to the prospering of an exceptional river-lake ecosystem as well as fishing. The suspended sediment flux, estimated for decades to be around 150–170 Mt/year of silt and clay, seemed to ensure the longevity of mining. The sand flow was ignored, even though it was measured in small quantities.

140

Sedimentary Crisis at the Global Scale 1

Figure 4.7. The extraction of sand (red) and gravel (orange) from the Mekong in 2011–2012; the circles are proportionate to the volume extracted. Thailand and particularly Cambodia and Vietnam are the primary removers (source: [BRA 13a]). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

From Hills to the Ocean: Production, Transfer and Trapping

141

In fact, it is remarkable that no one, to our knowledge, imagined that measuring silt-clay and extracting sand did not have the same significance in terms of the sedimentary budget. Two recent studies performed for WWF-Greater Mekong and IKPM (Information and Knowledge Management Program, Mekong River Commission) led to progress in explaining the sedimentary budget, considering the volume of extraction, until then unknown, and real sediment transport. This budget was most likely affected, starting in 1994, by the construction of storage dams on the Lancang, the Chinese Mekong [BRA 13a, KOE 12]. The real sediment flux would be lower regarding the portion of silt-clay (effect of upstream retention), with an annual flow of 77 Mt/year, but greater for the sandy fraction, the transfer of which had never been measured on the floor or in saltation (1–4 Mt/year). The estimates concerning extraction on the Mekong downstream of China, based on declarations that are most likely less than the real figures, provide a total of 34.5 Mm3/year or 56–57 Mt/year (Figure 4.7). A practical consequence of this first evaluation is as follows: the extraction of sand is much greater than the natural flux; the river is lowering upstream from its delta; and its connection with the Tonlé Sap and the provision of nutrients has been reduced. Moreover, groundwater levels are dropping and the flooding of the immense floodplain made up of the Cambodia alluvial plain and the Mekong delta are less able to maintain the fertility of the agricultural land, the stability of the barrier beach and the fertility of ocean waters. In the Mekong’s lower valley, sand is first and foremost used locally. It has served to feed rampant construction in Ho-Chi-Minh City and in the delta and to backfill many square kilometers of the Mekong’s banks downstream of Phnom Penh, to develop industrial areas or to backfill natural depressions in the alluvial plain likely to be converted into urban areas [PIE 08]. Extraction has seen a recent boom south of Vientiane (the Loatian capital), where Thai enterprises extract sand with suction dredges and excavators to provide for Chinese construction in Vientiane. Women, helped by their children, sort the heaps of coarse gravel extracted from the live channel with excavators to gather white quartz pebbles that are put in bags for a few dollars per day and sold in garden centers11. These extractions run too deep and too close to the edge and so the river’s high banks collapse, bringing buildings down with them (79 km of banks are affected along a 179 km stretch). The construction of concrete banks is, in principle, the chosen protection method, but the work to be performed exceeds the financial means of Laos, except in the city of Vientiane12.

11 AFP, July 27, 2016. 12 Vientiane Times, July 8, 2016.

142

Sedimentary Crisis at the Global Scale 1

As an example of highly contestable usage, the sand of the Mekong has contributed to the construction of polders in the city of Singapore, which now exceeds five million inhabitants and is undergoing rapid growth. Owing to a lack of space on the continent, where it is surrounded by its border with Malaysia, Singapore has chosen to create artificial platforms, increasing its original area (130 km2) by more than 20% in 40 years; and with 515 Mt of sand in 20 years, Singapore was the top importer in the world. One project plans to create an additional 52 km2 on the sea by 2030. Indonesia faced the highest demand for sand for Singapore until 2002, with marine and island sand dredging. Since the Indonesian moratorium, sand has come from neighboring countries, notably Vietnam and Cambodia. These imports are partially illegal as they stem from unauthorized sites and corrupt mafia extractors, however, they elude the tax authorities and guarantee comfortable revenue to certain political figures. Extracting has led to an outcry from NGOs. Sand exports from the channels of the Vietnamese Mekong in 2009 were equal to those from the 10 previous years owing to falsified contracts (fraudulently dated 2008 or earlier). This boom was a direct consequence of the Cambodian moratorium, with Singapore’s importers falling back on lower-quality Vietnamese sand (it being richer in silt in the Mekong estuaries). Barges deliver the sand to ships carrying 10,000 tons [GLO 09]. Moreover, Singapore’s MND (Ministry of National Development) hopes to reduce the nation’s dependence and move towards self-sufficient resource management through recycling, polderization following in the Netherlands’ footsteps (sea dykes and drainage of protected areas to be below sea level) and, lastly, thanks to the construction of waterfronts protected by sea dykes supporting concrete tiles [AUY 17]. A model of the smart city, which is widespread in Asia, Singapore is innovating in transport, information and communications technology, as well as Big Data [PIS 17]. There is no question that Singapore’s brand sells its systemic vision of the city of the future and sustainable development well, but any contestation by citizens of the substrate (physical) on which the city is constructed is forbidden. In short, we are in the presence of an export of know-how based on cutting-edge technology, but which is partly to the detriment of the weaker neighboring countries, which do not correspond to the notion of sustainability, quite the contrary. This predation may permanently deprive these nations of potential economic wealth. Vietnam’s Ministry of Transport began work on a directive in 2015 aiming to reinforce sand control measures and legislation. This process involves the Minister of Construction, Industry and Commerce, as well as the Minister of Natural Resources and the Environment on a large national scale, the goal of which is to orient the country’s popular committees towards mining management plans and careers for the period 2016–2020, as well as to severely limit contraband13. The 13 Viet Nam News, July 4, 2016.

From Hills to the Ocean: Production, Transfer and Trapping

143

reaction of Vietnam’s authorities was quick and firm, but it probably came too late for a river whose dams are blocking supplies. Public opinions are becoming tenser; States are replacing their formerly very lax policies with stricter ones, such that Singapore is seeing its vulnerability. The issue of corruption is on Cambodia’s agenda, where a survey has been started at the request of the NGO Mother Nature and where the Ministry of Mines and Energy faces demands for justification concerning the disappearance of 700 million dollars for the sale of sand to Singapore, with only 5 million dollars having been reported; Cambodian law imposes a 20% tax on the product being sold in favor of the Cambodian State, provided that these sales are officially declared. Moreover, there is no transparency in the issuance of licenses to exploit sand [HAN 16a, HAN 16b]. It is even more surprising that a master plan for extraction was launched in 2013 at the initiative of the Cambodian Prime Minister, who also inaugurated a port in the district of Kien Svay (southeast of Phnom Penh) to facilitate sand extraction and to increase the nation’s revenue. This port was financed by China thanks to a low-interest loan. As was ambiguously stated by Prime Minister Hun Sens, “We must use the river to save the river”14. The stated goal is to homogenize the topography of the river’s channel through dredging, as deeply as possible. The most plausible hypothesis is that the Cambodian authorities are forcing things before an international plan is decided upon.

14 Radio Free Asia, January 22, 2012.

5 The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

The complexity of processes involving sediment entry into fluvial systems, transit, natural trapping, anthropogenic trapping and removal is puzzling a priori. It is not possible to generalize or even group some rivers into truly homogeneous families. This is the reason why we will go through a few case studies that are representative of the diversity of situations, some of which are archetypes in their own right. We will begin this chapter with the Amazon River, the most powerful river in the world, which has been changed little by humankind. This will be followed by a presentation of Southeast Asia’s three rivers, noting that the Huang-He1 has been the world’s primary contributor of sediment to oceans, that the Three Gorges Dam makes the Yangtze a major laboratory for analyzing the impacts of a very large dam and finally that the Mekong is being transformed into an enormous staircase of dams, to the detriment of its delta. We will then look at the example of the Mississippi, a very large river that has long been modified by humankind, before presenting three overexploited rivers: the Nile and the Ebro, Mediterranean tributaries, and the Colorado.

1 The ancient Yellow River is called “Huang He” in pinyin or “Huang-He” in English and French. We no longer speak of the Blue River; it is the “Yangzi Jiang” or “Changjiang” in pinyin (the “long river”), the “Yangzi” in French and the “Yangtze” in English.

Sedimentary Crisis at the Global Scale 1: Large Rivers, from Abundance to Scarcity, First Edition. Jean-Paul Bravard. © ISTE Ltd 2019. Published by ISTE Ltd and John Wiley & Sons, Inc.

146

Sedimentary Crisis at the Global Scale 1

5.1. A river in its natural state, the Amazon 5.1.1. The river in its basin The Amazon, the second longest river in the world after the Nile (6,437 km), holds the world records for the total drainage area in its basin (6,122,000 km2) and its discharge at its outlet into the estuary (209,000 m3/s). The Amazon alone contains 18% of the planet’s freshwater (5 times the output of the Congo and 12 times that of the Mississippi). Three of the Amazon’s large tributaries, one on the right bank, the Madeira, which starts in Bolivia (32,000 m3/s), and two on the left bank, the Río Negro (29,300 m3/s) and the Japurá (18,600 m3/s), are also among the largest rivers in the world2; they are only surpassed by the Congo, the neighboring Orinoco and the Yangtze. Two Andean floods, brought about by the southern rains (December to February) and then by the rains from the northern hemisphere (April to July), spread downstream; their waves combine on the Solimões to produce an enormous general flood reaching its apogee in June, then spreading throughout the year to Óbidos, keeping in mind the delayed tributary contributions; its largest part ends on the floodplain. The Amazon flows to the east for thousands of kilometers. It rises due to tectonic activity as it leaves the Andes; it then flows towards the passive margin of the Atlantic Ocean, which is the final “well” for sediment. The river’s downstream segment is located in a gutter with a metamorphic substrate that separates two large cratons, the Guiana Shield (1,000–1,500 m) and the Brazilian Shield [MEA 07]. We will not expand upon a new hypothesis stating that plate tectonics may be responsible for flow inversion: the uplifting of the Andes during the Miocene may have blocked the original flow originating in the east, creating the current flow, a complete reversal of drainage. Occupying only 12% of the basin’s surface, the Andes provide the river with more than 90% of its sediment, even more if we consider the fact that the sediment removed by the erosion of the river banks also stems from the Andes, but in a secondary position, i.e. deposited in the basins of the floodplains, then freed after several centuries or millennia. The Amazon’s load at its outlet is in the neighborhood of 1.2 Gt/year, rivaling the sediment load of the Ganges-Brahmaputra and the Huang-He before artificial changes. The Madeira alone contributes 0.488 Gt/year from the Andes, while the Río Negro, with a comparable flow, only contributes 0.007 Gt/year [MAR 89].

2 The name “Amazon” applies to the river downstream of the confluence with the Río Negro; upstream and up to the Peruvian border, the river is known as the “Solimões”.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

River

147

Water flow (m3/s)

Sediment load (Gt/year)

Basin area (millions of km2)

Madeira

24,400

0.488

1.42

R. Negro

28,400

0.007

0.69

Japurá

18,620

Amazon

209,000

0.268 1.24

6.112

Table 5.1. Characteristics of the Amazon in Óbidos (800 km from the outlet) and of three of its large tributaries

5.1.2. River function The long periods of high water levels or floods spread water over land along more than 90,000 km2 of the Amazon. The variety of habitats present between the high natural levees (the restingas) and the adjacent basins with varying depths have created exceptional landscape variety and biodiversity, in terms of both flora and fauna. This river and its tributaries are unique in the sense that the river and ecological processes have remained perfectly visible thanks to the scarcity of adjustments, which has the happy consequence of providing lateral freedom to its channels. The floodplains formed by the slow piling up of Andean material are actually fertile thanks to the sediment, which provides turbidity to the rios brancos, compared with the plains bordering the rios negros. The latter drain metamorphic shields that are poor in useful soluble substances; their waters are acidic (pH 3.8–4.8) and rich in dark polymerized organic material (these are the rios da fome, the “rivers of hunger”, which earned their names from the lack of minerals in the locals’ diets). These plains, with their large expanses of lakes bordering the rios negros, are varzeas; they are sprinkled with deltas built by the tributaries and with blind arms, the furos and iguarapés that meander below the forest canopy [JUN 97]. As the plain slowly aggrades due to sedimentation, this dynamic is compensated for by downstream shifting and cutting off of meanders. This ensure that nonconsolidated sandy material will be picked up again and transported towards the ocean. The extremely high levels of biodiversity in the Amazon could stem from the coexistence of “primary” riparian successions (created by the active river dynamics in basins as a form of mobilization and a regeneration factor) and mosaics of firm land belonging to old floodplains raised by tectonics (raising of tectonic arches and basin subsidence) [SAL 86]. Moreover, comparisons between river segments have revealed that the speed of migration, convex bank construction and cutting off of meander could be conditioned by the volume of the sediment load imposed upon them; it is not surprising that heavily loaded rivers flowing from the Andes have higher migration rates. These high rates create vast low-lying spaces at their

148

Sedimentary Crisis at the Global Scale 1

margins, which favor the storage of suspended material and thus maintain the sediment equilibrium. Inversely, storage dams, by greatly reducing transfer, lead to risks for the dynamics of river milieus and their complexity. Season floodwaters invade the floodplain and its depressions, meeting again at the outlet, whereas, as we have just seen, a significant fraction of the sediment from the Andes is deposited along the way. It is estimated that the Andes provide 2.3–3.1 Gt/year to the Amazon basin, but only 1.4 Gt/year can reach the foreland and 1.24 Gt/year can reach Óbidos. These deposition/erosion compensation processes are so intense that the volumes of sediment exchanged between the alluvial plain and the Amazon may be greater than three times the volumes exported to the river outlet (Figure 5.1). The average residence (or recycling) time upstream of the confluence with the Madeira may be between 2,000 and 3,000 years, a long enough time to dissolve clay particles and reinforce the sandy fraction [MER 96].

Figure 5.1. Satellite view of Iquitos, Peru, a city built on the left bank of the Marañón River (downstream from the confluence with the Ucayali River, the two rivers take on the name Amazon). Iquitos, once built on the waterfront, now finds itself distanced from the Marañón River after the meander was cut off. Image taken during flooding, June 2, 2002. The great amount of suspended load from the Andes backfills the channels near the primary channel (source: NASA, ID-ISS004-E-12717). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

149

The last measurement station on the river is located in Óbidos, 600 km from the river mouth at the upstream end of the estuary. The sediment load is therefore 1.24 ± 0.13 Gt/year [MER 07]. The estuary, 200 km wide, surrounds the vast island of Marajó and the estuary of the Pará River, which communicates with the Amazon via furos, lagoons (bahias) and marshes [DEN 27]. The estuary’s morphology is under the control of eustatism* and regressive erosion probably went back to the Solimões, upstream of the confluence with the Japurá. The retreat of waters in the Grimaldian during the last glaciation period led to the broad, deep incision of the estuary zone, which was invaded by waters during the Flandrian transgression for about 6,000 years; the minor fluctuations of the ocean level, such as the maximum during the Dunkerkian about 3,000 years BP, could explain the presence of semi-submerged or semi-perched river-delta landforms, but the Amazon River system’s load was not great enough to compensate for the effects of the rising base level, and subsidence intensified this phenomenon [TRI 77]. Geographer Pierre Gourou coined the term ria fluviale* to classify this type of mouth. The sediment flux descends during low tide, as witnessed at the oceanfront; it is stopped and reverses when the tide rises, with a tidal range of 3 m. The decantation of gray mud, tijuco, takes place along half of the estuary, particularly the right bank, while the tidal bore* erodes the left bank, the rest being deposited on the continental platform as a 1,200 km-wide underwater delta dropping down 5,000 m. Continental drift remobilizes a portion of this sediment (0.20–0.25 Gt/year) and transports it to the northwest, to the Orinoco’s delta. Along the way, it forms vast mud flats, for instance in French Guiana, which are colonized by mangroves. 5.1.3. The threat of dams Dams that have already been built and those planned threaten the functioning of the Amazon River system. The large Tucuruí dam, one of the 40 works planned on the Tocantins River, was built between 1976 and 1984; it has an installed power of 8,370 MW. Specific problems were discovered in the reservoir (decomposition of organic matter, risk of silting connected to the agricultural colonization of the basin, hygiene problems along the banks) and downstream from the reservoir (flood reduction, obstacles to fish migration, reduced natural fertilization due to a drop in nutrient contributions) [BAR 87]. The impacts of 40 dams planned in the Tapajós basin, which converges downstream of Óbidos (total power of 25,000 MW) could be even more significant.

150

Sedimentary Crisis at the Global Scale 1

5.2. Adjusted rivers in China and Southeast Asia The great investment efforts made in constructing reservoirs on Chinese rivers are revealed through its efforts concerning sediment dynamics. Measurements have been taken at the mouth of rivers flowing into oceans or regional “seas”, from the Heilong (Amur) in the north to the Mekong in the south. On a hydrological level, the output of the 10 largest rivers in China has varied rather little since the 1950s. The interannual fluctuations are marked and do not reveal any tendency. The only undisputed aspect is that the rivers north of the Huang-He have less water. However, the results are clear concerning the effects of sediment transport. The total load of Chinese rivers has dropped tendentiously from 2.09 Gt/year in the period 1955–1968 to 0.575 Gt/year in the period 1997–2007. The country’s two largest rivers, the Huang-He and the Yangtze, make up for 83% of this amount, regardless of the cause, whether reversible (climatic) or not (of human origin) [LIU 09]. Next, we will deal with the issue of the Mekong, which has been clearly affected, even though hydroelectric adjustment is recent.

5.2.1. The Huang-He downstream of the Loess Plateau: contemporary generalities The 5,464 km-long Huang-He drains a basin covering 752,000 km2. This basin receives precipitation levels reaching approximately 480 mm/year on average. The water flow of the Huang-He averages 46.4 km3 at Sanmenxia, at the entrance to the floodplain. The concentration of suspended sediment here is one of a kind around the world, equal to 10–20 times that of other large rivers. The average annual load is 1.6 Gt/year in Sanmenxia, i.e. before reaching the massive deposition that starts well upstream of the delta. The hydrological rhythm follows a seasonal pattern; it is characterized by a flow brought about by monsoon floods between July and October (60% of the annual total), which ensures the transport of 85% of the load. The concentration thereof can increase as high as 1,000–1,500 kg/m3 or even more over the median course and more than 900 kg/m3 in Longmen and Sanmenxia [ZHE 89]. The Huang-He alone contributes 6% of the globe’s sediment transport (Table 5.2)3.

3 For more details on the characteristics of the basin and the river across the Loess Plateau, see Chapter 3.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

151

River

Water flow (km3)

Average concentration (g/m3)

Load (Gt/year)

Huang-He

40–43 (delta)

25–37

1,6 in Sanmenxia

Yangtze

910 (Datong)

4.7

0.5 (1953–1976)

Table 5.2. Comparing some characteristics of the Huang-He and the Yangtze (source: [REN 98])

5.2.1.1. The influence of agriculture on the Loess Plateau after 1950 Sediment transport influenced by agriculture (excluding reservoirs) is naturally more important than under “geological” conditions. In the period 1950–1977, prior to the effect of reservoirs, the fluxes from the Loess Plateau to Sanmenxia were on the scale of 1.6 Gt/year; the peak in sediment transport dates back to the 1980s. Above, we saw the effects of soil protection policies (with limited effects) and of sediment trapping policies on the Loess Plateau, at least in the upper ravines. The period prior to 1980 also experienced storage dam construction, which involved significant filling. However, it is important to distinguish between dams built along the tributaries of the river draining the Plateau on the one hand and those built on the river itself on the other. In the former case, that of the Xinqiao and Jiucheng reservoirs (on the Wuding River and its tributary, the Luhe River), were quickly backfilled and the river’s new longitudinal profile returned to its original form, but 40 m higher on a 40 km reach. This resulted in a useful reduction in mass movement and sediment transport, the rise in alluvia having reduced destabilization of the mountain sides [GON 87]. In total, the sediment transport from the Loess Plateau may have dropped by nearly 20% thanks to soil protection measures and reservoir construction. 5.2.1.2. The factors affecting the recent decline in the sediment load of the mid-valley The decline in the solid load recorded on the Huang-He can be attributed to two primary factors: – the massive derivation of winter irrigation water with little sediment load has significant effects. Agricultural water consumption has increased to 30 km3/year, yet the sediment load remains relatively limited; – in contrast to the preceding result, climatic variations and the moderate decline in precipitation levels are accompanied by a more significant reduction in sediment

152

Sedimentary Crisis at the Global Scale 1

transport [LU 13]. One study of the available data, performed in the long term, reveals that the sediment load may have dropped by 69% between the predevelopment period (1919–1978) and the post-development period (1979–2008); 81% of the reduction may be due to human activities and 19% due to climate change [MU 12]. In total, the positive effect of soil restoration is decreasingly established as we learn to better estimate the weight of other factors. While the 40% reduction due to Loess land control measures were previously advanced, the effect of derivation and water consumption, as well as the effect of reservoirs, have been reevaluated at a lower level, while the climatic factor has recently assumed greater importance. Research efforts have been successfully encouraged in the Loess Plateau region; other factors supplement these today, and there is no lack of importance in the effects of excessive water derivations that deviate part of sediment load and particularly impose a serious handicap to the sediment transport capacities downstream. 5.2.1.3. The role of large storage dams in sediment management on the river Large dams were designed by the Yellow River Conservancy Commission in the early 1950s and built primarily to store part of the Huang-He’s flood discharges. The 103 m high hydroelectric reservoir Sanmenxia, completed in 1960 at the gorges outlet, quickly revealed its first weakness: the absence of bottom flushing gates caused nearly all sediments to be trapped. The dam had been designed with no real study of sediment transfer. Sedimentation upstream from the Sanmenxia Dam also affected the tributary on its right bank, the Wei Ho, which flows into the Huang-He at the upstream end of the Sanmenxia reservoir, near the city of Tongguan [WU 07]. In the first 2 years after it was put into service, deposition in the Sanmenxia reservoir was 1.7 Gt, taking up 43% of the reservoir’s capacities (Figure 5.2). The length of the area of the river reach affected by deposition and rising water level along the Wei Ho was estimated at about 135 km, with more than 4 m of aggradation. The opening of bottom gates through the concrete-made Sanmenxia Dam allowed flushing that partially regulated the problem, but massive dredging of the Wei Ho’s aggraded bed was also necessary, as was the erection of enormous dykes along this tributary. Recent modeling shows that equilibrium has not been achieved, because sediment storage and release in the reservoir depend on unmastered external factors, notably climate, water and sediment input (accumulation increases when flow is reduced), trapping by upstream storage dams, and the functioning of bypasses used for irrigation [ZHE 89].

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

153

Figure 5.2. Image of flushing operations at the Sanmenxia Dam on July 26, 2013 during the monsoon flood (source: Wikimedia Commons, Rolfmueller (CC BY-SA 3.0)). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

Several large dams built on the large rivers in the Loess Plateau region to retain floodwaters pose similar sediment-related problems (Liujiaxia Dam, 1968; Longyangxia Dam, 1985); the last dam built was that of Xilaolangdi (1997), downstream of Sanmenxia. Sediment filling of the reservoirs created by hydroelectric dams thus poses a serious problem in China (Figure 5.3). It only took a few years for the sediment storage capacity of the river’s reservoirs to decrease from 50% to nearly 90%, and many current reservoirs located on tributaries have been abandoned. In the Shaanxi province, the lost capacity is estimated at 80 hm3/year, but this was compensated for through the construction of new reservoirs with a capacity of 260 hm3 [ZHE 89]. The effect of dams has proven to be very complex, as it combines: 1) the impact of sediment flushing; performed for strong flows. They are heavily loaded, cause extensive deposition, modify the channel and reduce the section downstream from Sanmenxia; 2) weak, but strongly concentrated flows [MA 12].

154

Sedimentary Crisis at the Global Scale 1

Figure 5.3. Impact of the start of operations of the Sanmenxia Dam on the sediment load of the Huang-He in Lijnin in the period 1951–2007. The black curve is the mobile average of the output (annual values in blue); the orange curve is the mobile average of the sediment (average values in green) (source: [LU 13]). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

5.2.1.4. Fluctuations in the ENSO and solid transport It has been observed that the years with abundance of flow on the Huang-He and the years of limited discharge alternate on a cycle of approximately 8–10 years, which is the same as that of the ENSO (El Niño Southern Oscillation, which affects the atmospheric pressure in the Northern Hemisphere). One significant correlation has been proven between the indicators of the ENSO and the flow conditions in the fall (end of the monsoons) and during certain winter months. The rain follows at a gap that may be several months, but this observation allows the intensity of the rainfall to be anticipated, which is important for predicting floods, sediment transport and water supplies during the dry winter months [LU 10]. The discharge ratio between these extreme years is high, about 4:1, but the ratio regarding the interannual variability of the suspended load is 8:1. Between 1951 and 2007, the sediment flux produced by eight large rivers in China dropped from 1.5 to 0.6 Gt/year. The analyses of precipitation decrease, temperature increase and water samples from basins have shown very interesting relations with sediment load. The measurements show that the reduced flow is always less extreme than the reduction of the sediment load. This tendency is certainly less significant than the effects of reservoirs and extraction, but it is nevertheless very significant for planning in the decades to come, which will, in principle, see an intense climate change [LU 13].

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

155

5.2.1.5. Flood control reservoirs Another flood control technique was to create vast storage basins on the floodplain, behind and below the river levees. That of Dongping Lake (418 km2), itself surrounded by dykes, farmed and inhabited, can contain 3 km3 of water and receive the Huang-He’s discharge up to a limit of 10,000 m3/s. Frequently used, Dongping Lake underwent heavy aggradation through sediment deposition that progressively reduced its efficiency. Another basin, the North Basin of the Gold Levee (2,316 km2), protecting more than 1.5 million inhabitants, was designed to serve the northern reach of the river if needed, but large tributaries brought other threats. 5.2.1.6. Seasonal drying-up of the Huang-He and related issues One remarkable situation, first observed in 1972, is the seasonal drying-up of the river; from a permanent river, the Huang-He has become an intermittent one, while the annual water flow dropped as well. In 1997, the dry season lasted 226 days and the year’s discharge dropped to 1.7 km3 in Lijin, 4% of the average discharge measured in the long term. Because this drying-up went back to Kaifeng, 700 km from the China Sea, it would not be wrong to say that 90% of the Lower Huang-He is periodically dry [REN 98]. Because the water discharge is insufficient to transport the sand load to the outlet, sand aggrades the lower river’s channel at a rate of 1 cm/year, which explains why the channel has a relative altitude of 5–10 m compared to the alluvial plain. The monsoon floods rise higher and lap against the dykes, which increases the risk of breaching and flooding. Among the undesirable consequences is also an increase in water pollution due to lack of dilution (and thus threats to human health), as well as the salinization and alkalinization of the estuary branch entering the delta. One of the causes of this drying-up is the 5–10% drop in precipitation since 1950; other factors include various forms of water consumption that have increased along the Huang-He River due to agriculture, industry and settlement. The Huang-He’s load also suffers from water derivation, particularly to the Hebei and the Shandong Peninsula. 5.2.2. The Yangtze and the Three Gorges Dam 5.2.2.1. Generalities Starting in the Qinghai-Tibet Plateau, where it has its source 6,600 m above sea level, at the China Sea, the Yangtze has highly contrasted reaches. The upper river has been incised into the gorges for 1.8 million years. Downstream of Yichang, Jingjiang is the name given to the middle reach of the plain, which stretches more than 350 km. It is characterized by meanders that were extremely mobile throughout the river’s history. In the section downstream from Jingjiang, the Yangtze braids with stable islands and channels. The low degree of hydrological variation works in favor of island stability. This is a subsidence area connected to neotectonics, where

156

Sedimentary Crisis at the Global Scale 1

80% of the river flow converges. The Jingjiang more than compensates for subsidence by flowing over an accumulation of very young sediment. This suspended river bed, before the Three Gorges Dam became operational, increasingly quickly aggraded between its dykes (the floor was 2 or 3 m above the floodplain in 2000). The Jingjiang is subject to disastrous floods connected to an insufficient evacuation capacity and the conjunction of peak floods from the river and its tributaries. The water level between the dykes can rise from 6 to 17 m above the floodplain. As for the Yangtze delta’s dynamics, they are controlled by the tide, which has a magnitude of 2–4 m, although the waves are quite weak (1 m tall). The tides can be felt up to 600–650 km inland (just downstream of Datong, the farthest hydrometric station downstream where fluxes are measured). 5.2.2.2. Fluxes With an average discharge of 14,300 m3/s in Yichang (at the foot of the mountain and downstream of the Great Dam) and 28,300 m3/s in Datong (upstream of the mouth of the Yangtze, but beyond the tide’s influence), a dry season flow of 2,900 m3/s in Wuhan and floods exceeding 70,000 m3/s, the Yangtze is the third largest river in the world. The water and sediment flux regime is based on that of the summer monsoon, with nearly 90% of discharge from May to October. Draining a basin of 1.8 million km2, the Yangtze transported nearly 0.5 Gt/year of sediment to the sea before the Three Gorges Reservoir was made operational; this was little in relation to its neighbor in Northern China (Table 5.2). At China’s scale, the Yangtze provides a remarkable example of impacts, those recorded downstream of the Three Gorges Dam, in the reach called the “Jingjiang”. The first dam built on the Yangtze was the Gezhouba, which began operations in 1988 (47 m tall with a reservoir volume of 1.66 km3). It blocked the pebbly bed load and reduced the size of suspended particles, because the coarsest ones were trapped in the reservoir; it thus temporarily contributed to a slight reduction in the load in Datong. The Three Gorges Dam was made partially operational in 2003 before complete filling in 2009, and its first impacts were analyzed with no ambiguities. It had modified both the hydrological regime and the sediment regime at the scale of the downstream Yangtze, particularly in the Jingjiang reach. The measurements taken at the two stations, Yichang and Datong, demonstrate a highly significant load reduction. What could be the causes of this reduction? Two reaches impose themselves as analytical supports, the Upper Yangtze, between the Three Gorges Dam and Yichang, and the Lower Yangtze, between Yichang and Datong (Table 5.3) [LIU 07, XIQ 05, YAN 15].

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

Years

Sediment load (MES) (Gt/year) in Datong and to the ocean

1950–1968

0.507

1953–1976

0.482

1977–2000

0.389

Post Three Gorges Dam

0.145

157

Table 5.3. Evolution of the sediment flux of the Yangtze in Datong (source: [LIU 07, XIQ 05, YAN 15])

A recent evaluation showed that the Yangtze’s average annual flow has decreased by 7% in relation to the period 1950–2002 (−67 km3/year). The largest part of this decrease is likely due to the reduction in precipitation. As for the sediment flow, it decreased by 70% between 1960 and the years after 2003. The Three Gorges Reservoir is responsible for 65% of this reduction, the rest being attributed to the drop in precipitation, the effect of other reservoirs and soil protection. The changes at work in the Yangtze’s upper basin are as follows: contrary to common opinion, the efficiency of soil protection measures is limited, because these efforts have affected few sectors in the Jialing basin; the tendency was rather towards a clearing and exacerbation of erosion, at least until 2000. Without the negative effects of reservoirs and extraction, the Yangtze’s budget would have increased in suspended load. The large reservoirs built on the Jinsha (a river that flows into the Yangtze in Chongqing) are major traps. The impact studies prior to the construction of the Three Gorges Dam foresaw a total sedimentation in the reservoir of 8.57 Gt over 30 years and an equilibrium that would be reached within 80 years. However, dams built on the upper tributaries, partially to achieve this goal, relieved the sedimentation in the Three Gorges Reservoir. In 2000, the Yangtze’s upper basin had 12,700 reservoirs with a cumulative volume of 23.4 km3. The 29,600 reservoirs on the Yangtze’s mid-basin have a total volume of 116 km3, but they are in less of a position to intercept the products of erosion than those on the upper basin. The changes in the Yangtze’s mid-basin (downstream from Yichang) are more complex [LI 07, LI 16, WAN 13b]. First, the reduced sediment input into the Jingjiang, due to the effect of interception ensured by the Three Gorges Reservoir (interception ranges from 65–85% of the solid load, depending on the year), has

158

Sedimentary Crisis at the Global Scale 1

translated into the immediate incision of the Yangtze’s sandy-bottomed channel, which lost a volume of 1 km3 in the period 2002–2010. The intensity of the incision decreases downstream due to the sediment input caused by bank erosion and tributary contributions. The latter is however limited because artificial reservoirs, like the Danjiangkou on the Han, also trap sediment. There is a tight connection between the hydrological regime influenced by the Three Gorges Dam, the behavior of the Yangtze channel and the fate of sediment. On the one hand, the flow impacted by the Three Gorges Dam shows a certain regularization of the seasonal regime, with a rise, in the winter, of the low water’s discharge and levels (+0.20–2.10 m); there is also a reduction in the discharge of high monsoon waters and autumnal levels (+0.70–3.35 m). From 2009–2012, the incision of the channel reduced the winter and fall rise in the water levels by half. On a local scale, the impacts appeared quite quickly. Since the Three Gorges Dam became operational, in a few years, the meandering reach of Jianli, a segment of the Jingjiang about 100 km long, reacted strongly: accentuation of meandering, incision of deep pools and enlargement of the channel to the detriment of the banks, whose sand banks collapse spectacularly. This is the image of great instability. The relative increase in the floodwaters and that of the water heights in the Yangtze (with the same discharge) increase pressure downstream. It can be expected that the disturbances recorded by the Yangtze’s channel will continue in the future and the attenuating and delaying effect on the floods will be even further reduced. During flooding, the river is actually connected to natural lakes at its southern margin created through subsidence (Dongting Lake, Poyang Lake); they contribute to the protection of 10 million inhabitants. Mastering the Jingjiang’s flooding strongly depended on filling Dongting Lake with river floodwaters; this is achieved through a vast complex of river channels and lakes serving as the natural expansion field for floods, located on the right bank of the Yangtze between Ychang and Wuhan. The lakes were fed naturally, controlled by three spillways built in 1958 in the Jianli reach (southern point of the Yangtze’s bend between Yichang and Wuhan). In just 7 years (2008–2014), with the same discharge, the drop in the flood level reduced the discharge rerouted towards the lake in the Upper Jingjiang. Moreover, a sediment deposit affected the spillway sectors and contributed to reducing the rerouted discharge. Also, in the spring, the incision of the Yangtze reduced the blocking effect the river had on the lateral lakes (Dongting Lake or Dongting-Hu), as well as the potential benefit of the water level rise in the summer. Dongting Lake alone stored 4.93 Gt of sediment between 1956–1995, but the Yangtze’s contributions decreased, as did their contribution to Dongting Lake as a result. Moreover, each lake has an area that is reducing because of the river

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

159

contributions and the contributions of tributaries in their basin, in spite of the outlets to the larger river.

Figure 5.4. Impact of the start of operations of the Three Gorges Dam and basin management on the discharge and sediment load of the Yangtze in Datong in the period 1951–2007. The black curve is the mobile average of the discharge (annual values in blue); the orange curve is the mobile average of the sediment load (average values in green) (source: [LU 13], modified). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

The Yangtze’s channel has been the site of sand extraction since the 1950s, and this removal, partially legal and in all cases excessive, has skyrocketed since the 1980s, at the risk of catastrophic bank collapses. This extraction has risen to 40–80 Mt/year since the late 1980s. At the same time, the mid-Yangtze was trapping sediment. The balance of these contradictory tendencies can be found in the fact that the Yangtze’s contribution to the sea greatly decreased between 1953–1976 and 1977–2000 (Figure 5.4). Moreover, since regulations have protected the Yangtze channel, extraction has migrated to Poyang Lake (since 2001). In 2005 and 2006, 236 Mm3 was extracted there, supplying 9% of China’s overall need. This is the first production site in the world, but the loss of biodiversity due to the high turbidity led to the end of this extraction in 2008 [LEE 10].

160

Sedimentary Crisis at the Global Scale 1

5.2.3. The Mekong 5.2.3.1. Generalities The Mekong is the second largest basin in the world after the Amazon in terms of fish diversity, and it is the site of the most productive fisheries on the planet. In the mid-2000s, the Mekong’s hydrologists believed that this river was one of the most natural rivers on the globe, particularly due to the low population of its mountain basin (10 inhabitants/km2) and its low development level, even in regard to navigation [WAL 09b]. It is true that traffic has always been hindered by the rocky outcrops that create a rough bed, as well as by the low water levels during the dry season. The primary resource was abundant and highly diverse fishing. The Mekong begins in Tibet and flows more than 4,800 km to the estuary branches of its delta. Called “Lancang” in its Chinese reach (the “turbulent river”), the Mekong flows through its gorges between folded chains with no significant tributaries, which provides it with a greatly reduced watershed and a steep slope. Its whole course being located between mountain chains that only provide water from short tributaries, the Lancang’s elongated basin only covers 795,000 km2. Downstream of its Chinese reach, the Mekong maintains its mountainous character until Vientiane; upstream of Luang Prabang, it receives the Nam Ou, and its basin is locally broadened. The Mekong has a relatively young course as it travels through mountains that are approximately 6–8 million years old upstream. Downstream of Vientiane, the river flows through low metamorphic and volcanic plateaus deformed by tectonic activity. Its course is extremely recent from a geological standpoint, as it flows in volcanic plateaus dated from the Quaternary Period (maybe Holocene) since the upper basin’s waters were rerouted onto today’s course. Until the Mekong enters Cambodia, the channel and its narrow, discontinuous floodplains flow between reliefs whose history are not well-known to us. The channel offers few possibilities for sand storage except at the foot of rocky slopes, but clear quartz is highly present in the dark landscape at the water’s edge, marking the highest levels reached by the annual flood that reworks the low banks. Flooding, so important for the fertilization of the low plains, only takes place in the Cambodian and Vietnamese reaches. In Cambodia, the tectonic basin of the great lake Tonlé Sap communicates with the river, which fills it when it floods and dries up once the flooding subsides [GUP 07]. It is only starting in the Cambodian reach that the Mekong can develop a narrow alluvial plain before building its delta in Vietnam. Hydrology Relatively sheltered from the effects of the summer monsoon, the upper basin only receives 1,000 mm/year on average, whereas the lower Mekong, fully open to water fluxes (June to October), receives an average of 1,670 mm/year, irregularly

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

161

spaced out. The volume discharged from the river only amounts to 65 km3 when leaving Chinese territory, but 350 km3 upon entering Cambodia. The downstream Mekong has an average discharge of 14,500 m3/s. After the first rise in the water levels due to the melting of the snow in the upper basin, the Mekong enjoys abundant monsoon rains from June to October; the high waters are rather regular from one year to another and spread downstream following the same rhythm, notably with a peak occurring in August in Laos, September in Cambodia and October in the delta (Vietnam). While low in April, during the dry season, the water is only 2,100 m3/s, and the monsoon floods, highly regular from one year to the next, are an immense hydrological pulse occurring from May to September, with annual increases to an average 40,000 m3/s; peaks caused by heavy cyclonic rain add to the regular rise in the waters. The floodwaters are transferred to the delta unchanged, because the overflow spaces are highly restricted until the Cambodian reach. This is the flood pulse spoken of by ecologists, and it has very important ecological effects. In its natural state, the flood covered the delta plain starting at the distributary channels, the largest two being the Mekong itself and the Bassac; these two branches divide further downstream. Sediment load The Mekong’s load is heavy due to the contributions from the mountainous basin; the granite and sandstone outcrops provide sand, an extremely important resource for the channel and the maintenance of the delta’s dunes and beaches. Based on the measurements of the suspended load taken starting in 1962 but interrupted multiple times by wars, for decades, the sediment load was estimated at 70 Mt/year in Chiang Saen, at the entry of the lower Lancang (drainage of 20% of the basin’s area) and 150–170 Mt/year, depending on the year, in Paksé (upstream from Cambodia, although the measurement is valid for downstream). The publications admitted – without validation until quite recently – that in Kratie (upstream of the Cambodian border), there are about 160 Mt of suspended silt and clay transported and the floods reduce this progressively on the plains of the downstream course. The load is partially stony upstream of Vientiane, but this has not been measured; the sandy load that, according to the stations, is transited in suspension or on the channel bottom was also considered negligible due to a lack of measurement. This figure, 160 Mt/year, is particularly low in comparison with other regional values, and, without the mountain, it would be even lower. Essentially, silt and reddish clay are rarely seen on the banks of the Mekong, which can only be explained by complete transport in the upstream segments, by deposition on the downstream floodplains behind the levees and finally by transport to the ocean

162

Sedimentary Crisis at the Global Scale 1

[CAR 09]. The visible sand and gravel are generally considered part of the bed load, due to a lack of measurements in suspension, but observations made in the late 2000s suggested that sand is easily transported in suspension. Moreover, the longitudinal variations in the MES fluxes measured remained mysterious, particularly the decreases at certain downstream stations; they brought to mind contributions from unknown tributaries or losses of transport capacity, but without it being possible to identify deposition areas, or even sediment “wells”, the location of which was still unknown. A paradox of the Mekong is thus the existence of a significant volume of sand, visible in the channel and on the banks, but whose flux, and thus also its regeneration capacity, are unknown, while this is the object of intense extraction (see Chapter 4) and even though the reservoirs built and being built threaten its transfer towards the delta, where it plays a central role. A recent study looked more in depth at the specialized, but important, point of the method of sand transfer, because this conditions our understanding of the river’s functioning during floods as well as the measurements revealing the reality of transport. Systematic analyses were performed to characterize the sand of the banks all along the river. These helped us understand that the sand is transited largely in suspension, along with silt; sometimes it is coarse sand, sometimes fine sand, depending on the turbulence of the flow, which itself depends on the river’s energy, depending on the channel’s geometry, notably its slope. Why were the measurements in the water mass unable to detect sand? Probably because it is transported particularly during flood peaks and because the measurements were not taken in these difficult conditions, and possibly also for reasons related to the measurement techniques [BRA 13b]. Parallel to this research, the question of measuring sediment loads has been fully revisited. The following are the primary contributions: – contributions to the sea were corrected to 72 Mt/year based on new measurement techniques, knowing that the largest part of the reduction recorded is due to the beginning of operations of dams on the Lancang (Figure 5.5); – it has been confirmed that sand is indeed present in the suspended load during high discharge periods, but that it migrates as bed load during lower flood discharges (14–15 Mt/year); moreover, the upstream–downstream variations in the load are explained by the mechanisms specific to each river reach, which joins the contribution of geomorphological analysis; – sand extraction exceeds the upstream contributions by at least 25 Mt/year [KOE 14].

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

163

Figure 5.5. Comparison of the annual sediment load in Chiang Saen, downstream from the Lancang (Yunnan, China) and in Pakse (Laos), the last station before the Mekong delta. The load dropped downstream of the Lancang dams, more so than downstream from the basin, which has been less modified to date. In blue, the historic data (Mekong River Commission (MRC)) in Walling [WAL 09b]; in orange, the results of recent measurements (2009–2013) taken by Lois Koehnken (source: [KOE 14]). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

These results will make it possible to stall the reality of the impacts much better than it was possible through modeling based on contestable figures. In all, the two studies that were just briefly presented emphasize that the breakthroughs that could come about as a result of storage dams in terms of the longitudinal division of energy will have heavy consequences on sediment transit, particularly sand. 5.2.3.2. Cooperation and economic development projects concerning water The integrated adjustment of the Mekong is an old idea dating back to the period after World War II. In 1949, a UN commission was created to promote cooperation on the large rivers; the Mekong was selected in Southeast Asia for the struggle against flooding and the development of hydraulic resources. In 1957, a development plan was presented to the Mekong Committee responsible for promoting it in countries where one of the political objectives of Western nations was to oppose the progression of Marxism under the aegis of the Economic and Social Commission for Asia and the Pacific, created by the UN (1957–1977). Relatively in-depth studies foresee a chain of

164

Sedimentary Crisis at the Global Scale 1

dams on the river and its tributaries – part of this has been built in Thailand – as well as projects pertaining to irrigated agriculture, fishing, navigation and flood control; the Asian Development Bank and the World Bank have provided funding for this. Due to the American withdrawal following the Vietnam War and the political changes on the peninsula – three of the four countries along the downstream segment of the Mekong (Cambodia, Vietnam and Laos) adopt Marxism – the Mekong Committee entered a period of instability; it sought to relaunch the development initiative, but on new political bases, while maintaining close ties, notably financial ones, with the UNDP4. Technical measures were limited to the hydrometereological monitoring, to studies intending to adjust the river for navigation and for the construction of dams on the tributaries. Then, the positions vis-à-vis the very nature of development were disrupted, with the breakthrough of the concept of sustainable development (1987). The question of sediment and that of the environment appear in 1992 in expert reports ordered by the Interim Committee, which led to the Global Environment Program, launched in 1991 and executed following a phased approach. In 1993, France participated financially alongside the UNDP in a study proposing the revision of development plans for the series of dams on the river, although without modifying a general schema strongly inspired by the USA (the US Bureau of Reclamation has long influenced general designs by promoting the Tennessee Valley Authority’s model). This study, finished in 1994, promotes the notion of derivation dams, to the detriment of large dam-reservoirs, with the goal of achieving better environmental insertion; it is important here to note the direct influence of the Compagnie Nationale du Rhône (CNR). The era of the Mekong Commission began in 1995, responsible for promoting cooperation between States in a liberal economic framework, taking into account the political specificities of the different countries on the peninsula and the direct competition between Thailand and Vietnam in allocating water. China chose not to participate in the Mekong Commission, which helps it avoid taking sides that could risk placing it in environmentalist territory, thus allowing it to launch its first dam without excluding a long-term economic policy that would be offensive in the Lower Mekong region. The 1995 Mekong Agreement makes all nations bordering the Lower Mekong subject to the unanimous agreement of their partners5. When the ASEAN (Association of Southeast Asian Nations) is globally favorable to the nations of the Lower Mekong, China is upstream and thus holds a dominant position, awaiting its time. The boom in dams on the Mekong and in its basin The first dam made operational on the Lancang was the Manwan (1994); after 20 years, a chain of eight dams was completed in China with the Xiaowan Dam (2010, 4 UNDP: United Nations Development Programme, created in 1965. 5 The highly documented work by L. Lacroze [LAC 98] belongs to a highly pro-development mentality, but it manifests a certain degree of impatience and pessimism concerning the possibility of finishing the projects initiated in 1949. The acceleration in construction under the influence of the Chinese economic boom was not anticipated by the author.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

165

height: 292 m) and the Nuozhadu dam (2012, height: 260 m). Across the basin, the 2000s saw an explosion of projects and construction in the tributaries and on the river. In 2015, 31 dams were under construction and 136 were planned for the next 40 years [MRC 11]. These construction projects are highly alarming, as the reservoirs for these dams are meant to control a water volume of 97 km3 in the long term, i.e. 25% of the Mekong’s annual discharge. China has thus oriented itself towards integrated hydroelectric development of the Mekong on its territory. This choice ensures control of the discharge, and this series of dams will control the entire sediment flux (the efficiency of trapping would be 60% for the Manwan reservoir alone). For a number of scientists, the impacts would stem less from the seasonal alteration of the hydrological regime than from the total loss of longitudinal connection for living species and sediment. The administrators only know how to measure water discharge well; they have a poor perception of the negative consequences of a chain of dams, and hence they advocate integral development. The concerns regarding the sediment flux dynamics did not manifest themselves in the first years, because the Chinese data were difficult to access and because the monitoring of the MRC was uncertain. Two elements were even considered favorable by the MRC or as likely to compensate for the effects of the dam: on the one hand, the erosion of sand deposited along the Lower Mekong could be a source of sediment downstream; on the other hand, tributaries could contribute even more to the sand and pebble flux, e.g. the Nam Ou, which is upstream of the confluence with Luang Prabang and partially regenerates the load. A major breakthrough came in 2012 when Laos and Thailand decided to secretly build the Xayaburi Dam, freeing themselves of the principles of the Mekong Agreement, which imposes shared information among the countries of the Lower Mekong and their prior agreement before any modifications of such a scale are made. Thailand wanted energy and Laos wanted to benefit from the revenue expected from selling electricity to its neighbor. The opposition of the downstream nations, Cambodia and Vietnam, was not enough to block the project, whose construction started downstream of Luang Prabang in 2012, to be put in service in 20196. Moreover, the implication of a Western company, the CNR (ENGIE group), in managing sediment through flushing during floods, has become an argument in favor of the credibility of this so-called “green” energy; future management is not meant to pose an obstacle to sediment transit (dams should be “transparent” vis-à-vis the solid flux, as stated by engineers)7. Some elements of the file, however, raise doubt concerning the feasibility of the sediment flushing for both technical and institutional 6 The creation of a series of dams on the Nam Ou and the Xayaburi Dam will somewhat serve as a shield against the impacts brought about by the series of dams on the Lancang. 7 Scientific works financed by USAID work towards the same goal.

166

Sedimentary Crisis at the Global Scale 1

reasons; the methods of managing flushing would depend on decisions that are not guaranteed by any apparent procedure. Thus, the Mekong Commission, which was a powerful regional cooperation tool, lost part of its credibility and its strength, even though it is still financed by donors. The door has since opened wide for integral adjustments of the Lower Mekong. Moreover, recent years have seen an explosion of dam projects on the river and its tributaries (the Nam Ou, the Se San, the Srepok and the Sekong, which drain the Annamite Range in Southeast Laos). The experience acquired by the CNR in Xayaburi allowed it to impose itself on Laotian projects in 2017 in partnership with the company China Power International Development, which solidifies China’s grasp on the hydroelectric energy of the Laotian Mekong and the CNR’s return to the regional scene, 20 years after its initial study8. With the MRC fading away, in 2016, China proposed a new mechanism, the Lancang-Mekong Cooperation Mechanism, which aims to promote cooperation between the basin’s states. China will provide loans to the Asian Development Bank for the improvement of infrastructures as part of the Greater Mekong Subregion Project, which associates the Yunnan and the countries along the Lower Mekong; it also proposes significant financing by the Asian Infrastructure Investment Bank. Moreover, China has expressed its good will by releasing water from its reservoirs when drought affected the hydrology downstream. The crisis thus allowed China to present its storage dams not as environmental disruption factors, but as a new form of mutual assistance. 5.3. The Mississippi, an altered river in a new country 5.3.1. Basin and hydrology The Mississippi owes its current name to an Algonquin word meaning “large river”; it was also the “Father of Waters”. The third longest river in the world at

8 China also exercises great pressure in mastering agricultural land in riverside nations. In 2011, Cambodia developed its land concession policy for economic use, started in 1995. Property titles often having been destroyed under the Khmer Rouge, the villagers cannot oppose expropriation. As an example, in the north of the Preah Vihear province, the land of 25 villages were handed over to Chinese sugar companies; the conceded area amounts to 40,000 ha, affecting 25 villages and 25,000 people, who lost their rice fields and forest access. In total, more than 10% of the country’s area was conceded in 2012, 2.1 million ha, displacing more than 770,000 people; China obtained 0.4 million ha (Rina Chandran for Reuters, November 9, 2017). This dynamic interferes with the limitation of the resources that the population can draw from the Mekong once the dams are built.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

167

3,770 km (behind the Amazon and the Nile), the Mississippi drains a basin covering 3,225,000 km2, which is actually the fourth largest in the world (after the Amazon, the Congo and the Nile). The Mississippi floodplain is 70% occupied by the tributary basins subsequent to the Missouri River (45% of the area, but 14% of the mean annual discharge at the mouth and 84% of the suspended load in the 19th Century), the Ohio River (15%, but 43% of the module at the mouth) and the Arkansas River (15%) [KNO 07]. The hydrology of the Mississippi River basin is strongly influenced by the east–west dissymmetry that exists in the contributions from the tributaries flowing from the well-watered Appalachians (Ohio River, Tennessee River: 1,000 mm/year) and those from the Great Plains with their semiarid climate (Missouri River, Arkansas River, Red River: 400–1,000 mm/year); precipitation drops from Louisiana (+1,400 mm) towards the Upper Mississippi (600–800 mm/year). The geography of evapotranspiration and precipitation explains that the depth of run-off comprised between 10–100 mm in the west and 300 mm in the east of the basin. It is the Ohio River that exercises the greatest influence on the river’s hydrology, despite the small size of its basin, notably through its flood peaks at 12–13,000 m3/s, with a maximum of 30,000 m3/s. 5.3.2. Geology of the Mississippi basin Since the Jurassic Period, the Mississippi has occupied a tectonically formed trough inclined towards the Mississippi Embayment* and the Gulf of Mexico; its edges, the Rocky Mountains to the west and the Appalachians to the east, then rose and provided sediment through the tributary system that quickly appeared. The upper basin was subject to several periods of quaternary glaciation, the last two of which correspond to the so-called “Illinois” and “Wisconsin” glacial stages (55,000– 24,000 years BP), and the paleo-networks were greatly disturbed, which explains the fact that the current network is quite young. The glacial deposits, the river deposits from the Laurentide Ice Sheet and the periglacial Loess cover played an important role as sediment sources and in feeding the Gulf of Mexico. The glacial lake Agassiz is one of the several proglacial lakes* in the Laurentide Ice Sheet that, following the rupture of an ice dam, brutally flowed into the Upper Mississippi between 12,100– 9,400 years BP. Along with other lakes of this type, it is an ancestor of today’s Great Lakes. Downstream of the Upper Mississippi cut in the glacial deposits, the meanders succeeded the late glacial river braiding during the Holocene. Five meander belts from the Holocene have been discovered at the surface of the deposits that occupied the Lower Mississippi Embayment after 14,000 years BP (Figure 5.6). Once the sea level had stabilized, its influence on the aggradation of the Mississippi channel became manifest on a 700 km segment due to the river’s very heavy load. The avulsion of the river channel during severe flooding allowed for the construction of five vast delta lobes in the Gulf of Mexico, including the currently active lobe, which took shape about 1,275 years BP.

168

Sedimentary Crisis at the Global Scale 1

5.3.3. Aspects of the river Upstream of the confluence with the Ohio River, the Upper Mississippi flows over the substrate and, in the 19th Century, braided locally due to its abundant sandy bed load. For navigational purposes, the rapids were destroyed with dynamite starting in 1838; in the river segments, the channel was dredged and maintained by groynes or equipped with dams with locks and secondary arms that were cut off by the Corps of Engineers starting in the 1860s. A new large-scale development scheme, with 29 dams and locks, was developed between Saint Paul, in the State of Minnesota, and Saint Louis (1930–1935); the channel was shrunk through the deposition of material dredged from the navigation channel. From the confluence of the Missouri River to that of the Ohio River starts the open river, with artificially stabilized banks and a channel that was deeply incised in response to the sediment deficit resulting from the construction of dams on the Missouri River.

Figure 5.6. Relative age of the meander belts of the Mississippi as interpreted by H. Fisk between 1941–1944; they were shaped over several millennia. Extracted from folio 22–9, Cape Girardeau MO – Donaldsonville (Louisiana), located immediately upstream of the delta (source: USGS [FIS 44]). For full color image see: www.iste.co.uk/bravard/sedimentary1.zip

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

169

The Lower Mississippi extends downstream from the confluence with the Ohio River (altitude: 70 m). The channel has a rather steep slope up to 40 km north of Vicksburg, which allows it to displace an abundant sandy load. Neotectonics, however, has affected the visible substrate and formed irregularities in the profile. Moreover, the artificial cutoff of the meanders caused local slope adjustments, as well as variations in sediment transfer (through floor erosion), which affect the width of the channel and its sinuosity. Levees have protected the floodplain since the late 18th Century; the breaches caused by the 1927 flood led the Mississippi River Commission to combine this method of defense with the opening of spillways upstream of New Orleans and the construction of large storage dams on its tributaries. The progressive raising of the levees encouraged deposition on the edges of the low-water channel, to the detriment of silting on the floodplain and the delta (Figure 5.7). The lack of contributions to the floodplain is responsible for the increasing extension of the delta’s swamps. Furthermore, since the 1930s, the shortening of the channel – from 10–40% depending on the reaches – and its dredging in favor of navigation led to the steepening of slopes, followed by heavy incision upstream of the confluence with the Red River; bank protection also prevented the channel from developing sinuosity capable of lengthening the channel in Louisiana and compensating for the impact caused by development. Finally, the diversion of part of the discharge towards the Atchafalaya River downstream of the confluence with the Red River has reduced the river’s transport capacity downstream of the Old River Control Structure’s water intake; this method of management causes the aggradation of the Mississippi’s main channel [WIN 94, KES 89].

(a) September 1934

170

Sedimentary Crisis at the Global Scale 1

(b) August 1936

(c) May 1946 Figure 5.7. The alteration and change of the Mississippi floodplain’s landscape. a) In 1934 (low water), natural state with a channel blocked with sand; b) in 1936, after longitudinal dykes and groynes were installed; c) in 1946, after the development of river vegetation (source: USGS [ALE 00])

5.3.4. Modifications to the sedimentary budget Storage dams have had an effect on flood hydrology and sediment fluxes. Table 5.4, which is incomplete, shows the reduction in annual sediment fluxes. The first reservoirs on the large tributaries of the fluvial system were those of the Pathfinder Dam (1909) on the North Platte (Missouri basin), the Fort Peck Dam (1940) on the Missouri River9 and the Kentucky Dam (1944) on the Tennessee River. By 1944, these three dams already controlled an area of nearly 300,000 km2, i.e. 10% of the 9 The dams built on the Missouri are, from upstream to downstream, Fort Peck (1940), Garrison (1953), Oahe (1959), Big Bend (1963), Fort Randall (1956) and Gavins Point (1957). The first was built in 1933; the others were decided upon by the Pick-Sloan Flood Control Act in 1944 and completed in 1964.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

171

basin. Upon beginning operations, the Fort Randall storage dam, finished in 1956 on the Lower Missouri, controlled 683,000 km2 of the basin’s area, i.e. approximately 20% of the basin. In addition to the effects of the storage dams (the top cause of the sediment deficit) we must mention several other factors: the success of soil protection practices; sedimentation between the transverse groynes in the Lower Missouri; the effect of dredging the Mississippi (−5 to 7 Mt/year), which placed deposits outside the reach of river erosion and finally, significant sedimentation between the levees on the river banks. These types of traps are less effective today because they are largely saturated and the speed of the decline in the river load has reduced [MEA 10]. River

Pre-dam in Mt/year

Dam construction

Upper Mississippi Tennessee

10.5

1930–1940 1936

Ohio

16% of the river’s 19thCentury load

1885–1929 lock dams

Missouri

289 84% of the river’s 19thCentury load

Mississippi downstream + Missouri Arkansas Mississippi downstream, Old River Control Contribution to coast

1909–1964

Post-dam in Mt/year 0.8

0 (downstream of dams) 78.4 (upstream of Mississippi)

250–275

150–180

93

11

500

150–300

400

170

Table 5.4. Reduced suspended sediment flux at different points in the Mississippi River system [ALE 12, KEO 86, MEA 10]

The reservoirs regulated the hydrological regime and reduced the floods, such that the morphology of the rivers downstream of the Missouri dams was adjusted with the stabilization of the bed, the metamorphosis of the braids into meanders and the colonization of the active tracts by vegetation. The trapping in the large reservoirs of the Tennessee River was drastic and unable to be reduced in the very long term. Sediment trapping by the Missouri’s reservoirs was total, as well as the adjustment of the Kansas River, a tributary to the Missouri, exacerbated this deficit. However, the river progressively recovered downstream of the chain of dams by rearranging the materials on its bed and banks thanks to the low water discharges strongly raised by hydroelectric production. The Missouri River, downstream of the Gavins Point storage

172

Sedimentary Crisis at the Global Scale 1

dam (just downstream of the Fort Randall Dam), still provides the delta with 65% of its suspended load and sand; this contribution takes place through degradation of the channel downstream of the dam, over a 300 km reach ending in Omaha. The Missouri unloaded a very significant volume of material, which led to a more than 3 m drop in the floodwater levels. This impact was thereafter transferred to the Mississippi. The river had certainly recorded a great sediment deficit since the 1960s, the load having dropped by half, but the Mississippi’s levels rose between Minnesota and Louisiana (from 1–4 m for large floods) due to the blockage of the bed. The US-ACE had to raise the relative height of the levels several times; their summits have stood 10.5 m above the alluvial plain since 1978. It goes without saying that this river reach is threatened by avulsions or course changes in the case that breaches open in the levees. Over the river’s downstream course, the rise in water levels is also due to the damming effect caused by the ocean’s surface, i.e. a loss of energy slowing transport capacities. From the so-called source-to-sink perspective, i.e. the transport of sediment from their continental origin to the ocean, which is the ultimate sink, multiple forms of trapping have played and continue to play a major role. The mass of sediment mobilized during the Anthropocene has been fragmented into a number of natural and artificial deposition sites. The most significant impact was the isolation of the artificialized channel vis-à-vis the alluvial plain, isolation that eliminated a major trapping area [BEN 16]10. However, this impact coincides with the stronger period of sediment production (intensive agriculture and soil erosion) and with the absence of dams. This anthropogenic peak in sediment had effects on the style of progradation of the current delta (see Volume 2, Chapter 2). Once the main reservoirs were in place, starting in 1950, the decline in sediment contributions in Tarbert Landing (diffluence of the Mississippi and the Atchafalaya, 500 km from the mouth) was fast and continuous (Table 5.5); moreover, the number of reservoirs today exceeds 40,000! It is remarkable that the Mississippi River system was placed under the influence of hydraulic equipment located more than 2,000 km from the delta given that downstream, the channel is nothing more than an embanked conduit. Period 1950–1953 1970–2013

Flux of suspended material Mt/year and % 463 130 (−72%)

Flux of suspended sand Mt/year and % 78 28 (−65%)

Table 5.5. Decline in MES flux and sand flux at Tarbert Landing between 1950 and the 2010s

10 Source-to-sink goes back to the transport of sediment from their source (production) to successive local trapping on the alluvial plain, the delta and the ocean.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

173

The US-ACE’s large dams blocked the river’s natural tendency to adjust its lateral migration, its avulsions and the shifting of the delta lobes in the framework of an autogenic response (i.e. self-adjustment of fluxes and landforms on timescales ranging from decades to centuries). Despite more than 150 years of costly efforts and engineering science, the Missouri–Mississippi system is still evolving: it incises downstream of the dams and deposits sediment downstream of the river and at its mouths in response to spatial energy variations. Despite the best efforts from generations of engineers, the leveed, gated and dammed Mississippi still demonstrates the same tendency for self-regulation that confronted 19th-Century engineers [BEN 16]. A spatial approach to the river’s behavior has put the changes recorded since the 1930s, just before a period of heavy work, into perspective [KES 03]. At that time, the river was rising on its alluvia nearly all along its course. In this context of sediment retention in the riverbed, the suspended load reaching the Gulf was, despite all inhibiting factors, 0.270 Gt/year and the bed load, 0.130 Gt/year. Two-thirds of the contributions were due to the sapping of the concave banks of the meanders developed between Cairo (downstream of the confluence with the Ohio River) and Red River Landing (station located between Natchez and Baton Rouge). Two-thirds of the storage took place in the short term in the channel as sandbanks strongly connected to bank erosion and retreat. However, the situation was reversed downstream of Red River Landing, two-thirds of the sediment being deposited on the floodplain (thus for the long term) and the rest into the river channel; as for the sandy bed load, it more often reached the end of the delta and the Gulf (Figure 5.8). The large “coating” works on the banks had the effect of reducing 90% of the lateral erosion even before the effects of dam construction on the Missouri and Arkansas Rivers appear, then those of building dyke fields, responsible for trapping a considerable fraction of the sandy bed load after 1955 (Figure 5.9). The sinuous channel reaches, isolated in the 1930s by cutting them off (in the form of oxbow lakes), were also subtracted from the sedimentary budget. At the same time, the sandy bed load accumulated partially in the dammed channel in the New Orleans district. In short, the hydrosedimentary history of the anthropized Mississippi River system is quite young, having been formed less than a century ago. The adjustments to the river, in the reaches that were not equipped with reservoirs, have taken place over the last 50 years, so the situation has not yet stabilized. The river channels have become the primary source of materials destined for the delta. The current deficit can thus only intensify.

174

Sedimentary Crisis at the Global Scale 1

a)

b) Figure 5.8. Decline in the load in the Mississippi basin between 1800–1980. The load of the tributaries on the right bank has been greatly reduced (impact of dams), whereas that of the Ohio River has increased, but to a lesser degree (soil erosion) (source: [MEA 10] and USGS redrawn [ALE 00] for this volume)

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

175

Figure 5.9. The river poplar forest (P. nigra) in the Saurel Bayou, a rather natural landscape (source: J.-P. Bravard)

To date, there is a consensus on the fact that the fluctuations in the Mississippi’s floods were correlated with the El Niño-Southern Oscillation in the Pacific and with the multi-decadal Atlantic oscillations; climate change could explain the modification in the precipitation–evaporation budget, as well as the increases in flows. For some researchers, however, climate phenomena alone cannot explain the 20% increase in the frequency and intensity of 100-year floods* monitored over the past 500 years. Recent studies have been performed on dead arms of the river to

176

Sedimentary Crisis at the Global Scale 1

measure the grain size of the sand deposits brought by floods over the past 500 years and the growth of the tree rings in the hardwood alluvial forest that record the excess dampness related to floods; they have allowed extension of the time series regarding the discharges monitored since the first flood marker was fixed in 1897 in Vicksburg. The authors of these studies believe that the development works on the river and its tributaries (dams, levees) could explain 75% in the worsening of flood conditions [MUN 18]. 5.4. Overexploited rivers in regions with a water deficit 5.4.1. The God River and the Aswan Dams The Nile, stretching more than 6,850 km, has a 3.25 million km2 watershed. Today, it provides for 200 million people. Its water comes from the African Great Lakes (the White Nile) and the Ethiopian mountains (the Blue Nile, which drains the mountain range in Ethiopia to the west and converges with the White Nile in Khartoum, and the Atbarah River, which drains the northern portion of the Ethiopian mountains). As it enters Egypt, the Nile’s flow was 84 km3/year on average before the hydraulic adjustments, but it fluctuated from one year to the next, with a minimum of 12 km3 in 1913–1914 and a maximum of 155 km3 in 1878–1879. The discharge fluctuated throughout the year between values of 900 m3/s at its low point and 10,000 m3/s during the flood. The discharge rate peaked in late August; its rise, which started in July, was faster than the fall, which finished in late February. The flood peak came from the Atbarah River and the Blue Nile, while the water from the recession phase came from the White Nile, fed, with some delay, by the equatorial rains falling on the Great Lakes’ plateau. Two-thirds of the discharge in Egypt and 95% of the sediment come from the Blue Nile, which drains the highlands of the Ethiopian rift. The suspended sediment load averaged 160 Mt/year and the (sandy) bed load 20 Mt/year. During the millennia of hydraulic civilization, serving irrigation and agricultural production in the Nile Valley, the river flowed unhindered. The only modifications (which are non-negligible), were the organized flooding of the plain beyond the natural limit of free flooding; there was also organized sedimentation, the Nile’s silt being responsible for the fertility of the flooded areas. The concern to irrigate in a vast if not permanent manner materialized in the construction of dams at the head of the delta on the branches in Damietta, then Rosette (1843–1861); the dams divided the water between canals designed for gathering the entire discharge arriving to the delta. All that remained was to break away from the constraints of dry years. The former equilibrium could not resist the progress that demanded reinforced irrigation in terms of both volume and duration. This all played out at the Aswan site. We will see that the first dam built starting in the late 19th Century attempted to respect the

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

177

Nile’s nature as best it could. The Great Dam, on the contrary, introduced a fundamental break with this, as it succeeded in eliminating the annual flood and its attributes. 5.4.1.1. The Old Aswan Dam Ever since Antiquity, the Nile’s alluvial plain has been equipped with channels rerouting its silty flooding into vast developed parcels of land, the so-called “hods”. The latter only function when the flood level exceeds 7 m. The first storage dam in Aswan was built to hold back a volume of water capable of ensuring sufficient discharge to the hods, as well as to overcome the long periods of low waters that are harmful to cotton production. The project aroused a heated geopolitical debate between the UK, who had watched over their protectorate since 1882, and the French, who contested this. In 1890, Victor Prompt, a graduate of the École Polytechnique who was not free of political intentions in the context of opposing Marchand and Kitchener, proposed the construction of a storage dam with no bottom valve on the Fachoda site in Sudan; one project that benefitted from the watchful eye of Mahdi, the nation’s “savior” in the face of Ottoman domination. The Aswan Dam project was UK’s reaction to the French project, which was somewhat provocative and would most likely have been less useful for irrigation. The Technical Commission on the reservoirs met in Cairo in 1894 with the goal of providing the Egyptian government with an evaluation regarding the UK project to irrigate using the waters from the permanent part of the Nile and to ensure protection against heavy floods [GAR 94]. The principle of a dam was kept for the site at the first cataract, 7 km downstream of Aswan. The dam would have two characteristics: 1) the evacuation of floodwaters and silt from the annual flood through gates closed by rising metal doors; 2) the retention of water in the winter in a long reservoir with a rather modest capacity. Construction took place between 1899 and 1902 according to the plans drawn by engineer William Willcocks. This first dam was a wall made of stacked granite blocks, built in a straight line 1,960 m long, with a height of 30.5 m and lined with 180 evacuation gates in two superimposed rows, 140 of which could be found in the lower row11. The height was raised twice to increase the water reserve. The second

11 An ingenious device associates one granite basin gathering the water that passes through one part of the gates and a “current counter” that measures a partial discharge. Through extrapolation, this allowed the “primary” Nile’s discharge to be measured. One of a kind, this device was considered the starting point for reduced models subsequently used when designing hydraulic dams, notably the other dams on the Nile.

178

Sedimentary Crisis at the Global Scale 1

instance, in 1934, created the “old dam” we can still see today, which controls a water reserve of 5.1 km3; however, this volume only represents 6% of the Nile’s annual discharge, such that the old dam could not significantly modify the river’s seasonal regime (Table 5.6). Aswan Dams

Construction dates

Height (m)

Reservoir (km3)

Reservoir length (km)

First

1899–1902

18,35

1

225

Second

1907–1912

25

2.4

295

Third

1929–1934

31

5.1

360

Table 5.6. The Old Aswan Dam and its successive improvements (source: [BES 57, HUR 54])

The reservoir evened out the discharge variations and the opening of the gates allowed excess flow to be evacuated. The water stored from October to January was released between February and July towards the irrigation canals. Why, then, were so many gates installed in the dam? The goal was to homogenize the floodwater speed throughout the width of the reservoir and, as such, optimize the evacuation of silt through the dam’s wall. It is remarkable that the late 19th-Century engineers sought to reduce sedimentation in the reservoir; it is true that Willcocks, one of the best engineers of his day, had acquired a wealth of experience in India, Australia and South Africa. The dam, despite the reservoir’s length, could not hold the silt from the annual flood, on the one hand, so that the reservoir did not lose its capacity and, on the other hand, to guarantee the proper renewal of fertility in the downstream irrigated parcels of land. Before the construction of the Aswan High Dam, the average concentration of silt at the entry into Egypt was relatively low, with 1.6 g/l from August to October, the highest level measured having been 4 g/l. The annual flux of “silt” (a rather restrictive term given that the load was partially sandy) was, on average, 100 million tons at the entry into Egypt according to Hurst, 30 of this being fine sand, 40 being actual silt and 30 being clay; the sediment load was not greater than 57 Mt in Cairo12 12 Important measurements provided the knowledge that in 3 years (1929–1931), the Nile lost 52.5 Mt of “silt” between Wadi Halfa (at the entry into Egypt), Aswan and Cairo during the high waters. This was thus about half of the sediment load deposited along the way. The fine portion was deposited in the canals, the hods and the drains at a rate slightly below 2.5 t/ha. The average contribution to the fields along the Nile logically decreased downstream; it was approximately 1 mm/year in Upper Egypt, 0.3 mm in Mid-Egypt and 0.06 in Lower Egypt. Fine sand was missing because it was not measured downstream; in a less turbulent flow, it was transported along the channel floor to the mouths of the delta.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

179

and the sandy portion was reduced downstream, while that of clay increased. The principle of the dam was to fully evacuate the load in the initial phase of the flood, then to only store the water in the Aswan reservoir when it had become clear again13. The measurements taken on the reservoir floor do not, according to Hurst, show the deposition “of appreciable quantities” of silt [BES 57, HUR 54]14. This original management principle was modified after the rising of the dams required more water to fill a reservoir with a greater volume of water. The closing of the gates was moved forward from late October to mid-October to better trap the end of the flood. The quantity of trapped silt could have then increased, but an assessment made in 1963 stated that the technique used had prevented any deposition, which was a great success in itself. The Old Aswan Dam was completed by the deviation dams located further downstream and simply meant to raise the waters during the low-water period towards the irrigation canals; these were the Assiut (1898–1902, then 1938), Zifta (1902), Esna (1906–1909, then 1947), Nag-Hamadi (1928–1930) and Edfina (1951) Dams, the last of which was built on the Rosette branch to limit the water volumes heading towards the sea. These ensured the long-term irrigation allowed by the Aswan reservoir. Their performance is still considered excellent today, even though they had the effect of creating a maximum limitation on the discharge entering the Mediterranean [HAR 96]. 5.4.1.2. The first impacts of the Old Dam One negative impact of the increase in the water volumes allocated to irrigation and particularly their long-term use was quickly noted. The dam had made more water available, to the point that it reached the irrigated parcels in excessive quantities. In the Lower Egyptian delta, where the groundwater level is near the surface, it rose, favoring ascending capillary action and leading to a drop in cotton, rice and corn yield returns associated within winter crop rotation. Drainage then became necessary.

13 The silty water was collected, weighed with its silt, then weighed again with only the silt after evaporation; such works, performed over several seasons, provided the experience on which the dam’s initial management was based. 14 The most informative work on this matter was written by a British engineer who actively took part in the measurements, Hurst; in office along the Nile from 1905 to 1946, he worked until the late 1960s [HUR 54]. The very good book by J. Besançon is less precise regarding the technical aspects of the dams than Hurst’s [BES 57]. The latter was dedicated to the Lord of the Mud, the fellah, while the former was dedicated to British engineers.

180

Sedimentary Crisis at the Global Scale 1

5.4.1.3. The Aswan High Dam The first studies were performed at the site of the Fourth Cataract, with the initial goal of eliminating the danger of very heavy floods like those of 1878 and 1946; they also aimed to store water for the dry season at an Egyptian site, thus one independent of geopolitical considerations seen as unstable by the Egyptians in the upper basin. The project was approved by several Western governments, but their financing project failed due to Soviet implication in the context of the Cold War [HOW 94, SAI 93]. The dam was built in the 1960s; it was locked in 1964 and filled in 196715. The Old Aswan Dam was kept and serves for energy production (500 MW) and to regulate the daytime lockages of the High Dam. The effects of the High Dam on the reservoir’s hydrology and morphology and those of the downstream Nile have been well-studied. The currently managed water discharge has been reduced to 55 km3/year, considering losses through evaporation and infiltration. The sediment load transported by the Blue Nile fills the “dead zone” of Lake Nasser, the total deposition measured between 1964–1998 rising to 3.5 billion m3. In the early 1990s, sediment storage in the dead zone, upstream of the reservoir, had progressed to 230 km since the locking of the dam (out of the 500 km of the reservoir). An unknown portion of backfill is caused by sand transported from dune fields by wind energy, which is no longer carried away by the Nile’s annual flood. Downstream, the Great Dam limits the maximum discharge to 2,500 m3/s in favor of reservoir storage, which is meant for perennial irrigation. A high-up representative of the Egyptian Public Authority for Drainage Projects stated that the “annual loss of water to the sea” before the construction of the Great Dam was one of the “Nile’s problems” [SAL 76]; this problem, if it was one, has largely disappeared. Agricultural specialists also note an increase of 1 ppm/year in the total concentration of dissolved salt in the lake waters since the dam was locked. The almost complete optimization of the resource for agriculture was one of the set economic goals of this development scheme, but 10 years later, there was absolutely no mention of potential impacts on the delta, even in regards to agriculture. Moreover, the Great Dam has practically reduced the sediment load transported to Aswan to zero. The silt concentration in the water at El Gaafra dropped from 3.8 g/l before the dams to 0.05–0.01 g/l after the dam was locked (the trapping efficiency is nearly 100%). However, pessimistic predictions regarding bed erosion (between 2–10 m depending on the model used), the surplus energy no longer

15 Total volume of the reservoir: 162 km3, useful storage: 107 km3, dead storage: 55 km3 and installed power: 2,100 MW.

The Recent Hydrosedimentary History of Some of the Globe’s Largest Rivers

181

dissipating during transport, were largely denied by the facts. The erosion of the bed took place downstream of some dams between Aswan and the delta, but it occurred mostly during the first years after the High Dam was locked, its depth being less than 1 m, and it stopped in the mid-1980s. Water release management at the reservoir for agricultural purposes is realized at rates comprised between 153–230 hm3/day, a discharge that corresponds to the needs along the 950 km between Aswan and the Delta Dam, a distance to which should be added the length of the Damiette (225 km) and Rosette (245 km) branches. It has also been acknowledged that bed erosion takes place when the daily discharge exceeds 230 hm3, thus a value that should not be exceeded. However, the Nile seems to be in equilibrium and the material proceeding from the channel’s morphological evolution should not represent a significant volume reaching the sea. On a large scale, bank erosion and wind contribution can locally aggrade the river bottom and heavy floods can scour the channel. The Nile has actually kept little energy to erode its banks and transport its residual load due to the regulation of its discharge; the local sinuosities have developed little, and islands have been attached to the banks in such a way that the bed has shortened and contracted. One of the additional consequences of suppressing silt transfer along the Nile is the cessation of sediment deposition that benefited 7,000 brickworks all along the river. Companies transferred themselves to the floodplain, where 120 km2 were taken back each year from agricultural production to be used for extraction. The government banned extraction from agricultural land in 1984, but its decision was followed by a drastic loss of 1,000 km2 of land, without any real effects of public regulation in the following years [WHI 88]. More difficult to understand is the question of nutrient loss suffered by agriculture due to the drastic reduction of the silt flux previously contained in the water. A reasonable synthesis of the effects recorded could be as follows: 1) drop in contributions in the hods fed directly by the flood before the dam completion; 2) absence of notable change for the land parcels irrigated throughout the year for decades; 3) absence of negative evolution downstream from Cairo, where the bulk of silt contributions was evacuated to the sea with no benefit for the land parcels, except for 10% of them, near the canals;

182

Sedimentary Crisis at the Global Scale 1

4) despite its reputation for fertility, the Nile’s silt is naturally poor in nutrients (nitrogen), so fertilizer has long been used in the valley to increase productivity [BIS 12, SAI 93]16. Dates

Khartoum

Aswan

Delta

Ql km3/year

Qs Mt/year

Ql km3/year

Qs Mt/year

Ql km3/year

Qs Mt/year

Before 1902

110 wet phase

200

110 wet phase

200

62.5

180 MES 160

1902–1964

80

200

80

120–160

50

100–120

Today

80

230 ± 20

55 Evap. 10 Sudan 15

0

12