Advances In Geosciences (A 4-volume Set) - Volume 28: Atmospheric Science (As) And Ocean Science (Os) : Atmospheric Science and Ocean Science 9789814405683, 9789814405676

Citation preview

A d v a n c e s

i n

Geosciences Volume 28: Atmospheric Science (AS) and Ocean Science (OS)

8474hc.v28.9789814405676-tp.indd 1

11/7/12 10:08 AM

ADVANCES IN GEOSCIENCES Editor-in-Chief: Kenji Satake (University of Tokyo, Japan) A 5-Volume Set Volume 1: Solid Earth (SE) ISBN-10 981-256-985-5

A 6-Volume Set

Volume 2: Solar Terrestrial (ST) ISBN-10 981-256-984-7

Volume 17: Hydrological Science (HS) ISBN 978-981-283-811-7

Volume 3: Planetary Science (PS) ISBN-10 981-256-983-9

Volume 18: Ocean Science (OS) ISBN 978-981-283-813-1

Volume 4: Hydrological Science (HS) ISBN-10 981-256-982-0

Volume 19: Planetary Science (PS) ISBN 978-981-283-815-5

Volume 5: Oceans and Atmospheres (OA) ISBN-10 981-256-981-2

Volume 20: Solid Earth (SE) ISBN 978-981-283-817-9

A 4-Volume Set

Volume 21: Solar Terrestrial (ST) ISBN 978-981-283-819-3

Volume 6: Hydrological Science (HS) ISBN 978-981-270-985-1

Volume 16: Atmospheric Science (AS) ISBN 978-981-283-809-4

Volume 7: Planetary Science (PS) ISBN 978-981-270-986-8

A 6-Volume Set Volume 22: Atmospheric Science (AS) ISBN 978-981-4355-30-8

Volume 8: Solar Terrestrial (ST) ISBN 978-981-270-987-5

Volume 23: Hydrological Science (HS) ISBN 978-981-4355-32-2

Volume 9: Solid Earth (SE), Ocean Science (OS) Volume 24: & Atmospheric Science (AS) ISBN 978-981-270-988-2 Volume 25: A 6-Volume Set Volume 10: Atmospheric Science (AS) Volume 26: ISBN 978-981-283-611-3

Ocean Science (OS) ISBN 978-981-4355-34-6 Planetary Science (PS) ISBN 978-981-4355-36-0 Solid Earth (SE) ISBN 978-981-4355-38-4

Volume 11: Hydrological Science (HS) ISBN 978-981-283-613-7

Volume 27: Solar Terrestrial (ST) ISBN 978-981-4355-40-7

Volume 12: Ocean Science (OS) ISBN 978-981-283-615-1

A 4-Volume Set

Volume 13: Solid Earth (SE) ISBN 978-981-283-617-5

Volume 28: Atmospheric Science (AS) & Ocean Science (OS) ISBN 978-981-4405-67-6

Volume 14: Solar Terrestrial (ST) ISBN 978-981-283-619-9

Volume 29: Hydrological Science (HS) ISBN 978-981-4405-70-6

Volume 15: Planetary Science (PS) ISBN 978-981-283-621-2

Volume 30: Planetary Science (PS) and Solar & Terrestrial Science (ST) ISBN 978-981-4405-73-7 Volume 31: Solid Earth Science (SE) ISBN 978-981-4405-76-8

A d v a n c e s

i n

Geosciences Volume 28: Atmospheric Science (AS) and Ocean Science (OS)

Editor-in-Chief

Kenji Satake

University of Tokyo, Japan

Volume Editor-in-Chief

Chun-Chieh Wu

National Taiwan University, Taiwan

Jianping Gan

Hong Kong University of Science and Technology, Hong Kong

World Scientific NEW JERSEY

•

LONDON

8474hc.v28.9789814405676-tp.indd 2

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

11/7/12 10:08 AM

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

ADVANCES IN GEOSCIENCES A 4-Volume Set Volume 28: Atmospheric Science (AS) and Ocean Science (OS) Copyright © 2012 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN 978-981-4405-66-9 (Set) ISBN 978-981-4405-67-6 (Vol. 28)

Typeset by Stallion Press Email: [email protected]

Printed in Singapore.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

EDITORS

Editor-in-Chief:

Kenji Satake

Volume 28: Atmospheric Science (AS) and Ocean Science (OS) Editor-in-Chief: (AS) Chun-Chieh Wu Editor-in-Chief: (OS) Jianping Gan Editors: (AS) Kevin K. W. Cheung Hyun Mee Kim Tieh-Yong Koh Mong-Ming Lu Seon-Ki Park Editor: (OS) Minhan Dai Volume 29: Hydrological Science (HS) Editor-in-Chief: Gwo-Fong Lin Editors: Kwan Tun Lee Sanjay Patil Srivatsan Vijayaraghavan Volume 30: Planetary Science (PS) and Solar & Terrestrial Science (ST) Editors-in-Chief: (PS) Anil Bhardwaj Vikram Sarabhai Editor-in-Chief: (ST) Andrew W. Yau Editors: (PS) Takashi Ito Paul Hartogh Editors: (ST) Yusuke Ebihara Susan Mckenna-Lawlor Gang Lu

v

July 13, 2012 9:9

vi

AOGS - AS and OS

Editors

Volume 31: Solid Earth Science (SE) Editor-in-Chief: Ching-Hua Lo Editors: Yih-Min Wu J. Gregory Shellnutt

9in x 6in

b1387-fm

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

REVIEWERS

The Editors of Volume 28 would like to acknowledge the following referees who have helped in review of the manuscript publish in this volume: Byung-Gon Kim Hee-Sang Lee Xianxiang Li Tsung-Ren Peng Vijay Tallapragada Kei Yoshimura Jun Sun Satya Prakash Joji Ishizaka Zhongming Lu Hiu Suet Kung

Hung-Chi Kuo Young-Hee Lee Chun Fung Lo Tetsuya Takemi Qin Xu Sixiong Zhao Wen-Chen Chou F. Fang Ken T. M. Wong S Peng

vii

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-fm

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

PREFACE

The present volume set (volumes 28 to 31) of Advances in Geosciences (ADGEO ) is the sixth round of ADGEO series edited by the Asia Oceania Geosciences Society (AOGS), and contains papers presented at the eighth annual meeting held in Taipei in 2011. The AOGS is an international society legally registered in Singapore, aiming to cooperate and promote discussion on studies of the Earth and its environment, as well as the planetary and space sciences. To achieve this objective, the AOGS has held its annual meetings since 2004. The AOGS has six sections, Atmospheric Sciences (AS), Hydrological Sciences (HS), Ocean Sciences (OS), Planetary Sciences (PS), Solar and Terrestrial Sciences (ST), and Solid Earth Sciences (SE). In the current set, papers presented at AS and OS sections are included in volume 28, those at HS section are in volume 29, at PS and ST sections are in volume 30, and at SE section are in volume 31. ADGEO is not a scientific journal, but a monograph series or proceeding volumes of the AOGS meetings. Only papers presented at the AOGS meetings are invited to ADGEO series, and are published after peer reviews. The first (volumes 1 to 5), second (6 to 9), third (10 to 15) sets corresponded to the second, third and fourth AOGS annual meetings. The fourth volume set (16 to 21) included papers presented at the fourth and fifth annual meetings, and the fifth set (22 through 27) included those at the sixth and seventh meetings. As a young scientific society, AOGS needs to develop ways to promote information exchange and interaction among scientists in Asia and Oceania region, in the era of internet. Until we establish a journal or other means of publication, ADGEO is expected to serve as a publication tool among the AOGS members and society at large. Finally, I would like to thank authors, reviewers, volume editors and volume editors-in-chief for their timely efforts to publish the current

ix

July 13, 2012 9:9

x

AOGS - AS and OS

9in x 6in

b1387-fm

Preface

volume set, Meeting Matters International (MeetMatt) for developing and maintaining a system for submission, review and editorial processes, and World Scientific Publishing Company (WSPC) for the editorial, publication and marketing processes.

Kenji Satake Editor-in-Chief

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

PREFACE TO AS VOLUME

This volume contains papers contributed from AS section of the AOGS scientific meeting 2011 in Taipei, Taiwan. Since the public interest in atmospheric research is increasing as concern grow regarding severe weathers associated with tropical cyclones and heavy rainfall events, climate change and air pollution and its effect on monsoon variability, climate change due to anthropogenic and natural processes, and issues on special observations, numerical modeling and data assimilation. Scientists from around the world were invited to present their original research, and requested them to submit full manuscripts for peer review and finally for publication in AOGS volumes. The submitted papers cover four different areas of atmospheric fields, such as 3DVAR data assimilation in numerical model, investigation of numerical prediction for tropical cyclones, understanding of stable isotopes in precipitation, and analyses of dust deposition over land and seas. The publication of this AS volume would not be possible without the cooperation of the authors, the ADGEO AS editorial team as well as ADGEO editorial team, WSPC, and the AOGS Secretariat Office. Therefore, I would like to take this opportunity to thank the Volume Editors in Atmospheric Science who are the driving force in making ADGEO possible: Prof. Kevin K. W. Cheung (Macquarie University), Prof. Hyun Mee Kim (Yonsei University), Prof. Tieh-Yong Koh (Nanyang Technological University), Dr. Mong-Ming Lu (Central Weather Bureau), and Prof. Seon-Ki Park (Ewha Womans University). They have worked very hard to ensure both the quantity and quality of the published papers in AS volume of ADGEO.

Chun-Chieh Wu AS Section Editor-in-Chief

xi

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-fm

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

PREFACE TO OS VOLUME

Ocean science is a branch of geosciences and plays an important role in climate and environment variability. This volume collects some of the papers in ocean science that presented in 2011 Asia Oceania Geosciences Society annual meetings in Taipei. These papers report recent research results about biogeochemical response to climate change and about Tsunami dynamics in the oceans of Asia Oceania region. The studies are based on field and remote sensing measurements, computational fluid dynamics and laboratory measurements.

Jianping Gan OS Section Editor-in-Chief

xiii

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-fm

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-fm

CONTENTS

Editors

v

Reviewers

vii

Preface

ix

Preface to AS Volume

xi

Preface to OS Volume

xiii

Forecast Skill and Computational Cost of the Correlation Models in 3DVAR Data Assimilation M. Yaremchuk, M. Carrier, H. Ngodock, S. Smith and I. Shulman Real-Time Tropical Cyclone Prediction Using COAMPS-TC J. D. Doyle, Y. Jin, R. M. Hodur, S. Chen, H. Jin, J. Moskaitis, A. Reinecke, P. Black, J. Cummings, E. Hendricks, T. Holt, C.-S. Liou, M. Peng, C. Reynolds, K. Sashegyi, J. Schmidt and S. Wang Factors Controlling the Spatial Distribution of Stable Isotopes in Precipitation over Kumamoto, Japan Masahiro Tanoue, Kimpei Ichiyanagi, Jun Shimada and Naoki Kabeya Asian Dust Deposition Over the Land and Seas in 2010 Estimated by the ADAM2 Model Soon-Ung Park, Moon-Soo Park and Anna Choe

xv

1

15

29

41

July 13, 2012 9:9

xvi

AOGS - AS and OS

9in x 6in

b1387-fm

Contents

Distribution of Biogenic Silica in the Upwelling Zones in the South China Sea Yang Liu, Minhan Dai, Weifang Chen and Zhimian Cao

55

Influence of Velocity Distribution and Density Stratification on Generation or Propagation of Tsunamis Taro Kakinuma, Kei Yamashita and Keisuke Nakayama

67

Sea surface PCO2 in the Indian Sector of the Southern Ocean during Austral summer of 2009 Suhas Shetye, Maruthadu Sudhakar, Rengaswamy Ramesh, Rahul Mohan, Shramik Patil and Amzad Laskar

79

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch01

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


FORECAST SKILL AND COMPUTATIONAL COST OF THE CORRELATION MODELS IN 3DVAR DATA ASSIMILATION M. YAREMCHUK∗ , M. CARRIER, H. NGODOCK, S. SMITH, and I. SHULMAN Naval Research Laboratory Stennis Space Center, MS 39529, USA ∗ [email protected]

Many background error correlation (BEC) models in data assimilation are formulated in terms of a positive-definite smoothing operator B which simulates the action of correlation matrix on a vector in state space. To estimate the efficiency of such approach, numerical experiments with the Gaussian and spline models „ « ∇ν∇ −m , B = exp(∇ν∇); Bm = I − m have been conducted. Here I is the identity operator and ν is the diffusion tensor, whose spatial variability is derived from the forecast field and m is the spline approximation order. Performance of these BEC representations are compared in the framework of numerical experiments with real 3dVar data assimilation into the Navy Coastal Ocean model (NCOM) in the Western Tropical Pacific. It is shown that both BEC models have similar forecast skills over a two-month time period, whereas the second-order spline model is several times more efficient computationally, if the cost function is minimized in the state space.

1. Introduction In recent years, heuristic background error correlation (BEC) modeling has become an area of active research in data assimilation (DA). This interest has been fueled by development of the ensemble DA techniques and rapid

1

July 13, 2012 9:9

2

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

increase of the data streams, driven by remotely sensed observations from satellites and introduction of new autonomous observational platforms in oceanography. Traditional BEC models based on the explicit definition of the error covariances by the families of parameterized correlation functions1,2 tend to loose computational efficiency with the growth of the number of observations and with the necessity to introduce more complex observation operators into the DA algorithms. Because of that, there is a growing tendency to estimate error correlations directly in the model space (MS) or in its subspaces spanned by the appropriately selected basis functions.3–5 Of particular interest are the MS correlation models based on positive functions of the diffusion operator D = ∇ν∇, such as the exponent or the inverse of its binomial approximation.6,7 Both types of models were extensively used in many applications8–11 because of their convenience and ease of numerical implementation. Another advantage is their flexibility in approximation of inhomogeneous and anisotropic covariance functions.12,13 Numerically, these models are implemented by integrating the diffusion equation (DE) using either explicit or implicit scheme. The computational cost of the DE integration by explicit methods increases substantially when the local decorrelation scale, ρc , becomes larger than the model grid step δx. This is because the minimum number of multiplications by D, is proportional to (ρc /δx)2 — a constraint imposed by numerical stability of the integration. In such situations it may be advantageous to employ implicit integration schemes,11,14 which tend to converge fast enough to deliver considerable computational gain. Additional gain can be obtained if the implicit approximation is implemented within the MS formulation. In this case, the iterative solution of the system in data space (DS) which embeds an iterative cycle of the implicit scheme, is no longer needed. This study compares the forecast skill and computational cost of two BEC models: The first model (C∞ ) is described by the propagator of the DE and implemented numerically by its explicit integration; the second BEC model (Cm ) is defined by the inverse of a mth-order binomial of D, that approximates C∞ and can be interpreted as a result of m-step DE integration with the implicit scheme. Assimilation experiments were performed using both DS and MS formulations of Cm and a realistic regional ocean model with real data. It is shown that in certain situations it is computationally advantageous to employ the second (spline) model combined with the MS solution to the normal equation.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models

3

2. Covariance Modeling in MS 2.1. MS and DS approaches in 3dVar In its basic formulation, the 3dVar analysis determines the optimal model state increment x that minimizes the following cost function, J(x) =

 1
T −1 x B x + (Hx − d)T R−1 (Hx − d) → min . x 2

(1)

Here B is the BE covariance matrix and d is the innovation vector d = y − Hxb , where y denotes observations, xb is the background model state, H is the observation operator (linearized) in the vicinity of xb , and R is the covariance matrix of observation errors. To simplify the notation, the variables in both model and data spaces are non-dimensionalized by x ← C−1/2 x and d ← R−1/2 d, where C is the (diagonal) background error variance matrix. In order to keep J invariant, the matrices B, and H are non-dimensionalized by B ← C−1/2 BC−1/2 ; H ← R−1/2 HC1/2 . The cost function (1) is minimized by solving the normal equation which sets the gradient of J equal to zero: (B−1 + HT H)x = HT d,

(2)

so that the solution to the normal equation is: x = (B−1 + HT H)−1 HT d.

(3)

Solving Eq. (2) for the model state increment x is the basic tool of 3dVar analysis. If B−1 has full rank, the solution (3) is unique and can be rewritten in the dual form15 : x = BHT (HBHT + I)−1 d,

(4)

which is often called the DS solution to the variational problem (1). Note that if B−1 does not have full rank, defining B as its generalized inverse does not guarantee that solution (4) will coincide with the solution (3) of the original minimization problem. This is because the DS solution (4) is always orthogonal to the null space of B, whereas in general, the minimizer (2) of (1) is not constrained by this condition. It should also be noted that the majority of the BEC models are based on direct computation of the matrix elements of B from experimental data,

July 13, 2012 9:9

4

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

and therefore, require 3dVar solutions in the forms, involving B.√These are ˜ = B16,17 : the DS solution (4), or the MS solution, preconditioned by B ˜ +B ˜ T HT HB) ˜ −1 BH ˜ T d. x = B(I

(5)

Therefore, the basic MS solution (3) has been used in practice more rarely for two reasons: (a) it requires solving the linear system in MS, which has many more dimensions than the DS; and (b) estimation of B from the data is more straightforward than estimation of B−1 .

2.2. The Gaussian and spline BEC models The two types of BEC models considered here are based on the polynomials of D. The major idea is to model the resulting action of the BEC operator B on a vector x by integrating the corresponding DE 1 ∂x = Dx ≡ ∇ν∇x ∂t 2

(6)

for a certain “time period” T , thus setting B = exp T D. The diffusion tensor ν is represented by 3 × 3 positive-definite matrices whose entries depend on the coordinates x in physical space. The eigenvalues λ2i , i = 1, . . . , 3 of νT are all positive, have the dimension of length squared, and in the homogeneous case (ν = const) they are naturally interpreted as the squares of the decorrelation scales ρi in the directions of the respective eigenvectors of ν. In the inhomogeneous case, the decorrelation scales are defined locally in a similar manner, whereas the integration time T plays the role of a global scaling parameter for the distribution of ρ2i (x ). Therefore, setting the value of T is equivalent to specifying the square of the mean decorrelation scale ρ for a given distribution of ν(x ). Throughout the remainder of this paper, we keep in mind this equivalence and replace T with ρ2 where appropriate. Numerically, the action of the Gaussian BEC operator exp(T D) is usually represented by integrating (6) with an explicit time-stepping scheme, xt+δt = xt + δtDxt , such that the result of multiplication of a vector x0 by B is  n TD xT ≡ Bx0 = I + x0 ≈ exp[ρ2 D]x0 , (7) n where n = T /δt is the total number of time steps. Expression (7) shows that numerically, the Gaussian BEC model is a high-order polynomial in D. In data assimilation problems the n-step “time integration” (7) is embedded

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models

5

in the iterative loop that solves linear equations, whose system matrices are under the inversion signs in either (4) or (5). Therefore, reducing n (increasing δt) provides major computational savings. The minimum value for n is limited, however, by the stability condition which constrains eigenvalues of the operator I + ρ2 D/n in (7) not to exceed 1 in magnitude: n≥

1 2 ρ λ, 2

(8)

where λ is the absolute value of the largest eigenvalue of D. Numerically, the minimum value of n in 3d is proportional to the square of the largest ratio ρ˜ between decorrellation scale and the local grid step taken over the entire grid. In realistic applications, ρ˜ may easily exceed 10, substantially increasing the cost of computing the action of B on a vector. For ρ˜ > 10 the computational burden can be reduced by considering a spline BEC model  −m ρ2 D  exp[ρ2 D], (9) Bm = I − m which specifies the inverse BEC as a polynomial in the powers of −D and converges to the Gaussian model as m → ∞ (Fig. 1). The BEC operator in (9) can be implemented numerically in two ways, distinguished by the order of the operations of inversion and raising to the mth power. The first method requires m inversions of I − ρ2 D/m, and this approach can be interpreted as integration of the DE by an implicit scheme18 with the “time step” δt = ρ2 /m. The second method involves only one inversion of the matrix whose condition number is cm , where c = cond(I − ρ2 D/m). 1

10−3

m=∞ m = 10 m=3 m=2 m=1

correlation

10−4

10−5

10−6

100

m=∞ m = 10 m=3 m=2 m=1 NCODA

0.8

0.6

0.4

0.2

101

102

103

(a)

104

105

106

0 0

0.5

1

1.5

2

2.5

3

3.5

4

(b)

Fig. 1. Normalized spectra of the Gaussian (m = ∞) and spline BEC operators with different approximation order m (a) and the corresponding correlation functions (b). Horizontal axis is normalized by the correlation radius. The dashed line shows correlation function used in the experiments with NCODA Cd model (see Sec. 3.2).

July 13, 2012 9:9

6

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

Numerically, each iteration of the first inversion process is approximately equivalent to a one-step integration with the explicit scheme (7) as the corresponding matrices I+ρ2 D/n and I−ρ2 D/m differ only by the numerical factor before the diffusion operator. The second method of computing the action of Bm is more expensive than the first one, because the total number of iterations grows exponentially with m, unless an efficient preconditioner is available. On the other hand, the possibility to directly compute the action of m is advantageous in solving the MS 3dVar problem (2), B−1 m = (I − D/m) as it requires only one MS iterative cycle to invert (I − D/m)m + HT H. In ˜ m -preconditioned MS solution (5) contrast, the DS solution (4) and the B involve the product of two cycles: Each iteration of the respective DS/MS system solvers contains an MS iterative cycle required for computing the ˜ on a vector. action of B (or B) Spectral properties of the low-order spline models differ considerably from the Gaussian one: Their spectra exhibit more gentle slopes and weaker damping of the short (near-grid) scales (Fig. 1a) and the correlation functions decay faster than the Gaussian at small distances (Fig. 1b). The difference may affect the forecast skill of the assimilation system and not worth the computational gain when applied to real data. This and other related issues have been examined by means of numerical experimentation.

3. Numerical Experiments Setup Experiments were performed with the Relocatable Navy Coastal Ocean Model system (RNCOM) consisting of two primary components: The NCOM provides forecasts of the ocean state, and the Navy Coupled Ocean Data Assimilation (NCODA) uses a 3dVar algorithm to assimilate observations into the model forecast state.19

3.1. Numerical model and observations NCOM has a free-surface and is based on the primitive equations under the hydrostatic, Boussinesq, and incompressible approximations. The Mellor Yamada Level 2/2.5 turbulence models are used to parameterize vertical mixing. Most terms are treated explicitly in time, except for the propagation of surface waves and vertical diffusion, which are treated implicitly. For the present study, the model was configured on two grids with homogeneous grid spacing of 3 km and 10 km in the horizontal. In the vertical, there

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models °

0

°

120 E

4

2

0

4

8 2 4

6

2

2

0

2

02

2

4

4

0

2

0 2

4

6

2

2

2

2

4

2

4

2

6

0

2

2

246

2

2

4

0

0

0

18° N

2

4

2406

0

2

0

0

2

4

2

0

4

2

6

4

8

20° N

2

20 18 1620 14 181

2

2

2

22° N 0

2

2 4

4

4

2 4

12

4

6

2

0

2

2

2

6

8

2

0

2

4

0

6

20

0

2

2

0

0

2

0

18 1602 14 118 0

20

0

4

2

2

2

°

24 N

2

18° N

132 E

0

2

2

4

2

4

2

4

2 4

20° N

6 8 24 42 0

2

4

2

6

66

°

130 E 2

4

0

°

26 N

0

6 4

2 0

°

128 E 4

28 N

0

4

6 0 2

2

2 4 6

0 2

0

0 22° N

8106

68 2

4

°

24 N

2

68

4

0

8

02

6 2 4 20

°

126 E

4

°

2

2

2

6 4

4

°

26 N

°

124 E

4

6 0

°

122 E

6

2

°

118 E

0

0

°

132 E

2

8

°

28 N

°

130 E

4

2

°

128 E

4 2

°

126 E

4

2

°

124 E

2

°

122 E

2

°

120 E

2

°

118 E

7

0

4

(a)

(b)

Fig. 2. The model domain and sea surface temperature increments at 10 m on September 2, 2007 for the Cd (a) and C2∗ covariance models (b) (see Sec. 3.2). Contour interval is 0.1◦ .

were respectively 46 and 50 layers having grid steps varying between 1 m and 400 m. The number of grid points representing a 3d scalar field was M = 10,766,576 and M = 862,992 respectively. All the runs were conducted on the Dell R610 server equipped with 16 Xeon 5500 processors running at 2.8 GHz. Assimilation experiments were performed in the Okinawa Trough region (Fig. 2) in the time period from September 1 to October 31, 2007. The region and time period were selected to include extensive Navy observations from an air-deployed bathythermograph survey, a shipboard hydrographic survey, and eight gliders. Observations from this Navy exercise are an addition to the standard operational data stream used by NCODA, which consists of sea surface temperature (SST) and sea surface height anomalies obtained from satellites, and temperature/salinity profiles acquired by buoys, floats, CTDs, and XBTs. The total number of observations processed during the ¯ = 17,507 2-month assimilation period was 1,050,429, or approximately N points per 24-hour assimilation cycle.

3.2. Assimilation system NCODA uses a DS 3dVar data assimilation scheme with the analysis Eq. (4). The vector of analysis variables x contains temperature, salinity, geopotential (dynamic height) and velocity fields, but in contrast to the DE

July 13, 2012 9:9

8

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

approach, the BEC operator is defined by the explicit specification of its matrix elements via correlations in 3d using the correlation function shown by the dashed line in Fig. 1b. In the following, we will denote this BEC model by Cd , the Gaussian model and its mth-order spline approximations will be labeled by C∞ and Cm , and the asterisk will denote MS implementation (3) ∗ ). of the spline model (Cm In the assimilation experiments, the Cd model was replaced by the tested BEC models. Several assimilation runs with Cd and other BEC models were also performed over the 2-month assimilation period for comparison purposes. In these runs, the horizontal decorrelation scale was set to 45 km, while vertical scale varied in z in proportion with the vertical model grid step. Since the major goal of the present study is to compare computational efficiencies of the BEC models that are quite different, their forecast skill was monitored with respect to the operation of the NCODA system with the Cd BEC model whose forecast skill was used as a benchmark. This was done to ensure that the computational cost of the analysis was not reduced at the expense of reduction in assimilation quality q. The latter was estimated as the DS distance between the 24-hour model temperature/salinity forecast at observation points Tf , Sf and the observed values To , So : qT (t) = (Tf − To )T σT−2 (Tf − To )1/2 ,

(10)

qS (t) = (Sf − So )T σS−2 (Sf − So )1/2 ,

(11)

where σT,S are the observation errors and the angular brackets denote averaging over the observational locations. These DS distances were normalized to measure the forecast skill s of the tested models relative to the skill of the benchmark model Cd . s(t) =

qT (C) + qS (C) . qT (Cd ) + qS (Cd )

(12)

4. Results 4.1. Comparison of the forecast skills As it has been noted in Sec. 2, spline models are characterized by broader spectra and provide less attenuation at high spatial frequencies (Fig. 1) than the Gaussian model. This property causes a certain difference in the analyses increments (Fig. 2), which may result in substantial decrease of the

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models

9

s(t)

1.1 1 0.9 0.8 0

1.03 1.01

10

20

30 t, days

40

50

60

s(t)

1.1 1 0.9 0.8 0

1.02 1.01

10

20

30 t, days

40

50

60

Fig. 3. Forecast skill of the C2 (black) and C∞ (gray) BEC models implemented on coarse (10 km, above) and fine (3 km, below) resolution grids. The time-averaged skills s are shown in the low right corners.

overall forecast skill. The forecast skills for the 10 km and 3 km resolution configurations are shown in Fig. 3. It is seen that the forecast skill of both C∞ and C2 BEC models does not depend on the minor changes in the shape of the correlation function: The 2-month mean values shown in Fig. 3 do not differ significantly from 1. This result indicates that the analyses from the C∞ and C2 models are at least not degrading the benchmark forecast generated by the operational NCODA system, while both models demonstrate similar forecast skills. It is also remarkable that the forecast skill of the fine-resolution models appeared to be 13–15% below the skill of the respective coarse-resolution configurations. To some extent this phenomenon can be explained by the presence of small-scale motions in the 3 km configuration that are barely constrained by the available observations: On average, an observation supplies information for 610 grid points in the fine-resolution case against 110 grid points per observation for the coarse-resolution configuration.

4.2. Comparison of the CPU times The dependence of CPU time was explored on both the ratio ρ of the background decorrelation scale to the grid step and on the degree of

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

10

anisotropy of the correlations. A series of experiments were performed with different strengths of the anisotropy and different values of ρ for the selected date September 2, 2007 (23,970 observation points). In these experiments, the diffusion tensor was specified as follows. The background decorrelation scales ρi at every location were defined as a product of the local grid steps and the universal scaling parameter ρ. The smaller principal axis in the horizontal direction (corresponding to λ2 ) was set to be orthogonal to the local velocity vector  v . The length of the larger axis λ1 was set to be equal to λ2 · max(1, |v |/v), where v is a prescribed threshold value of |v |. A structure like this simulates enhanced diffusive transport of model errors in the regions of strong currents on the background of isotropic error diffusion (Fig. 4). The strength of anisotropy was controlled by changing the value of v: v = 10 m/s corresponds to locally isotropic diffusion (λ = 12, Fig. 4a), v = 0.2 m/s imposed moderately anisotropic covariances (λ = 50) in regions of strong currents, and v = 0.07 m/s corresponds to the strongly anisotropic case (λ = 165, Fig. 4b). In a series of experiments, NCODA observations on September 2, 2007 were analyzed using the C∞ and C1,2,3 BEC models in both state- and data-space formulations and the required CPU times for these analyses

60

°

70 18 N 80 80

0.2

0.4

0.6

30

50

40

40

50

6700

70

70

5 60 0

40

5 60 0

50

4

50

50

60 430020

30

3.5

50

70

3

60

50

°

60

60

50

(a)

2.5

70

0 60 7

20° N

40

6 70 0

80

50

2

80

50

70

1.5

40

60

6 70 0

80

70

40

50

60

70

40

22° N 40

60

50

1

2010 30

0 10

0

60

50

60

0.5

80

24° N

50

70

0 60 7

80

60

70

70 18 N 80 80

60

60

50

°

90

60

60

3400

50

40

70

20° N

°

130 E 132 E 20 3400500 80 6

10

20

70

2100

40

60 430020

70

26° N 80

60

50

°

128 E

50

28° N

50

6700

50

0

70

40

22° N

0 10

50

2010 30

50

24° N

2100

40

3400

50

26° N

°

126 E

90

°

70

°

122 E 124 E 30 40

20

70

60

°

120 E

30

50 30

60

28° N

°

118 E 90

70

°

2 100 0

°

130 E 132 E 20 3400500 80 6

10

60

°

128 E

70

°

126 E

90

°

70

°

122 E 124 E 30 40

70

°

120 E

2 100 0

°

118 E

70

0.8

1

1.2

1.4

1.6

1.8

2

(b)

Fig. 4. Composite maps of four columns (filled contours) of the temperature correlation operator for (a) isotropic (λ = 12) and (b) anisotropic (λ = 165) cases of the C∞ BEC model at 20 m. Thin contours show the SSH field. Contour interval is 10 cm.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models

11

Table 1. CPU times τ∞ for the Gaussian covariance model and the relative CPU times ∗ ). The accuracy of the mth-order spline models implemented in DS (τm ) and in MS (τm of system solutions (defined as the ratio between the norms of the residual and the rhs vectors) is ε = 10−6 . The fastest cases for a given m are boldfaced. τ∞ , min ρ

λ

C∞

∗ /τ τm ∞

τm /τ∞ m=1

m=2

m=3

m=1

m=2

m=3

3.0

12 50 165

3.47 14.9 54.0

0.83 0.38 0.16

2.00 0.85 0.32

2.66 1.10 0.40

0.06 0.02 0.01

0.31 0.21 0.15

1.36 1.82 1.78

4.5

12 50 165

8.72 36.4 156

0.54 0.19 0.09

1.25 0.41 0.21

1.54 0.74 0.29

0.04 0.01 0.00

0.23 0.15 0.09

1.58 1.62 1.83

7.0

12 50 165

24.0 129 418

0.24 0.08 0.04

0.62 0.19 0.11

1.06 0.27 0.16

0.02 0.00 0.00

0.16 0.07 0.06

1.42 0.84 6.68

were compared. Results of the 10 km grid size experiments are assembled in Table 1 where larger anisotropy corresponds to the larger maximum eigenvalues λ of the diffusion operator (column 2). Respectively, the tested values of ρ correspond to the decorrelation scales of 30, 45, and 70 km. As can be seen from Table 1, the numbers indicate improved computational efficiency of the low-order (m < 3) MS implementation of the spline model. For higher-order models, the MS solution appears to be less efficient due to exponential growth of the condition number of the system matrix. Additional experiments were made on a longer time scale, with the NCODA system using the generic correlation Cd model as well as the tested models with 24-hour analysis cycle (λ = 12, ρ = 4.5). These experiments have shown that the C2∗ model is 3 times faster than C2 and 3.5 times faster than C∞ for the 10 km configuration. Similarly, for the 3 km configuration, the C2∗ model was 3.3/4.2 times faster. CPU times of the C2∗ and Cd models are compared in Fig. 5. On average, the C2∗ model requires 30–50% more CPU time than the generic Cd model. However, when the number of observations exceeds 1.5–1.7 ·104 , the C2∗ model appears to be more efficient. Similar computations for 3 km resolution show that this critical number of observations increases only slightly to 1.8–2 ·104 despite a 12-fold increase in the dimension of the MS.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

M. Yaremchuk et al.

12

2 1.8 1.6

CPU ratio

1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

0.5

1

1.5

2

2.5

number of observations, × 10−4 Fig. 5. CPU time ratios between the Cd and C2∗ models against the number of observations for the 10 km configuration. Points above the horizontal line correspond to the cases of the faster C2∗ model.

5. Conclusions The forecast skill and computational efficiency of the Gaussian and spline covariance models were examined in the framework of 3dVar assimilation of real data into an operational ocean model. It is shown that the MS formulation of the second-order spline model has similar 24-hour forecast skill and 3–5 times better computational efficiency than the DS implementation of the Gaussian and spline models. At m < 3, the ∗ solutions is based on the low-cost computational efficiency of the Cm computation of the action of the inverse BEC operator B−1 = (I−ρ2 D/m)m which contributes to the system matrix of the normal equation (2). On the contrary, multiplication by I−ρ2 D/m (or by I+ρ2 D/n) has to be performed many times in the DS formulation to iteratively model the action of B, which in turn is immersed into the iterative loop required to find the solution (4) to the normal system in the DS formulation. It is also shown, that the difference in the BEC models has a negligible impact on the forecast skill of the 3dVar assimilation system. Comparison with the benchmark NCODA 3dVar algorithm has shown that the forecast skill remains virtually the same (Fig. 3), whereas the C2∗ model appears to be more efficient computationally than the operational BEC model when the number of observations exceeds 15–12·103. Numerical experiments have

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch01

Forecast Skill and Computational Cost of the Correlation Models

13

also shown that spline models become especially advantageous when the background decorrelation scale is well resolved by the model grid (ρ > 3) and the diffusion operator is strongly anisotropic/inhomogeneous (Table 1). The results of this work suggest that studying the applicability of the anisotropic higher-order spline BEC models to 3dVar assimilation is worth consideration for at least three reasons: (1) they are computationally efficient in processing large number of observations, (2) they are flexible enough to accommodate covariance information from the structure of the background flow, and (3) they can be easily extended to include modelgenerated covariances extracted from model statistics. Acknowledgments This study was supported by the Office of Naval Research (Program element 0602435N, project ECoVarDA’).

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

G. Gaspari and S. Cohn, Q. J. R. Met. Soc. 125 (1999) 723. G. Gaspari, et al., Q. J. R. Met. Soc. 132 (2006) 1815. J. Verron, et al., J. Geophys. Res. 104(C3) (1999) 5441. A. Deckmyn and L. Berre, Mon. Wea. Rev. 133 (2005) 1279. O. Pannekoucke, Mon. Wea. Rev, 137(9) (2009) 2995. Q. Xu, Adv. Atm. Sci., 22(2) (2005) 181. M. Yaremchuk and S. Smith, Q. J. R. Met. Soc. 137, in press, (2011). J. Derber and A. Rosati, J. Phys. Oceanogr. 19 (1989) 1333. R. J. Purser, et al., Mon. Wea. Rev. 131 (2003) 1536. A. Weaver and P. Courtier, Q. J. R. Met. Soc. 127 (2001) 1815. E. Di Lorenzo, et al., Ocean Modelling, 16 (2007) 160. O. Pannekoucke and S. Massart, Q. J. R. Met. Soc. 134 (2008) 1425. M. Yaremchuk and A. Carrier, Mon. Wea. Rev. 139, in press, (2011). H. Ngodock, et al., Mon. Wea. Rev. 128 (2000) 1757. A. C. Lorenc, Q. J. R. Met. Soc. 112 (1986) 1177. P. Coutier, et al., Q. J. R. Met. Soc. 124 (1998) 1783. A. T. Weaver, et al., Mon. Wea. Rev. 131 (2003) 1360. Mirouze and Weaver, Q. J. R. Met. Soc. 136 (2010) 1421. E. Coelho, et al., J. Marine Sys. 78 (2009) 5272.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-ch01

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch02

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


REAL-TIME TROPICAL CYCLONE PREDICTION USING COAMPS-TC J. D. DOYLE∗ , Y. JIN∗ , R. M. HODUR† , S. CHEN∗ , H. JIN∗ , J. MOSKAITIS∗ , A. REINECKE∗ , P. BLACK† , J. CUMMINGS‡ , E. HENDRICKS∗ , T. HOLT∗ , C.-S. LIOU∗ , M. PENG∗ , C. REYNOLDS∗ , K. SASHEGYI∗ , J. SCHMIDT∗ and S. WANG∗ ∗Naval Research Laboratory, Monterey, CA, USA †SAIC, Monterey, CA, USA ‡Naval Research Laboratory, Stennis, MS, USA

A new version of the Coupled Ocean/Atmosphere Mesoscale Prediction System TM

for Tropical Cyclones (COAMPS-TC ) has been developed specifically for forecasting tropical cyclone track, structure, and intensity. The COAMPS-TC has been tested in real-time in both coupled and uncoupled modes over the past several tropical cyclone seasons in the Pacific and Atlantic basins at a horizontal grid spacing of 5 km. An evaluation of a large sample of real-time forecasts for the 2010 and 2011 seasons in the Atlantic basin reveals that the COAMPS-TC predictions have smaller intensity errors than other real-time dynamical models for forecasts beyond the 30 h time. Real-time forecasts for Hurricane Irene (2011) illustrate the capability of the model to capture both the intensity and the fine-scale features (e.g., eyewall, rainbands), in agreement with observations. The results of this research highlight the promise of highresolution deterministic and ensemble-based approaches for tropical cyclone prediction using COAMPS-TC.

1. Introduction A dramatic scenario played out during August 2011 as Hurricane Irene threatened many communities along the U.S. Eastern Seaboard, from Florida to New England. Basic questions such as where Irene would track and how strong it would become had profound implications for the millions of people in its path and billions of dollars in property that were vulnerable. The potential impact of tropical cyclones on military operations can also be enormous. An extreme example is the infamous Typhoon Cobra, also known as Halsey’s Typhoon after Admiral William Halsey, which struck 15

July 13, 2012 9:9

16

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

the U.S. Navy’s Pacific Fleet in December 1944 during World War II. Three destroyers were lost, and a total of 790 sailors perished. More recently during Irene, the decision to sortie Navy assets from Norfolk, VA and other ports along the Eastern Seaboard many days in advance of the storm was critically dependent on forecasts of Irene’s track, intensity (maximum sustained wind speed at the surface), and storm structure (such as the size of the storm or radius of key wind speed thresholds). In the N.W. Pacific, recently the Navy Pacific Fleet in the Philippine Sea was impacted by Typhoon Nanmadol, which exhibited erratic movement and was poorly forecasted. The demand for more accurate tropical cyclone forecasts with longer lead times is greater than ever due to the enormous economic and societal impact. There has been steady and methodical improvement of tropical cyclone (TC) track prediction; a three-day tropical cyclone track forecast today is as skillful as a one-day forecast was just 30 years ago. However, there has been almost no progress in improving TC intensity and structure forecasts due to a variety of reasons ranging from a lack of critical observations under high wind conditions and in the TC environment to inaccurate representations of TC physical processes in numerical weather prediction (NWP) models. Advances in high-resolution TC modeling and data assimilation are thought to be necessary to significantly improve the performance of intensity and structure prediction. To this end, the Naval Research Laboratory (NRL) in Monterey, CA, has developed the Coupled Ocean/Atmosphere Mesoscale Prediction System for Tropical Cyclones (COAMPS-TC), a new version of COAMPS designed specifically for highresolution tropical cyclone prediction.

2. COAMPS-TC Description The COAMPS-TC system comprises data quality control, analysis, initialization, and forecast model sub-components. The Navy Variational Data Assimilation System (NAVDAS1 ) is used to blend observations of winds, temperature, moisture, and pressure from a plethora of sources such as radiosondes, pilot balloons, satellites, surface measurements, ships, buoys, and aircraft with a model first guess field to produce the analysis. Enhancements to the NAVDAS system2 for COAMPS-TC include the addition of synthetic observations that define the TC structure and intensity based on the TC reports in real-time from the National Hurricane Center (NHC) and the Joint Typhoon Warning Center (JTWC). Also, as part of the TC analysis procedure, the pre-existing circulation in the COAMPS-TC

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Real-time Tropical Cyclone Prediction Using COAMPS-TC

b1387-ch02

17

first guess fields is relocated to allow for an accurate representation of the TC position during the analysis. Following this step, the analyzed fields are initialized to reduce the generation of spurious, high-frequency atmospheric gravity waves. The sea surface temperature is analyzed directly on the model computational grid using the Navy Coupled Ocean Data Assimilation (NCODA3 ) system, which makes use of all available satellite, ship, float, and buoy observations. Both the NCODA and NAVDAS systems are applied using a data assimilation cycle in which the first guess for the analysis is derived from the previous short-term forecast. The COAMPS-TC atmospheric model uses the non-hydrostatic and compressible form of the dynamics and has prognostic variables for the three components of the wind (two horizontal wind components and the vertical wind), the perturbation pressure, potential temperature, water vapor, cloud droplets, raindrops, ice crystals, snowflakes, graupel, and turbulent kinetic energy.4 Physical parameterizations include representations of cloud microphysical processes,5 convection,6 radiation,7 boundary layer processes,4 and surface layer fluxes.8,9 The COAMPS-TC model contains a representation of dissipative heating near the ocean surface, which has been found to be important for tropical cyclone intensity forecasts.10 The model also contains an optional advective scheme for scalars that preserves monotonicity and positive definiteness. The COAMPS-TC system uses a flexible nesting design that has proven useful when more than one storm is present in a basin at a given time as well as special options for moving nested grid families that independently follow individual tropical cyclones. The COAMPS-TC system has the capability to operate in a fully coupled air–sea interaction mode.11 The atmospheric module within COAMPS-TC is coupled to the NRL-developed Navy Coastal Ocean Model (NCOM12 ) to represent air–sea interaction processes. The COAMPS-TC system has an option to predict ocean surface waves, and the interactions between the atmosphere, ocean circulation, and waves using the Simulating WAves Nearshore (SWAN) model. A sea spray parameterization can be used to represent the injection of droplets into the atmospheric boundary layer due to ocean surface wave breaking and shearing.

3. Real-time Tropical Cyclone Forecasts COAMPS-TC has been tested in real-time in both coupled and uncoupled modes over the past several tropical cyclone seasons in the Pacific and

July 13, 2012 9:9

18

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

Atlantic basins. These real-time tests have been conducted in conjunction with the National Oceanic and Atmospheric Administration (NOAA) sponsored Hurricane Forecast Improvement Project (HFIP), which is focused on accelerating the improvement in hurricane intensity forecasts. In these real-time applications, the atmospheric portion of the COAMPSTC system makes use of horizontally nested grids with grid spacing of 45, 15, and 5 km. The 15- and 5-km grid-spacing meshes track the TC center, which enables the TC convection to be more realistically represented on the finest mesh in an efficient manner. The forecasts make use of the Navy or the NOAA global model for lateral boundary conditions. The model is typically run four times daily for the W. Atlantic, E. Pacific, and W. Pacific regions and is triggered by the NHC and JTWC warning message (which contains observational estimates of the storm position and intensity) when a storm reaches a 30 knot intensity. The forecasts are routinely disseminated in real-time to NHC, JTWC, and HFIP researchers. The forecast graphics are also available in real-time at http:// www.nrlmry.navy.mil/coamps-web/web/tc. 3.1. COAMPS-TC atmospheric model forecasts Real-time COAMPS-TC forecasts have been conducted using U.S. Department of Defense High Performance Computing (HPC) platforms over the past several years. An example of the intensity forecast performance of COAMPS-TC for a large number of cases (more than 450 cases at the 24 h forecast time) in the W. Atlantic region for the 2010 and 2011 seasons is shown in Fig. 1 (for a homogeneous statistical sample). The COAMPS-TC model had the lowest intensity error of any dynamical model for the 36–120-h forecast times, which is an important period for forecasters and decision makers. Other numerical models included in this analysis are operational models run by NOAA (HWRF, GFDL), and the Navy’s current operational limited area model (GFDN). This promising performance is a result of a large effort devoted to developing and improving COAMPS-TC over the past several years. Key aspects of the COAMPS-TC system that have contributed most to the forecast intensity skill improvement (as deduced from sensitivity tests) include: (i) new synthetic observations and data assimilation system modifications; (ii) modified turbulence parametrization (in particular the mixing length) in and above the boundary layer; (iii) surface drag, heat, and moisture exchange10 consistent with recent field campaigns; and (iv) fidelity of the microphysical parametrization.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch02

Real-time Tropical Cyclone Prediction Using COAMPS-TC

19

Intensity error, NHC criteria 25

COAMPS-TC HWRF GFDN GFDL

MAE (kt)

20

15

10

5

Sample size

0

0 611

12

532

24

36

48 72 Lead time (h)

453

395

336

241

96

120

182

125

Lead time (h)

Fig. 1. Wind speed mean absolute error (MAE) (knots; 1 knot = 0.514 m s−1 ) as a function of forecast time for the 2010 and 2011 seasons in the Atlantic basin for a homogeneous statistical sample. The numerical models included in this analysis are the Navy’s COAMPS-TC, operational models run by NOAA (HWRF, GFDL), and the Navy’s current operational limited area model (GFDN). The number of cases is shown at the bottom. The intensity errors are computed relative to the best track.

An example of a real-time COAMPS-TC forecast for the recent Hurricane Irene is shown in Fig. 2. The composite National Weather Service radar reflectivity is shown on the left panel near the time of landfall in North Carolina at 1148 UTC 27 August 2011 and the COAMPS-TC predicted radar reflectivity at 36 h valid at 1200 UTC is shown on the right panel. The COAMPS-TC forecast shown in Fig. 2 is for the model second grid mesh (15 km horizontal resolution). The model prediction was accurate in the track (skill similar to other dynamical models), eventual landfall location (Fig. 2), storm intensity (Fig. 4), as well as the structure and size (from a qualitative perspective), an especially important characteristic of this particular storm in such close proximity to the U.S. East Coast. One noteworthy aspect of Irene was its large size, with tropical storm

July 13, 2012 9:9

20

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

Fig. 2. The NWS composite radar reflectivity (from NOAA) valid at 1148 UTC 27 August 2011 (left) (source NOAA) and the COAMPS-TC 36-h forecast radar reflectivity performed in real time and valid at 1200 UTC 27 August (right) for Hurricane Irene. The COAMPS-TC reflectivity is shown for the second grid mesh, which has a horizontal resolution of 15 km.

force winds (17.5 M s−1 or 34 knots) extending radially outward from the eye for nearly 300 km. The large size of Irene is also apparent in the observed radar reflectivity in Fig. 2. The COAMPS-TC prediction captures the large areal extent of the precipitation field, as well as its asymmetry about the TC center (most of the precipitation is north and east of the center). This large shield of heavy precipitation caused severe river flooding as it slowly moved north through the mid-Atlantic and Northeast U.S. The simulated radar reflectivity for the COAMPS-TC grid mesh 3 (5 km horizontal resolution), shown in Fig. 3, illustrates the capability of the model to capture the finer-scale features, such as the eyewall and rainbands, in generally good agreement with the observed reflectivity. Sensitivity tests (not shown) indicate that the synthetic observations and TC-related data assimilation system modifications, along with the in-cloud turbulent mixing representation, are important for proper simulation of Irene’s structure. It should also be noted that COAMPS-TC makes use of a 1.5 order closure turbulence parametrization that predicts turbulent kinetic energy (TKE) and includes advection of TKE. We hypothesize that the advection of TKE may be important for the proper representation of the turbulence evolution within the hurricane boundary layer beneath and near the eyewall. Overall, the Navy’s COAMPS-TC real-time intensity predictions of Hurricane Irene were more skillful than the other leading operational

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Real-time Tropical Cyclone Prediction Using COAMPS-TC

b1387-ch02

21

Fig. 3. The COAMPS-TC 36-h forecast radar reflectivity valid at 1200 UTC 27 August for the third grid mesh, which has a horizontal resolution of 5 km.

governmental forecast models, as shown in Fig. 4. All of the available models except for COAMPS-TC had a tendency to over-intensify Irene, often by a full Saffir-Simpson category or more. These real-time COAMPSTC forecasts were used by forecasters at the National Hurricane Center as part of an experimental HFIP multi-model ensemble. The COAMPSTC consistently provided accurate real-time intensity forecasts during the period 23–28 August 2011, when critical decisions were made by forecasters and emergency managers including evacuations. 3.2. COAMPS-TC ensemble forecasts While research is ongoing to improve deterministic atmospheric forecasts through advancements to the forecast model and more accurate estimates of the initial state, simultaneously there has been interest in obtaining probabilistic information derived from ensemble forecasts. An ensemble

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

22

Intensity error, NHC criteria COAMPS-TC HWRF GFDN GFDL

40

Mean absolute error (kt)

35 30 25 20 15 10 5

Sample size

0

0

12

24

36

48 72 Lead time (h)

23

21

19

17

15

11

96

7

120

3

Lead time (h)

Fig. 4. Wind speed MAE (knots) as a function of forecast time for Hurricane Irene for a homogeneous statistical sample. The numerical models included in this analysis are the Navy’s COAMPS-TC, operational models run by NOAA (HWRF, GFDL), and the Navy’s current operational limited area model (GFDN). The number of cases is shown at the bottom. Only forecasts after Irene has moved away from Hispaniola are shown here. The intensity errors are computed relative to the best track.

of forecasts from equally plausible initial states and model formulations offer a computationally feasible way of addressing inevitable forecast uncertainties, offering improved forecasts through ensemble statistics such as mean quantities, as well as quantitative estimates of forecast error and variance. Although the concept of ensemble modeling is relatively simple, the performance of an ensemble forecast system is very sensitive to the basic ensemble architecture. At the Naval Research Laboratory, we are designing new ensemble methods for both the global and mesoscale atmospheric forecast systems. Because of the high computational demands associated with ensemble development and verification (especially when one is interested in severe or rare events), exceptional computational resources are necessary to perform this research.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Real-time Tropical Cyclone Prediction Using COAMPS-TC

b1387-ch02

23

A new COAMPS-TC ensemble system that is capable of providing probabilistic forecasts of TC track, intensity, and structure has been developed by scientists at NRL in Monterey, CA. This system makes use of the community-based Data Assimilation Research Testbed (DART13 ) developed at the National Center for Atmospheric Research, which includes various options for Ensemble Kalman Filter (EnKF) data assimilation. The COAMPS-TC DART system constitutes a next generation data assimilation system for tropical cyclones that uses flow dependent statistics from the ensemble to assimilate observational information on the mesoscale. A real-time COAMPS-TC ensemble system was run in a demonstration mode in 2011 for the W. Atlantic, E. Pacific, and W. Pacific regions. The system was comprised of an 80-member COAMPS-TC cycling data assimilation ensemble on three nested grids with horizontal spacing of 45, 15, and 5 km. For each TC initiated by either NHC or JTWC, the COAMPS-TC ensemble was initialized by interpolating global forecast fields from the 80-member GFS-EnKF system14 to the three nested grids, which were centered on the storm. Six-hour forecasts were made four times daily to provide background estimates for the assimilation of observations from surface stations, ship data, radiosonde ascents, cloud-track wind retrievals, and aircraft data using the DART EnKF. In addition to these conventional datasets, the NHC and JTWC TC position estimates were directly assimilated with the EnKF system. Under-sampling issues in the data assimilation procedure associated with the finite ensemble size were controlled by limiting the spatial influence of observations with a static localization radius of 1,000 km, as well as applying a spatially and temporally varying inflation factor15 to the prior ensemble perturbations. For effective usage of the high-resolution capability of COAMPS-TC, a two-way interactive data assimilation procedure was implemented. In this algorithm, the innovations were defined using the highest resolution nest that contained the observation. These innovations were used to update the COAMPS-TC fields on all three grids. Furthermore, observations contained outside of a nest were allowed to update the fields within the nest. Ten-member forecasts were performed twice daily to five days using the same three nested grid configuration as the data assimilation ensemble. The first 10 members of the data assimilation ensemble were used to define the forecast ensemble. Lateral boundary conditions for the forecast ensemble were drawn from the GFS-EnKF ensemble forecast. Examples of probabilistic products for Hurricane Irene are shown in Fig. 5 for both track (top panel) and intensity (bottom panel). This is a real-time

July 13, 2012 9:9

24

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

Fig. 5. Probabilistic products from the COAMPS-TC ensemble for Hurricane Irene corresponding to the track (top panel) and intensity (bottom panel). This real-time forecast was initialized at 1200 UTC 23 August, which is approximately four days prior to landfall. The probabilistic track product shows the TC position from the individual ensemble members every 24 h and ellipses that encompass 1/3 and 2/3 of the ensemble members. The intensity (knots) distribution is shown as a function of forecast lead time (hours) with the minimum value, maximum value and various quantiles of the ensemble distribution shown as denoted by the legend.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Real-time Tropical Cyclone Prediction Using COAMPS-TC

b1387-ch02

25

forecast initialized at 1200 UTC 23 August, which is four days prior to landfall. The probabilistic track product shows the TC position from the individual ensemble members every 24 h and ellipses that encompass 1/3 and 2/3 of the ensemble member forecast positions. Note that the observed landfall location of the eye (see Fig. 2) was within the ensemble distribution, although the ensemble mean landfall was approximately 12 h later than observed. The probabilistic intensity product (lower panel) shows a considerable spread among the members, particularly beyond 84 h, just prior to landfall. These products can be extremely valuable to assess the uncertainty in both track and intensity forecasts, and NRL is currently developing these capabilities and products further.

3.3. Coupled COAMPS-TC forecasts The COAMPS-TC system was run in a fully coupled mode, interactive with NCOM, during the Office of Naval Research sponsored Interaction of Typhoon and Ocean Project (ITOP) during the summer and fall of 2010. An example of a fully coupled COAMPS-TC forecast for Typhoon Fanapi performed in real-time is shown in Fig. 6. The NCOM ocean model was applied using a 10-km horizontal grid increment in this example, and the finest mesh for the atmospheric model used a 5-km grid increment. The atmosphere and ocean fluxes were exchanged every ocean model time step. The COAMPS-TC predicted track (red) from a 90-h real time forecast valid at 0600 UTC 19 September 2010 is quite close to the observed or best track (black). The sea surface temperature, shown in color shading, indicates significant cooling was predicted by COAMPS-TC during the passage of Fanapi due to enhanced mixing by the strong near-surface winds of the typhoon. The predicted cooling of the sea surface temperatures of 2–4◦C is in agreement with estimates from in situ and remote sensing observations in this region. The negative impact of the ocean cooling underneath the tropical cyclone can reduce the TC intensity and broaden the tropical cyclone secondary circulation, which underscores the importance of properly representing these air–sea interaction process.4,11 A joint Navy/Air Force Hurricane Hunter program was in its demonstration phase in 2011 with Airborne Expendable Bathythermographs (AXBTs) being deployed from WC-130J hurricane reconnaissance aircraft in order to improve the initialization and validation of coupled models such as COAMPS-TC. Over 30 AXBTs were deployed from Air Force Hurricane Hunter aircraft in Irene as it approached landfall on Cape Hatteras.

July 13, 2012 9:9

26

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al.

Fig. 6. The COAMPS-TC predicted track (red) for Typhoon Fanapi from a 90-h realtime forecast valid at 0600 UTC 19 September 2010 and the observed best track (black). The sea surface temperature (◦ C) is shown in color shading and indicates significant cooling during the passage of Fanapi. The surface currents are shown by the white vectors.

Assimilation studies are currently being carried to assess the impact of these AXBTs on the coupled model skill. These new observations supplement previous studies4,11 of hurricane-induced cooling and may further help document the existence of ocean mixing or advective processes (and cooling of the sea surface) along the storm track and in coastal regions that may have slowed the intensification of Irene.

4. Summary The prediction of the tropical cyclone track, and even more so, the tropical cyclone intensity, remain among the greatest challenges facing meteorology today. The results of this research highlight the promise of high-resolution deterministic and ensemble-based approaches for tropical cyclone prediction

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Real-time Tropical Cyclone Prediction Using COAMPS-TC

b1387-ch02

27

using models such as COAMPS-TC. During the past several tropical cyclone seasons in the Pacific and Atlantic basins, COAMPS-TC has been tested in real-time in both coupled and uncoupled modes at a horizontal resolution of 5 km. An evaluation of a large sample of real-time forecasts for the 2010 and 2011 seasons in the Atlantic basin reveals that the COAMPS-TC predictions have smaller intensity errors than any other real-time dynamical model beyond the 30 h forecast time. Recent real-time forecasts of Hurricane Irene (2011) illustrate the capability of COAMPS-TC to capture both the intensity and the fine-scale features (e.g., eyewall, rainbands), in agreement with observations. Real-time forecasts of Typhoon Fanapi in the W. Pacific performed in support of the ITOP experiment in 2010 accurately predicted not only the track and intensification, but also captured the sea surface cooling induced by the mixing and upwelling in agreement with satellite measurements. While real-time COAMPS-TC has accurately predicted the evolution of Irene, Fanapi, as well as other tropical cyclones (not shown), there are a number of examples that were not predicted as well. These storms, and the data collected during their life cycle, provide opportunities to study and obtain a greater appreciation of the complex physics and interactions that occur in these systems, and to use this information to further improve the COAMPS-TC modeling system. This research will lead to new capabilities in the form of mesoscale TC ensemble forecasts, providing the Navy with probabilistic forecasts of tropical cyclone intensity and structure for the first time. It is also expected that this research will help motivate new field campaigns, which focus on the key measurements needed to further advance our understanding of the convective structure and dynamics of these systems, as well as provide forecast validation. The flexibility of the COAMPS-TC design will also allow us to test more advanced physics and numerics in an effort to gain a better physical understanding of the model’s intensity forecast skill.

Acknowledgments We acknowledge the support of the Office of Naval Research’s (ONR) Program Element (PE) 0602435N and PMW-120 PE 0603207N, as well as the National Oceanic and Atmospheric Administration (NOAA) sponsored Hurricane Forecast Improvement Project (HFIP). We also appreciate support for computational resources through a grant of Department of Defense High Performance Computing time from the DoD Supercomputing

July 13, 2012 9:9

28

AOGS - AS and OS

9in x 6in

b1387-ch02

J. D. Doyle et al. TM

Resource Center at Stennis, MS and Vicksburg, MS. COAMPS-TC trademark of the Naval Research Laboratory.

is a

References 1. R. Daley and E. Barker, Mon. Wea. Rev. 129 (2001) 869. 2. C.-S. Liou and K. D. Sashegyi, Natural Hazards, http://www.springerlink. com/content/a840nrr1572w1387/ (2011). 3. J. A. Cummings, Quart. J. Roy. Meteor. Soc. 131 (2005) 3583. 4. R. M. Hodur, Mon. Wea. Rev. 125 (1997) 1414. 5. S. A. Rutledge and P.V. Hobbs, J. Atmos. Sci. 40 (1983) 1185. 6. J. S. Kain and J.M. Fritsch, AMS Meteor. Monogr. 46 (1993) 165. 7. Q. Fu and K.-N. Liou, J. Atmos. Sci. 50 (1993) 2008. 8. J. F. Louis, Bound.-Layer Meteor. 17 (1979) 187. 9. C. Fairall, E. Bradley, J. Hare, A. Grachev and J. Edson, J. Climate, 16 (2003) 571. 10. Y. Jin, W. T. Thompson, S. Wang and C.-S. Liou, Wea Forecasting 22 (2007) 950. 11. S. Chen, T. J. Campbell, H. Jin, S. Gabersek, R. M. Hodur and P. Martin, Mon. Wea. Rev. 138 (2010) 3579. 12. P. J. Martin, NRL Report: NRL/FR/7322-009962, Naval Research Laboratory, Stennis Space Center, MS, p. 42, (2000). 13. J. Anderson, T. Hoar, K. Raeder, H. Liu, N. Collins, R. Torn and A. Arellano, Bull. Amer. Meteor. Soc. 90 (2009) 1283. 14. T. M. Hamill, J. S. Whitaker, D. T. Kleist, M. Fiorino, S. G. Benjamin, Mon. Wea. Rev. 139 (2011) 3243–3247. 15. J. L. Anderson, Tellus A 61 (2009) 72–83.

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch03

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


FACTORS CONTROLLING THE SPATIAL DISTRIBUTION OF STABLE ISOTOPES IN PRECIPITATION OVER KUMAMOTO, JAPAN MASAHIRO TANOUE∗ , KIMPEI ICHIYANAGI, and JUN SHIMADA Graduate School of Science and Technology, Kumamoto University, Japan NAOKI KABEYA Forestry and Forest Products Research Institute, Japan

Precipitation samples were collected every 2 weeks between November 2009 and December 2010 at six points over Kumamoto in southwest Japan. The range of δ18 O and d-excess (= δD − 8∗ δ18 O) in precipitation ranged between −13.4
to −3.5
and from 2.6
to 35.6
, respectively. Precipitation δ18 O displayed spring maxima and summer minima. D-excess displayed summer minima and winter maxima. Summer (June to August) had prevailing southwesterlies, while winter (December to February) was characterized by prevailing northeasterlies. Monthly precipitation amounts were high from June to August. Directions of water vapor flux were calculated from the column total of water vapor flux by using meteorological reanalysis datasets. The apparent inland effect from the southwest (west or northwest) coast corresponded to the source direction of the water vapor flux. Spatial distribution of δ18 O in precipitation was affected by the apparent inland effect (decreasing δ18 O or δD in precipitation with increasing distance from the coast) and is related to the direction of water vapor flux, as well as the effect of the amount (decreasing δ18 O or δD in precipitation with increasing precipitation amount) and, in some cases, temperature (increasing δ18 O or δD in precipitation with increasing temperature). The apparent inland effect was assumed to depend on the rainout process from the coast to each station. As a result, δ18 O in precipitation was related to the distance from southwest, west or northwest coast during seven periods. Amount and temperature effects were found in eight and seven periods, respectively. The local spatial distribution of δ18 O in precipitation around

∗ Present

address: Scientific establishment in this work was carried out at Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan.

29

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch03

M. Tanoue et al.

30

Kumamoto in 14 out of 17 periods was controlled by the transport direction of water vapor flux, precipitation amount and, in some cases, temperature.

1. Introduction Stable isotopes in precipitation (δD and δ 18 O) are useful natural tracers for understanding the hydrological cycle and reconstructing paleoclimates.6,8,9 Since the 1960s, isotopic ratios of precipitation have been studied to elucidate the origins of precipitation and understand the dynamics of the global water cycle. Dansgaard1 reviewed the global and spatial-temporal distribution of isotopic ratios in precipitation, and related these to meteorological and geophysical factors (temperature, precipitation amount, latitude, altitude and the distance from the coast). Such observations have also been used to inform paleoclimatic reconstructions, as well as to verify General Circulation Models. However, most observations on the local scale were restricted to a single point. Hence, spatial distribution of isotopic ratios in precipitation on the local scale is not well understood. Moreover, there were a few studies that revealed seasonal variations of isotopic abundance over narrow regions. There are many Japanese studies on stable isotopes in precipitation; comparing their relationships to meteorological and geographical parameters (precipitation amount, temperature, and altitude), however, there are very few on the inland effect.2,3 These studies focused on the relationships between isotopic ratios and the minimum distance from a coast; unfortunately though, the study area (Kumamoto) was dominated by the effect of monsoonal seasonal winds (southwest in summer, and northwest in winter). Because the inland effect changes isotopic abundances in precipitation, spatial distribution of isotopic ratios in precipitation may be different from the transport direction of water vapor. Isotopic ratios decrease with increasing distance from the coast along the direction of cloud movement. On the other hand, isotopic ratios in precipitation related to precipitation amount, that is the amount effect, have been the subject of many studies.7 However, there are few studies that focused on the spatial distribution of precipitation amount on a local scale. In this study, spatial distribution of isotopic ratios in precipitation was considered at the local scale over Kumamoto (less than 5,000 km2 ), with emphasis on the inland and amount effects.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch03

Factors Controlling the Spatial Distribution of Stable Isotopes

31

2. Study Site and Method 2.1. Sampling Location of observational stations and topography over Kumamoto is shown in Fig. 1, and precipitation sampling periods are shown in Table 1. Precipitation samples were taken about every 2 weeks from November 2009 to December 2010 at Kumamoto University (Kumamoto: 130◦44 E, 32◦ 49 N), Uki (130◦ 42 E, 32◦ 53 N), Ohzu (130◦ 53 E, 32◦ 53 N), National Agricultural Research Center for Kyushu Okinawa Region (Koushi: 130◦ 44 E, 32◦ 53 N), the experimental watershed of the Forestry and Forest Products Research Institute (Kahoku: 130◦ 12 E, 33◦ 22 N) and the Center for Marine Environment Studies in Kumamoto University (Matsushima: 130◦ 25 E, 32◦ 38 N). Each sampling period was identified from P01 to P21. There are two seasonal monsoons that impinge on this study area (see Fig. 1). One is the oceanic monsoon with warm and humid air (OWH) in summer with prevailing southwest wind. The other is the continental monsoon with cold and dry air (CCD) in winter with prevailing northwest wind. The water vapor of these two monsoonal winds has different isotopic signatures.4,9,10 The OWH has low δ 18 O and a low d-excess (< 10
), and the CCD has high d-excess (> 25
) across Japan. However, the CCD has high and low δ 18 O along the coastal area of Japan Sea and Pacific Ocean, respectively.

Fig. 1.

Location of sampling stations and topography over Kumamoto.

Period

2009/11/15 2009/11/27 2009/12/23 2010/1/7 2010/1/19 2010/2/1 2010/2/17 2010/3/7 2010/3/20 2010/4/21 2010/5/7 2010/5/30 2010/6/10 2010/7/1 2010/7/29 2010/8/15 2010/9/9 2010/9/28 2010/10/19 2010/11/4 2011/11/26

Autumn Autumn Winter Winter Winter Winter Winter Spring Spring spring Spring Spring Summer Summer Summer Summer Summer Autumn Autumn Autumn Autumn

41 27 60 6 — 65 34 87 51 152 85 256 0 292 256 24 49 77 16 34 19

Minimum d-excess (
)

65 34 76 17 — 80 113 124 80 203 184 318 19 499 679 93 170 162 63 62 34

−9.1 −13.4 −8.5 −6.0 −12.1 −5.3 −4.2 −6.1 −7.0 −4.7 −6.0 −6.6 −4.3 −12.5 −7.8 −8.9 −6.6 −7.8 −8.9 −6.5 −8.4

−8.8 −11.2 −6.9 −4.3 −10.2 −4.6 −3.5 −5.1 −6.0 −4.2 −3.6 −4.5 0.0 −11.0 −6.0 −5.5 −4.6 −5.8 6.4 −5.6 −6.6

8.1 13.2 18.5 21.5 22.8 16.6 12.5 10.5 11.1 8.7 9.6 8.7 0.0 8.9 7.8 2.3 4.6 10.2 12.7 11.6 13.3

Maximum d-excess (
) 10.1 25.6 21.7 27.8 35.6 19.3 14.6 11.6 12.8 10.9 24.3 10.1 0.0 10.2 8.9 9.8 9.3 13.2 16.7 18.8 22.9

b1387-ch03

2009/11/3 2009/11/16 2009/11/28 2009/12/24 2010/1/8 2010/1/20 2010/2/2 2010/2/18 2010/3/8 2010/3/21 2010/4/22 2010/5/8 2010/5/31 2010/6/11 2010/7/2 2010/7/30 2010/8/16 2010/9/10 2010/9/29 2010/10/20 2010/11/5

Maximum δ18 O (
)

9in x 6in

Season

Minimum δ18 O (
)

AOGS - AS and OS

End

Maximum precipitation (mm)

M. Tanoue et al.

P01 P02 P03 P04 P05 P06 P07 P08 P09 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 P20 P21

Start

Minimum precipitation (mm)

July 13, 2012 9:9

32

Table 1. Precipitation amount, δ18 O and d-excess in each sampling period. Precipitation amount was not available at all stations during P05 since precipitation fell as snow.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Factors Controlling the Spatial Distribution of Stable Isotopes

b1387-ch03

33

2.2. Analytical methods The Isotope Ratio Mass Spectrometer (Thermo-Quest, Delta-S) was used to measure δ 18 O and δD of precipitation samples with a CO2 /H2 O equivalent method for δ 18 O and a Chromium reduction method for δD. Isotopic ratios are expressed by convention as parts per thousand (
), deviations relate to Vienna Standard Mean Ocean Water. The analytical errors for standard measurements of δ 18 O and δD are better than ± 0.1
and ± 1.0
, respectively.

2.3. Determining distance from the coast Schematic figure for determining distances from the northwest (NW), west (W) and southwest (SW) coasts to sampling sites are described in Fig. 2. Distances from all stations to these directional coasts were compared, and then a straight line was drawn along these directional coasts. Then straight lines were drawn from the stations to the coastal lines such that they intercepted each other at 90◦ . The length of these intercept lines was selected as the distance from the NW, W, and SW coasts. The direction of water vapor flux (WVF) was calculated by using the column total of water vapor in the Japanese 25-year Reanalysis (JRA-25) and JMA Climate Data Assimilation System (JCDAS). The WVF at each period is shown in Table 2.

Fig. 2. Schematic figure for determining distance from NW, W, and SW coasts to each station.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch03

M. Tanoue et al.

34 Table 2.

Summary of apparent inland effect depending on the direction of WVF. Inland effect (min)

Period WVF P01 P03 P06 P08 P09 P19 P20

r

NW — WNW — SW — SW — WSW — NW −0.75 N −0.63

Inland effect (SW)

Inland effect (W)

Inland effect (NW)

Slope (
100 km)

r

Slope (
100 km)

r

Slope (
100 km)

r

Slope (%100 km)

— — — — — −2.06 −3.65

— — −0.63 −0.95 — — —

— — −0.84 −1.51 — — —

— −0.94 — — −0.66 — −0.66

— −4.60 — — −2.31 — −1.55

−0.84 — — — — −0.91 —

−1.59 — — — — −5.92 —

r : correlation coefficient; WVF: water vapor flux.

3. Results and Discussion 3.1. Time series of stable isotopes in precipitation Time series of δ 18 O and d-excess in precipitation are shown in Fig. 3. The δ 18 O ranged from −13.4
to −3.5
, with spring maxima (from P08 to P12) and minima during P02, P05 and P14. All stations shared similar trends between P01 and P14 and from P20 to P21. However, this was not true between P15 and P19. High δ 18 O and negative d-excess in P13 was observed. During P13, there was little precipitation, and sample evaporation from collection bottles was a concern. The d-excess time series shared a similar trend among all stations from P01 to P12, P14 and from P18 to P21. The trends were not similar between P15 and P17. Similar trends of d-excess among all stations indicate that the source of water vapor to all stations was the same. On the other hand, dissimilar trend from P15 to P17 may indicate a different source of water vapor or a local precipitation system operating among them. The d-excess ranged from 2.3
to 35.6
with winter maxima and summer minima. Consistent with the results from previous work, the seasonal variation of stable isotopes in precipitation over East Asia was related to monsoonal water vapor with different isotopic ratios.4,9,10 In the study area, seasonal variations of stable isotopes in precipitation were affected by the OWH and CCD (see Fig. 1(a)). Stable isotopes in precipitation were affected by the OWH in summer (P14 to P17) giving low δ 18 O (−8
to −6
) with low d-excess (< 10
), and CCD in winter (P03 to P07) with high δ 18 O (−6
to −3
) with high d-excess (> 25
). With the exception of the spring (P08 to P12) and P05 and P14, the seasonal variations of isotopic ratio

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Factors Controlling the Spatial Distribution of Stable Isotopes

b1387-ch03

35

Fig. 3. Time series of (a) δ18 O and (b) d-excess in precipitation observed at six stations over Kumamoto.

in precipitation can be explained by these monsoons. Snow samples in P05 had low δ 18 O (−12.1
to −10.2
) with extremely high d-excess (22.8
to 35.6
). Low δ 18 O (−12.5
to −11.0
) with low d-excess (8.9
to 10.2
) were observed in P14 probably caused by the heavy rain at the Baiu front near Kumamoto. These characteristics of stable isotopes in precipitation in P14 were consistent with those for Tokyo and Ryori on the Pacific coast in summer.4 Therefore, the source of water vapor in P14 was likely to be from the Pacific Ocean. 3.2. “Apparent” inland effect The inland effect in each direction (NW, W, and SW) was compared with that of the nearest coast (Min). The result of the inland effect is shown in Table 3. A representative inland effect in P01, P03, P08, P19 is shown in Fig. 4. In the Min case, the inland effect was found during P07, P19, and P20. The slope of inland effect at Min ranged from −2.06
/100 km (P19) to −3.91
/100 km (P07). Yabusaki et al.3

.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch03

M. Tanoue et al.

36

Table 3. Summary of distance from the southwest/northwest coast to each station, altitude and annual mean temperature. Distance from the coastline (km)

Kumamoto Uki Koushi Kahoku Matsushima Ohzu

Min

NW

W

SW

Altitude (m)

11.5 3.9 27.9 29.7 0.0 18.9

126 138 121 99 129 130

98 95 100 97 70 114

100 85 106 121 57 116

21 41 115 307 44 147

Annual mean temperature (◦ C) 16.1 16.2 14.7 13.7 15.9 14.6

Min: distance from the nearest coast; NW: northwest; W: west; SW: southwest.

Fig. 4. Correlation between δ18 O in precipitation and distance from the coast in (a) P01, (b) P03, (c) P08, and (d) P19. Distance from the coast in left and right panels indicates distance from the nearest coast (Min) and meaningful distances (dependent on the WVF directions) (NW, W, and SW), respectively.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch03

Factors Controlling the Spatial Distribution of Stable Isotopes

37

and Taniguchi et al.2 reported that the inland effect for δ 18 O was −0.89
/100 km (Kanto Plain in eastern Japan) and −1.91
/100 km for δ 18 O (Lake Biwa in central Japan). The slope of the inland effect obtained in this study was higher than that obtained in previous studies, because the sampling period was less than 1 year. The apparent inland effect depends on how the transport direction of WVF is defined. The apparent inland effects at NW, W and SW were recognized during seven periods (P01, P03, P06, P08, P09, P19, and P20) mainly in autumn and winter. The values obtained for these periods were greater than the Min case. The meaningful distance from the coast is the one that represents the actual distance and direction traveled by the WVF, rather than the distance from the nearest coast. Slopes of the apparent inland effect ranged from −0.84
/100 km in P06 to −5.92
/100 km in P19. 3.3. Amount effect The result of isotopic effects is shown in Table 4. The amount effect, decreasing δ 18 O in precipitation with increasing precipitation amount, Table 4. Result of isotopic effects less (more) than −0.60 (0.60). The apparent inland effect with WVF of considered transported direction is shown. Shading indicates that the variation of δ18 O is not related to isotopic effects. Inland effect Period WVF P01 P03 P04 P06 P07 P08 P09 P10 P12 P14 P15 P16 P17 P18 P19 P20 P21

r

NW −0.84 WNW−0.94 WNW — SW −0.63 WSW SW −0.95 WSW −0.66 WSW — W — W — SW — W — NW — NW — NW −0.91 N −0.66 NW —

Temperature effect

Slope (
100 km)

r

Slope (
/◦ C)

−1.59 −0.05 — −0.84

— 0.73 — —

— 0.32 — —

−1.51 −0.02 — — — — — — — −5.92 −0.02 —

0.81 — — — — 0.68 0.81 — 0.67 — 0.73 0.75

0.14 — — — — 0.78 1.67 — 0.49 — 0.27 0.43

r : correlation coefficient; WVF: water vapor flux.

Amount effect r −0.84 — −0.77 — −0.72 −0.78 — — −0.60 — — −0.93 — −0.76 −0.75 — —

Slope (
100 mm) −1.08 — −11.19 — −0.53 −1.90 — — −0.87 — — −3.78 — −1.31 −4.08 — —

Altitude effect r

Slope (
100 m)

— — — — — −0.86 — — — — −0.91 — — −0.61 — — −0.87

— — — — — −0.27 — — — — −0.56 — — −0.40 — — −0.58

July 13, 2012 9:9

AOGS - AS and OS

38

9in x 6in

b1387-ch03

M. Tanoue et al.

was identified in eight periods (P01, P04, P07, P08, P12, P16, P18, and P19). The slope of the amount effect ranged from −0.53
/100 mm in P07 to −4.08
/100 mm in P19. P04 is an exception because the precipitation was extremely low (11 mm) during this period. In previous studies, the amount effect ranged from −1.0
/100 mm to −6.0
/100 mm in low and middle latitudes.5 Obtained values of the amount effect in this study were lower than those obtained from that previous study. The amount effect implies relationships between δ 18 O and precipitation amount among several stations in the same period. However, in previous work, it was calculated at only one station over a long-term period. 3.4. Temperature and altitude effects Temperature and altitude effects were found during P07 and P04, and during these periods, both altitude and temperature effects appeared together. Ichiyanagi5 reported a temperature effect at global scales, ranging from 0.3
/◦ C to 0.5
/◦ C. The temperature effect in this study, from 0.14
/◦ C in P08 to 1.67
/◦ C in P16, was lower than those found in previous studies, though it might be considered that the variation of temperature at a local scale was much lower than that on a global scale. 3.5. Non-correlated cases There were apparent inland, precipitation amount, and/or temperature effects in most periods (14 of 17 periods), but in only three periods (P10, P14, and P17) were none of these effects identified. Differences in δ 18 O among six stations in P10 were the smallest value (0.5
) over all periods. It may be that during this period, stratiform clouds covered the study area. The Baiu front dominated in P14, and this might be because of extremely low isotopic ratios in precipitation from other periods. Water vapor of the Baiu front usually originates in the tropics, so isotopic ratios are depleted as a result of the rainout process during long-term transportation. Local precipitation events with high rainfall near Kumamoto and Koushi on 18 August were found in P17. δ 18 O at these sites were relatively low (−5.5
and −4.2
) compared with other sites (−6.2
to −6.6
). 4. Concluding Remarks The seasonal differences in precipitation δ 18 O (high in summer) and d-excess (high in winter) were caused by the monsoonal winds in which water vapor

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Factors Controlling the Spatial Distribution of Stable Isotopes

b1387-ch03

39

has different isotopic ratios. A similar trend of d-excess among all stations indicates that the source of water vapor was the same in all cases. In most periods, apparent inland, precipitation amount, and/or temperature effects were present. The apparent inland effect is best calculated by using a distance from the coast that reflects the direction of WVF. In this work, in seven periods, this was greater than that for distance from the nearest coast. Even at the local scale over Kumamoto, the spatial distribution of δ 18 O in precipitation was controlled by the apparent inland effect of the transported directions of WVF, and in some cases temperature and precipitation amount.

Acknowledgments The authors are grateful to Dr. Atsushi Maruyama of the National Agricultural Research Center for Kyushu Okinawa Region for the installation of the precipitation sampler and to Mr. Hideyuki Shimasaki of the Center for Marine Environment Studies in Kumamoto University for collecting precipitation samples. We are also grateful to the two anonymous reviewers for their constructive comments to improve our manuscript.

References 1. W. Dansgaard, Tellus XVI (1964) 436. 2. M. Taniguchi, T. Nakayama, N. Tase and J. Shimada, Hydrol. Process. 14 (2000) 539. 3. S. Yabusaki, N. Tase and J. Shimano, Groundwater Response to Changing Climate, IAH book No. 16 (CRC Press, 2010). 4. L. K. Aragu´ as-Aragu´ as, K. Froehlich and J. Rozanski, Geophys. Res. 103 (1998) 28721. 5. K. Ichiyanagi, J. Assoc Hydrol. Sci. 37 (2007) 165. 6. K. Yoshimura, M. Kanamitsu, D. Noone and T. Oki, J. Geophys. Res. 113 (2008) D19108. 7. N. Kurita, K. Ichiyanagi, J. Matsumoto, M. D. Yamanaka and T. Ohta, Earth and Planetary Science Letters 102 (2009) 113. 8. C. Risi, S Bony, F. Vimeux, M. Chong and L. Descroix, Q. J. R. Meteorol. Soc. 136(s1) (2010) 227. 9. T. Peng, C. Wang, C. Huang, L. Fei, C. A. Chen and J. Hwong, Earth and Planetary Science Letters 289 (2010) 357. 10. K. Lee, A. J. Grundstein, D. B. Wenner, M. Choi, N. Woo and D. Lee, Clim. Res. 23 (2003) 137.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-ch03

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch04

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


ASIAN DUST DEPOSITION OVER THE LAND AND SEAS IN 2010 ESTIMATED BY THE ADAM2 MODEL SOON-UNG PARK, MOON-SOO PARK and ANNA CHOE Center for Atmospheric and Environmental Modeling, Seoul National University Research Park Rm. 515 San 4-2, Bongcheon-dong, Gwanak-gu,Seoul, 151-919, Korea

The Asian Dust Aerosol Model 2 (ADAM2) with the MM5 meteorological model has been employed to estimate dust emission, dust concentration, and wet and dry depositions of dust in the Asian region for the year of 2010. It is found that the domain mean annual total maximum dust emission is found to be 248 t km−2 in Central–Northern China (CNC) with decreasing emission rate away from this domain to 14 t km−2 in Heilongjiang province (HEI). The annual total dust emission from the whole Asian dust source region is found to be 540 Tg. The annual average maximum column-integrated concentration in the source region is found to be 2.1 mg m−2 in CNC and that in the downwind region is 0.90 mg m−2 over the Yellow Sea (YES). It is also found that the domain mean annual total maximum deposition of dust in the source region is 126 t km−2 (dry deposition, 117 t km−2 ; wet deposition, 9 t km−2 ) in the domain CNC and in the downwind region it is 27 t km−2 (dry deposition, 7 t km−2 ; wet deposition, 20 t km−2 ) over the YES. The estimated dust deposition could adversely impact the eco-environmental system in the downwind regions of the Asian dust source region, especially over the seas.

1. Introduction Asian dust storms occurring in the arid and semi-arid regions of northern China and Mongolia have caused major aerosol events well beyond the Asian continent. Some of them have been transported to the western parts of USA across the Pacific Ocean.1–5 Asian dust storms in the source region occur all year round with the maximum frequency during the spring season.6–9 The maximum PM10 concentration observed in the downwind region of Korea frequently exceeds 1,000 µg m−3 during the dust event period.6–9 This often causes severe 41

July 13, 2012 9:9

42

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

environmental impacts, including temporal closing of most of airports and elementary schools in Korea.6,10 Long-range transport and deposition of dust cause not only soiling of the exposed materials resulting in corrosion of buildings and valuable monuments but also respiratory disease (asthma) and a change in the chemical composition of soil.11 In addition, adverse effects are air pollution,4 cattle suffocation, damage to young plants, visibility reduction leading to airport closures and road accidents,6,12 damage to sensitive scientific and industrial equipments.13 Asian dust also affects significantly atmospheric chemistry, the radiative balance over the region causing possible climate change,14–18 and biological productivity in the North Pacific Ocean by providing bio-available iron in dust,19–22 which, in turn, by changing the rate at which atmospheric CO2 being fixed by oceanic biota, exerting a global-scale influence on climate.23−25 It is believed that Sahara dust carries bacteria and fungi across the Atlantic Ocean causing fungal disease, bacterial pathogens of rice and beans, foot and mouth disease and coral mortality in the Caribbean Island.26 The assessment of all these adverse impacts of dust require more accurate estimations of dust emissions in the dust source regions, longrange transport processes and wet and dry deposition for a long-term period. There are many numerical simulations of Asian dust transport models.1,3,4,6,8,27–29 However, there are very few estimates of total dust deposition in the Asian region for a year-long period that make it possible to assess the impacts of dust on human health and eco-environmental systems. There are a few estimates of Asian dust deposition in a particular region for a dust event case.29,30 Dust emission in the Asian dust source region and depositions of dust in the downwind regions for the month of March 201031 showed that the Asian Dust Aerosol Model 2 (ADAM2) model simulated quite reasonably the dust (PM10 ) concentration compared with the observed ones both in the dust source region and the downwind region of Korea. They also found that the monthly total maximum dust emission of more than 32 t km−2 and that of deposition of more than 25.4 t km−2 occurred in central northern China with the increasing ratio of the total dust deposition to the total emission towards the downwind direction in the source region from 0.4 in the Taklimakan desert to 0.8 in north eastern China.31 However, such a short-term case study may not be enough to understand the impacts of dust deposition on human health and the ecoenvironmental systems.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Asian Dust Deposition Over the Land and Seas

b1387-ch04

43

The purpose of this study is to estimate dust emission in the Asian dust source region and both dry and wet deposition of dust in the whole model domain for the whole year of 2010 to provide necessary data for the Asian dust impact studies on human health, eco-system and near-term climate change. The ADAM2 and the MM5 meteorological model are used to simulate Asian dust events occurred in 2010 when the highest dust event days (15 days) were observed in Seoul Korea in last 5 years.

2. Model Description 2.1. Meteorological model The meteorological model used in this study is the fifth-generation mesoscale model of non-hydrostatic version (MM5, Pennsylvania State University/NCAR) in the x, y, and σ coordinates.32,33 The model domain is given in Fig. 1 including dust source regions of Sand, Loess, Gobi, and Mixed soil region.6,7,27,34 The model has a horizontal resolution of 30 km with 25 vertical layers. The simulation has been conducted for the whole year of 2010. The 6-hourly reanalyzed

Fig. 1. The model domain and the Asian dust source region delineated by the total number of dust-rise reporting days for 9 years (1998–2006). The surface soil types of Gobi, Sand, Loess and Mixed soil in the Asian dust source region are indicated.

July 13, 2012 9:9

AOGS - AS and OS

44

9in x 6in

b1387-ch04

S.-U. Park et al.

National Center for Environmental Program (NCEP) data are used for the initial and boundary conditions for the MM5 model. 2.2. The ADAM2 model The detailed description of the ADAM2 model is given in Park et al. (2010a).7 The ADAM2 model, modified from the ADAM1 model6 includes the Asian dust aerosol only and uses 11 dust-size bins with near the same logarithmic interval for the particles of 0.1–37 µm in radius6,27 and uses time-dependent threshold wind speeds for dust-rise in the dust source regions delineated by the statistical analysis of the WMO dust reporting data for 9 years (1998–2006) in different soil surfaces in the domain (Fig. 1). This model also uses time-dependent dust emission reduction factors due to vegetation parameterized with the use of the Normalized Difference Vegetation Index (NDVI) values for 9 years (1998 to 2006) obtained from the Spot/Vegetation product of the maximum value composite syntheses for a ten-day period in a spatial resolution of 1×1 km2 (http://free.vgt.vito.be) in the Asian dust source region (Fig. 1). Dry deposition of dust is estimated with the dry deposition velocity multiplied by concentration of dust near the surface. The dry-deposition velocity is estimated with the use of the inferential method35–38 with taking into account a gravitational settling velocity. The wet deposition amounts of dust in the ADAM2 model are determined by the precipitation rate and the averaged dust concentration in cloud water estimated by the sub-grid cloud scheme followed by the diagnostic cloud model in the Regional Acid Deposition Model (RADM) version 2.6.39–41 The below cloud scavenging process is also included.38 The detailed dry and wet deposition of dust is given in other paper.8

3. Simulations of Asian Dust Events Occurred in 2010 The 2010 Asian dust events observed in Korea were characterized by having the highest dust-event days (15 days in Seoul) in last 5 years and they observed not only in spring but in autumn and winter in association with relatively cool and dry in winter and spring in the source regions of eastern Mongolia and northeastern China.42 The ADAM2 model has been used to simulate Asian dust events occurred in the Asian domain for the whole year of 2010 with the MM5 model. The ADAM2 model uses the NDVI distribution averaged for each

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Asian Dust Deposition Over the Land and Seas (a-1)

(b-1)

(a-2)

(b-1)

b1387-ch04

45

Fig. 2. The horizontal distributions of the averaged NDVI value in (a-1) January and (b-1) August with the use of 9-year data (1998–2006) and the emission reduction factor due to vegetation in (a-2) January and (b-2) August.

month using 9 years (1998–2006) data in the model domain. Even though NDVI data in 2010 are available, some of data are missing and erroneous so that a consistent year-long dataset cannot be obtained with one year data only. The spatial distribution of averaged NDVI and that of emission reduction factor due to vegetation in January (winter) and August (summer) are given in Fig. 2. The emission reduction factor is estimated by the empirical regression equation7 in the Asian dust source region. Most of the Asian dust source region (Fig. 1) in January (winter) show that the NDVI value is less than 0.15 except in northern India where the NDVI value is larger than 0.55 (Fig. 2(a-1)). Consequently, the emission reduction factor due to vegetation is almost negligible in most parts of the source regions except in northern India where the dust emission can be almost completely (emission reduction factor > 0.95) suppressed by vegetation (Fig. 2(a-2)). However, in August (summer time) large portions of the Asian dust source regions (Fig. 1) are covered with vegetations (NDVI value is larger

July 13, 2012 9:9

46

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

than 0.5) except for the sand and gobi deserts in Taklimakan, central northern Inner Mongolia and the Mongolia Gobi regions (Fig. 2(b-1)), so that the dust emission is greatly suppressed due to vegetations, especially in the Loess plateau and northeastern China and northwestern India where the dust emission reduction factor is as large as one (Fig. 2(b-2)). Time series of observed and simulated near surface PM10 concentrations in 2010 at a site located in the dust source region (Fig. 3(a)) and the downwind region of Korea (Fig. 3(b)) show that the model simulates quite well the starting and ending times and the peak PM10 concentration occurrence times of most dust events at these sites in both the source region and the downwind region. The overall performance test of the ADAM2 model31 showed that the hit rate of the model for dust events was more than 83%, implying the potential capability of the model to be used to evaluate impacts of Asian dusts on environment with less than 20% error.

Fig. 3. Time series of observed (dotted line) and simulated (solid line) surface PM10 concentration (µg m−3 ) by ADAM2 in May, August and November 2010 at (a) Jurihe in China and (b) Baeknyeongdo in Korea.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Asian Dust Deposition Over the Land and Seas

b1387-ch04

47

Fig. 4. Horizontal distributions of the annual mean (a) surface PM10 concentration (µg m−3 ) and (b) column-integrated PM10 concentration (kg km−2 ).

The annually averaged surface PM10 concentration and the columnintegrated PM10 concentration simulated by ADAM2 in 2010 over the model domain are given in Fig. 4. The annual mean surface PM10 concentration (Fig. 4(a)) over the Asian dust source region exceeds 50 µg m−3 with maximum of more than 200 µg m−3 over the Badain Jaran desert in Inner Mongolia (Fig. 4(a)). The annual impact of the Asian dust at the lower level extends to the East Sea to the east, the northern slope of Tibetan Plateau and the north of Taiwan to the south, the northern border of Mongolia to the north, and Central Asia to the west (Fig. 4(a)). The annual mean column-integrated PM10 concentration exceeds 50 kg km−2 in most Asian dust source regions with a maximum more than 300 kg km−2 over the maximum surface dust concentration region of the Badain Jaran desert (Fig. 4). The spatial distribution pattern of the annual mean columnintegrated PM10 concentration exceeding 10 kg km−2 is quite similar to that of the annual mean surface PM10 concentration (Fig. 4(a)), except for the further eastward extension of the column-integrated dust concentration over the northwestern Pacific Ocean (Fig. 4(b)). 4. Asian Dust Emission and Deposition Figure 5 shows the horizontal distributions of the annual total dust emission (Fig. 5(a)), dust deposition (Fig. 5(b)) and dry (Fig. 5(c)) and wet (Fig. 5(d)) deposition of dust in 2010 estimated by the ADAM2 model. The annual total dust emission of more than 10 t km−2 occurs in most dust source regions with a maximum of more than 1,000 t km−2 in the Badain Jaran desert area (Fig. 5(a)) where the maximum surface concentration occurs (Fig. 4).

July 13, 2012 9:9

48

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

Fig. 5. Horizontal distributions of the annual total (a) dust emission (t km−2 ), (b) dust deposition and (c) dry and (d) wet deposition (t km−2 ).

The horizontal distribution pattern of the annual total dust deposition (wet + dry deposition; Fig. 5(b)) is quite similar to that of the annual mean column-integrated PM10 concentration (Fig. 4(b)) with the maximum annual total deposition of more than 500 t km−2 over the high dust emission region in Fig. 5(a). The annual total dust deposition of more than 30 t km−2 extends eastward from the Taklimakan desert region to the Korean peninsula (Fig. 5(b)). Most of dust source regions are predominated by dry deposition with a maximum value of 522 t km−2 in the Badain Jaran desert region (Fig. 5(c)). However, wet deposition contributes largely to the total dust deposition over the seas, especially over the Yellow Sea (YES), East Sea and the northwestern Pacific Ocean. The annual total maximum wet deposition of 71 t km−2 occurs over the YES (Fig. 5(d)). To understand spatial and temporal variations of dust deposition and emission, six domains (TAK, NWC, CNC, NEC, HEI, and ENEC) in the dust source region and eleven domains (YES, KOR, ECS, SER, EAS, SJN, NWP3, NWP1, NWP2, SEA, and SCS) in the downwind region of the

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Asian Dust Deposition Over the Land and Seas

b1387-ch04

49

Fig. 6. Domains for the estimations of emission, concentration, deposition (wet and dry deposition). TAK: Taklimakan, NWC: Northwestern China, CNC: Central Northern China, NEC: Northeastern China, HEI: Heilongjiang Province, ENEC: East Northeastern China, YES: Yellow Sea, KOR: Korean peninsula, EAS: East Sea, NWP: Northwestern Pacific Ocean, SJN: Southern Japan, ECS: East China Sea, SCS: South China Sea, SEA: South-East Asia.

Asian dust source region are chosen (Fig. 6). The domain TAK is located over the Taklimakan desert (80–90E, 37–42N), the domain NWC over northwestern China including parts of eastern Xinjiang, northern Gansu, south-western Mongolia and western tip of Inner Mongolia (90–100E, 37–45N), the domain CNC over central northern China including the Badain Jaran desert, Tengger desert, Mu Us desert and southern Mongolia (100–110E, 37–45N), the domain NEC over northeastern China including parts of Inner Mongolia, Hebei, Shanxi province, and Beijing (110–120E, 37–45N), the domain HEI including parts of northeastern Inner Mongolia and Heilongjiang province (120–130E, 42–50N), the domain ENEC over east-northeastern China including Jilin province (120–125E, 40–45N). The downwind domain of YES (Fig. 6) is located over the Yellow Sea (120–125E, 32–40N), the domain KOR over the Korean peninsula (125–130E, 32–42N), the domain ECS over the East China Sea (120–130E, 22–32N), the domain SER over southeastern Russia (130–140E, 42–50N), the domain EAS over the East Sea (130–140E, 37–42N), the domain SJN over southern Japan (130–140E, 32–37N), the domain NWP3 over the

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

50

Source Region 22

240

I.CON. DRY WET EMIT

220

18

180

16

160

14

140

12

120

10

100

8

80 6

60 40

4

20

2

0

TAK

NWC

CNC

NEC

ENEC

HEI

Col. CON (10-2 g m-2)

EMIT/DEP (ton km-2)

200

20

0

Fig. 7. Annual total domain-mean dust emission ( t km−2 ), deposition ( wet; dry; t km−2 ) and annual area mean column-integrated dust concentration 10−2 g m−2 ) at the source region. (

northwestern Pacific Ocean to the south of Japan (130–140E, 22–32N), the domain NWP1 over the northwestern Pacific Ocean including northern Japan (140–150E, 42–50N), the domain NWP2 over the northwestern Pacific Ocean to the east of Japan (140–150E, 32–42N), the domain SEA over the South East Asia (95–110E, 15–22N) and the domain SCS over the South China Sea (110–120E, 15–22N). Figure 7 shows the annual mean column-integrated dust concentration, annual total dust emission and deposition (wet and dry) in the domains of the Asian dust source region. The highest annual total dust emission, likewise the annual total dust deposition and the annual mean columnintegrated concentration, occurs in the domain CNC. The annual total dust emission in this domain is about 248 t km−2 , of which about 50.8% (126.1 t km−2 ) are deposited in the domain through the dry deposition (116.7 t km−2 ) and the wet deposition (9.4 t km−2 ). The contribution of dry deposition to the total deposition is as large as 93%, suggesting the importance of dry deposition in the source region. The annual mean columnintegrated concentration in this domain is 2.1 mg m−2 . Figure 7 clearly indicates that the emission, deposition and the columnintegrated concentration decrease toward the downwind direction from the domain CNC to the domain HEI. The annual total emission, deposition, and the annual mean column-integrated concentration are, respectively,

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch04

Asian Dust Deposition Over the Land and Seas

51

248 t km−2 , 126.1 t km−2 , and 2.1 mg m−2 in the domain CNC, 81 t km−2 , 51.4 t km−2 , and 1.32 mg m−2 in the domain NEC, 54 t km−2 , 39.0 t km−2 , and 0.98 mg m−2 in the domain ENEC and 14 t km−2 , 14.6 t km−2 , and 0.49 mg m−2 in the domain HEI. However, the ratio of the total deposition to the total emission increases toward the downwind direction from 31% in the domain TAK, 40% in the domain NWC, 51% in the domain CNC, 63% in the domain NEC, 72% in the domain ENEC and 104% in the domain HEI. This increase of the ratio is caused by dust advection from the upstream source regions. The contribution of dry deposition to the total dust deposition is 75% in the TAK, 90% in the domain NWC, 93% in the domain CNC, 86% in the domain NEC, 62% in the domain ENEC and 47% in the domain HEI. Figure 8 shows the annual total dust deposition (dry + wet) and the annual mean column-integrated dust concentration in domains located in the downwind regions of the Asian dust source region (Fig. 6). The annual total dust deposition and the annual mean columnintegrated dust concentration decrease away from the source region from 26.6 t km−2 (dry deposition of 7.0 t km−2 and wet deposition of 19.6 t km−2 ) and 0.9 mg m−2 in the domain YES to 1.5 t km−2 (dry deposition of 0.5 t km−2 and wet deposition of 1.0 t km−2 ) and 0.12 mg m−2 in the domain NWP3. The impact of the Asian dust storms is negligible in the domains SCS (South China Sea) and SEA (South East Asia).

30

Downwind Region

10 I. CON DRY WET

27

9 8

21

7

18

6

15

5

12

4

9

3

6

2

3

1

0

YES

KOR

EAS

SJN

SER

ECS

NWP2 NWP1 NWP3

SCS

SEA

Col. CON (10-2 g m-2)

DEP (t km-2)

24

0

Fig. 8. Annual total domain-mean dust deposition ( wet; dry; t km−2 ) 10−2 g m−2 ) in the and annual area mean column-integrated dust concentration ( downwind region.

July 13, 2012 9:9

52

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

Much of the annual total dust deposition in the downwind region is largely contributed by wet deposition (more than 80%), especially over the seas (more than 90%), which is contrast to that in the source region where the dry deposition is predominated (Fig. 7). The annual total dust emission in 2010 from the whole Asian dust source region (Fig. 1) estimated by ADAM2 was about 540 Tg. This estimate is slightly lower than that of IPCC report of 800 Tg yr−1 .43 The difference might be caused by the area difference of the dust source region. The present model does not include West Asia where sand deserts are prevailing.

5. Summary and Conclusions ADAM2 with the MM5 meteorological model has been implemented to estimate the dust emission, dust concentration, wet and dry depositions over the Asian dust source region and the downwind regions for the year of 2010 to estimate the impact of Asian dust to the environment. The domain-mean annual total dust emission decreases toward the downwind direction from the highest value of 248 t km−2 in the domain CNC to the lowest value of 14 t km−2 in the domain HEI. The annual total dust emission from the whole Asian dust source regions is found to be 540 Tg in 2010, which is slightly lower than that of IPCC report of 800 Tg. The model simulated annual total dust deposition of 126.1 t km−2 (wet deposition of 9.4 t km−2 and dry deposition of 116.7 t km−2 ) in the domain CNC (Central–Northern China) is the highest value among all Asian dust source domains. Much of the total dust deposition is found to be contributed by dry deposition (more than 80%) due to the lack of precipitation in the Asian dust source region. However, the wet deposition contribution to the total dust deposition becomes important toward downwind domains of the source region such as in the domains ENEC (wet deposition of 38%) and HEI (wet deposition of 53%) due to frequent passages of synoptic frontal systems accompanied with precipitation. The ratio of the annual total deposition to that of emission is found to increase toward the downwind direction in the source region from 31% in the domain TAK to 105% in the domain HEI. The annual mean column-integrated dust concentration and the annual total dust deposition are found to decrease quite rapidly with distance away from dust source regions. The annual mean column-integrated dust concentration of 0.90 mg m−2 in the domain YES to 0.12 mg m−2 in the domain NPW3 and the annual total dust deposition of 26.6 t km−2

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Asian Dust Deposition Over the Land and Seas

b1387-ch04

53

(dry deposition of 7.0 t km−2 and wet deposition of 19.6 t km−2 ) over the YES decreases to 1.5 t km−2 (dry deposition of 0.5 t km−2 and wet deposition of 1.0 t km−2 ) over the Northwestern Pacific Ocean (NWP3). Much of the total dust deposition is found to be contributed by the wet deposition (more than 80%) in the downwind domains, especially over the seas where the contribution exceeds 90%. The presently estimated total dust deposition could affect the biological activities in the downwind regions of the seas including the YES and the East Sea. This study is mainly devoted to document the Asian dust emission, deposition (wet and dry) and concentration in the source regions and the downwind regions for a year long period to be used for the eco-environmental impact assessments of the Asian dust. This requires further studies on the anthropogenic aerosols. Acknowledgments This work was funded by the Korea Meteorological Administration Research and Development Program under Grant CATER 2008-3208. The observed surface PM10 concentration data are obtained from the National Institute of Meteorological Research in the Korea Meteorological Administration. References 1. R. Husar et al., J. Geophy. Res. 106(D16) (2001) 18317. 2. A. D. Clarke et al., J. Geophy. Res. 106 (2001) 32555. 3. F. E. Grousset, P. Ginoux, A. Bory and P. E. Biscaye, Geophy. Res. Lett. 30(6) (2003) 1277. 4. R. A. VanCuren, J. Geophy. Res. 108(D20) (2003) 4623. 5. N. C. Hsu, S. C. Tsay, M. D. King and J. G. Herman, Geosci. Remote Sens. 44 (2006) 3180. 6. S.-U. Park and E.-H. Lee, Atmos. Env. 38 (2004) 2155. 7. S.-U. Park, A. Choe, E.-H. Lee, M.-S. Park and X. Song, Theor. Appl. Climatol. 101 (2010a) 191. 8. S.-U. Park, A. Choe and M.-S. Park, Sci. Tot. Env. 408 (2010b) 2347. 9. S.-U. Park, A. Choe, M.-S. Park and Y. Chun, J. Sust. Energy & Env. 1 (2010c) 77. 10. H.-J. In and S.-U. Park, Atmos. Env. 36 (2002) 4173. 11. L. J. Hagen and N. P. Woodruff, Atmos. Env. 7 (1973) 323. 12. B. Buritt and A. Hyers, Geol. Soc. Am. Bull. Sp. Paper 186 (1981) 282. 13. A. Goudie, Pakistan Earth Surf. Proc. 2 (1977) 75. 14. I. Sokolik and O. B. Toon, Nature 381 (1996) 681.

July 13, 2012 9:9

54

AOGS - AS and OS

9in x 6in

b1387-ch04

S.-U. Park et al.

15. Y. J. Kaufman, D. Tanre, O. Dubovik, A. Karnieli and L. A. Remer, Geophy. Res. Lett. 28 (2001) 1479. 16. S.-U. Park, L.-S. Chang and E.-H. Lee, Atmos. Env. 39 (2005) 2593. 17. J.-I. Jeong and S.-U. Park, Sci. Tot. Env. 392 (2008) 262. 18. S.-U. Park and J.-I. Jeong, Sci. Tot. Env. 407 (2008) 394. 19. Y. Gao, R. Arimoto, M. Y. Zhou, J. T. Merrill and R. A. Duce, J. Geophy. Res. 97 (1992) 9867. 20. N. Meskhidze, W. L. Chamides and A. Nenes, J. Geophy. Res. 110 (2005) D03301. 21. R. A. Duce et al., Glob. Biogeochem. Cyc. 5(3) (1991) 193. 22. R. A. Duce and N. W. Tindale, Limnol. Oceanogr. 36(8) (1991) 1715. 23. G. Zhuang, Z. Yi, R. A. Duce and P. R. Brown, Nature 355 (1992) 537. 24. D. J. Cooper, A. J. Watson and P. D. Nightingale, Nature 383 (1996) 511. 25. S. M. Turner et al., Nature 383 (1996) 513. 26. S. Nickovic and L. Barrie, WMO SDS-WAS Workshop (2007). 27. S.-U. Park and H.-J. In, J. Geophy. Res. 108(D19) (2003) 4618. 28. I. Uno et al., J. Geophy. Res. 109 (2004) D19S24. 29. L. Su and O.B. Toon, J. Geophy. Res. 114 (2009) D14202. 30. S. Lee, Y. Chun, J.-C. Nam, S.-U. Park and E.-H. Lee, J. Meteor. Soc. Japan 82A (2005) 242. 31. S.-U. Park, A. Choe and M.-S. Park, Theor. Appl. Climatol. Doi:10.1007/ s00704-010-0380x (2011). 32. G. Grell, J. Dudhia and D. Stauffer, A description of the fifth-generation MM5 model, NCAR Technical Note, 1994. 33. J. Dudhia, D. Gill, Y. Guo, D. Hansen, et al., PSU/NCAR MM5 modeling system tutorial class notes, National center for Atmospheric Research, 1998. 34. S.-U. Park, A. Choe, M.-S. Park and E.-H. Lee, Tech Monitor Dec. 2008 24 (2008). 35. M. Wesely and B. Hicks, J. Air Pollut. Control Assoc. 27 (1977) 1110. 36. M. Wesely, J. Atmos. Sci. 43 (1989) 339. 37. J. Seinfeld and S. Pandis, Atmospheric Chemistry and Physics (1998). 38. S.-U. Park, J. Appl. Meteor. 37 (1998) 486. 39. R. Dennis, J. McHenry, W. Barchet, F. Binkowski and D. Byun, Atmos. Env. 27A (1993) 975. 40. C. Walcek and G. A. Taylor, J. Geophy. Res. 108(D20) (1986) 4623. 41. J. Chang et al., J. Geophy. Res. 92 (1987) 14168. 42. National Institute of Meteorological Research, 2010 Annual Report of Hwangsa (Asian Dust) (2010). 43. IPCC, Climate Change 2007: The Physical Science Basis (Cambridge University Press, 2007).

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch05

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


DISTRIBUTION OF BIOGENIC SILICA IN THE UPWELLING ZONES IN THE SOUTH CHINA SEA YANG LIU∗,‡ , MINHAN DAI∗,†,§ , WEIFANG CHEN∗ , and ZHIMIAN CAO∗ ∗State Key Laboratory of Marine Environmental Science, Xiamen University, 361005 Xiamen, China †[email protected] ‡Third Institute of Oceanography, State Oceanic Administration, 361005, Xiamen, China

Biogenic silica (BSi) is an important parameter to understand biogeochemical processes and paleoceanographic records in the ocean. Here we examined the distribution of BSi in the water column in two upwelling regions in the South China Sea (SCS), northwest to the Luzon island in winter and off Vietnam in summer. Also examined were the BSi contents in the sediments collected from the western SCS off Vietnam to investigate if BSi between the water column and sediment is coupled. In winter 2006, BSi concentrations in the water column ranged 0.60–1.83 µmol L−1 in the upwelling zone off Luzon, while the concentration range of 0.01–0.20 µmol L−1 was much lower in the western SCS off Vietnam in non-upwelling seasons. In summer 2007, however, BSi concentrations varied between 0.05 and 0.11 µmol L−1 in the areas off Luzon whereas the concentration significantly increased, up to 2.16 µmol L−1 in the water column off Vietnam. In the western SCS off Vietnam, BSi contents in the sediments displayed a general decreasing trend with increasing water depth, ranging from 2.42 to 4.29 wt.% as SiO2 . BSi in the surface sediments of this area was low and no apparent differences of BSi contents were observed in the surface sediments between the upwelling and non-upwelling zones in the SCS, suggesting decoupling of BSi between the sediment and the overlying water column.

1. Introduction Biogenic silica (BSi) refers to amorphous silica produced by a variety of aquatic organisms, mainly diatoms.1 Given the important role of diatoms playing in the biological pump, BSi is an important parameter to understand biogeochemical processes in the ocean.2–4 Moreover, due to the high §Corresponding

author. 55

July 13, 2012 9:9

AOGS - AS and OS

56

9in x 6in

b1387-ch05

Y. Liu et al.

preservation efficiency, BSi in the sediment can be potentially a useful proxy for paleoproductivity reconstructions.5 Upwelling brings nutrient-rich subsurface waters into the surface. The resultant eutrophic condition may alter the phytoplankton community structure and induce phytoplankton blooms, especially diatoms.6 Li et al.7 demonstrated a significant increase in opal accumulation rates since Middle Pleistocene, which is strongly linked to upwelling induced by the enhanced monsoonal circulation. On the other hand, Zhang et al.8 proposed that the content of BSi in the sediment can only indicate the strength of upwelling in the abyssal sea, whereas the preservation efficiency of BSi in the sediment on the continental shelf is easily disturbed by the sedimentary type and dilution of terrigenous matter. As a consequence, large uncertainties remain as to tracing the upwelling processes by using sedimentary BSi proxy in coastal seas. The South China Sea (SCS) is the world’s largest tropical-subtropical marginal sea where the surface water circulation is strongly driven by the East Asian Monsoon.9 Driven by favorable winds and currents, upwelling frequently occurs in different areas during different periods of time in the SCS.10 A number of studies11−14 have revealed upwellings in the northwestern off Luzon in winter and in the western SCS in summer. These upwellings typically result in enhanced primary productivity.15 For instances, significantly high chlorophyll-a (Chl-a) concentrations were observed in the western SCS off Vietnam in summer,16 while in winter high pigment concentrations covered an area of ∼2.58 × 104 km2 northwestern off Luzon, with a centre located at the position of ∼119◦ E, ∼19.0◦ N.17 In this study, BSi concentrations in the water column were investigated for the first time in both summer and winter, in order to examine the response of BSi production to the upwelling processes. At the same time, the South East Asian Time-series Study (SEATS) station established in the basin area was investigated as a reference station. We note that seasonal variations at station SEATS are relatively small.18 BSi contents in the sediment in the western SCS off Vietnam were also determined to examine the linkage between the water column and the underlying sediment.

2. Material and Methods 2.1. Cruise and sampling Sampling was conducted onboard the R/V Dongfanghong II during two cruises to the SCS. One was surveyed from 30 November to 16 December

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch05

Distribution of BSi in the Upwelling Zones in the South China Sea

57

2006 during which the SCS was largely influenced by the monsoonal wind northeasterly. The other one was conducted during the prevailing southwest monsoon from 29 July to 8 September 2007. During the winter cruise, stations Z98, Z97, Z96, and Z95 were sampled at the upwelling zone off Luzon while stations Yxx and transect M off Vietnam were surveyed in the non-upwelling season. During the summer cruise, samples were collected at stations G04 and H16 off Luzon in non-upwelling seasons while transects Yx0 and Y9 off Vietnam, including stations Y70, Y80, Y90, Y92, Y93, Y94, and Y95 were sampled in the upwelling season. In addition, station SEATS was investigated in both cruises and sediment samples were collected at stations Y10, Y20 and Y90 in summer and at station Y32 off Vietnam in winter (Fig. 1). Seawater samples were collected with a set of 12 Niskin bottles mounted on a rosette sampler. Suspended particle samples for BSi analysis were obtained by filtering ∼4 L of seawater through polycarbonate membranes of 1 µm pore size. The membranes were then dried at 50◦ C overnight and stored in polycarbonate dishes for analysis. BSi samples were collected throughout the upper 100 m at 5, 25, 50, 75, and 100 m layers at most stations. For stations Y22, Y32, M07, Z96, Z97, and SEATS, samples were collected down to 200 m. (b)

20

winter

(c)

Y22

China mainland

Taiwan Island

Y24

Luzon Strait

11 M cross transect

20

M07

Transect Z SEATS

o

25

Y32 Y33 Y34

13

Latitude ( N)

Y44

o

Latitude ( N)

15

18

Z98 Z97 Z96 Z95

16

Hainan Island

114

Longitude ( E) o

(d) Transect Yx0 Y70 Y80 Y93 Y95

o

Latitude ( N)

15

13

Vietnam

Y32 Y90 Y10 Y20

South China Sea

Sulu Sea

11

(a) 5 105

summer 110

112

114

Longitude ( E) o

Indonesia

110

115

120

119

121

Longitude ( E) o

20

10

Transect Y9

117

Philippines

Y90 Y92 Y94 Y10 Y20

9 108

14 115 15

summer

(e) G04

o

112

Latitude ( N)

110

Latitude (oN)

winter 9 108

18

SEATS

H16

16

125

o

Longitude ( E)

14 115

117

119

121

Longitude ( E) o

Fig. 1. (a) Map of the South China Sea showing the sampling area. The star symbols indicate the sampling stations for sediments. Also shown are the sampling stations off (b) Vietnam and (c) Luzon during the winter cruise and off (d) Vietnam and (e) Luzon during the summer cruise.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch05

Y. Liu et al.

58

Surface sediment samples were collected by a box sampler and subsamples were separated onboard and preserved at −80◦C until analysis. 2.2. Analytical methods BSi in the suspended particles was analyzed following the double wetalkaline digestion method.19 Membranes were subjected to a first digestion with 4 mL 0.2 mol L−1 sodium hydrate (NaOH) in a polymethypenten centrifuge tube at 100◦ C for 40 minutes. This digestion dissolves all the BSi and possibly part of the lithogenic silica (LSi).19 Si and Al concentrations ([Si]1 and [Al]1 ) in the supernatant were analyzed by a Technicon AA3 Auto-Analyzer and a Aglient 7500c Inductively Coupled Plasma-Mass Spectrometer, respectively. After rinsing and drying, the membrane was subjected to a second digestion identical to the first one. Si and Al concentrations obtained from this step determine the (Si:Al)2 , which is characteristic of the silicate minerals present in the sample. The corrected BSi concentration is given by the formula: [BSi] = [Si]1 − [Al]1 × (Si:Al)2 .

(1)

Blank and repeatability experiments indicated that the detection limit was about 5 nmol of BSi per membrane and the uncertainty through the entire procedure was below 10%. An optimized wet chemical extraction technique20 was employed to determine the content of BSi in the sediment. After removing the carbonate and organic matter by adding acid and peroxide, sediment samples were leached by 1% sodium carbonate (Na2 CO3 ) for 5 hours at 85◦ C. Aliquots were withdrawn at the 3, 4 and 5 hours and analyzed for dissolved silicate. BSi was generally completely dissolved in 2 hours and LSi continued to be leached throughout the digestion at a slower rate. The amount of BSi was estimated from the intercept of the linear regression line through the time course aliquots.21 The coefficient of variation (e.g., relative standard deviation) for five replicate extractions was 6–7%. Note that we measured four samples allocated by an inter-laboratory comparison experiment20 and our results agreed well with the reference values (Table 1). Depth profiles of Chl-a concentrations were determined on acetoneextracted samples using the standard fluorometric spectrophotometer method.22 The temperature and salinity were measured at each station using either an SBE 911 conductivity-temperature-depth (CTD) profiler or an SBE 917

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Distribution of BSi in the Upwelling Zones in the South China Sea

b1387-ch05

59

Table 1. Biogenic silica (BSi) contents (wt.% as SiO2 ) in four sediment samples from the inter-laboratory comparison experiment. Samples were provided by D. J. Conley.20 BSi content (wt. SiO2 %)

Sample1-Still Pond Sample2-Lewis Lake Sample3-R-64 Smaple4-Yellowstone

Conley20

This study

2.82 ± 1.17 44.3 ± 9.38 6.49 ± 2.06 38.2 ± 9.48

2.09 ± 0.13 38.6 ± 3.56 5.58 ± 0.45 38.4 ± 4.64

plus CTD profiler manufactured by the Sea Bird Corporation as described in Hu et al.23

3. Results and Discussion 3.1. BSi in the water column 3.1.1. The station SEATS At station SEATS, the concentrations of BSi in winter ranged from 0.03 to 0.14 µmol L−1 , with highest values observed in the surface water and at 70 m depth (Fig. 2(a)). BSi concentrations in summer varied in the range of 0.03–0.09 µmol L−1 , which were comparable to that in winter (Fig. 2(d)). A BSi maximum was observed at 75 m depth in summer, corresponding to the Chl-a maximum (data not shown here). Vertical distributions of BSi and Chl-a agreed well at station SEATS in both seasons, suggesting that BSi was mainly produced by the biological activities in the water column at this site. 3.1.2. The upwelling zone off Luzon In winter 2006, the concentrations of BSi in the upwelling zone off Luzon were significantly higher than at station SEATS. Station Z97 was at the core of the upwelling current (Figs. 2(b) and 2(c)), where the highest BSi value of 0.91 µmol L−1 was observed at the surface and the second highest value of 0.85 µmol L−1 was located at 50 m (Fig. 2(a)). At the adjacent stations Z98 and Z96, the highest concentrations of 1.83 µmol L−1 and 1.54 µmol L−1 were observed at 25 m, while the sub-maximum value of 1.51 µmol L−1 and 1.47 µmol L−1 were determined at the surface. In contrast, BSi concentrations below 75 m were relatively low and constant. The average concentration was near 0.70 µmol L−1 and still higher than that

July 13, 2012 9:9

AOGS - AS and OS

BSi (µmol L-1) 0.0

.1

.2 .5

Temperature (oC)

1.0 1.5 2.0 2.5

0

12

150 200

Vietnam

32

100 Y44 SEATS Z97 Z95

150

Luzon

(a)

(b)

200

0.0

.1

.2 .5

33.2 0

33.6

0

12

16

20

24

(d)

200

Depth (m)

34.8

150

Y44 SEATS Z97 Z95

(c) Salinity

28

32

33.2 0

33.6

34.0

34.4

34.8

50

100 150

34.4

200

50

G04 Luzon H16 SEATS Yx0 Y95 Y94 Vietnam Y93 Y92

34.0

100

Temperature (oC)

1.0 1.5 2.0 2.5

50

200

Salinity

28

50

BSi (µmol L-1)

150

24

Depth (m)

Y&M SEATS Z98 Z97 Z96 Z95

Depth (m)

100

100

20

50

50

0

16

H16 SEATS Y10 Y92

(e)

Depth (m)

0

Depth (m)

b1387-ch05

Y. Liu et al.

60

Depth (m)

9in x 6in

100 150 200

H16 SEATS Y10 Y92

(f)

Fig. 2. Vertical distributions of (a, d) BSi, (b, e) Temperature, and (c, f) Salinity in the upper water column of the upwelling zones off Luzon and Vietnam and at station SEATS in the South China Sea in winter (upper panel) and in summer (lower panel).

at the reference station of SEATS. The distribution pattern of BSi at station Z95 was similar to station SEATS. The highest value of 1.27 µmol L−1 was observed at 50 m and the concentrations at the surface and 100 m layer were relatively low (Fig. 2(a)). Chl-a concentration was highest (0.62 µg L−1 ) at 30 m layer at station Z97, while Chl-a maximum layers existed at the surface (0.50 µg L−1 ) at station Z98 and at 50 m (0.86 µg L−1 ) at station Z96, respectively. The depth of Chl-a maximum (DCM) agreed well with the BSi maximum layer at station Z95. It is interesting that the Chl-a concentration off the Luzon upwelling region did not show significant enhancement as compared to the reference station of SEATS where the maximum value of Chl-a was 0.45 µg L−1 . Note that the diagnostic pigment of diatom (fucoxanthin) varied more distinctly (3-fold) than total Chl-a (Huang et al., unpublished data). In summer 2007, the overall BSi concentrations at stations G04 and H16 were comparable to those at station SEATS (Fig. 2(d)). BSi concentrations increased generally with depth at station G04, varying from 0.08 µmol L−1 at the surface to 0.13 µmol L−1 at 100 m. The BSi concentrations at station

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Distribution of BSi in the Upwelling Zones in the South China Sea

b1387-ch05

61

H16 were lower than station G04, the highest value of which (0.11 µmol L−1 ) was observed at 75 m where the DCM was located (Fig. 2(d)). 3.1.3. The upwelling zone off Vietnam Influenced by the coastal upwelling in summer and mesoscale eddies (Fig. 2(e) and 2(f)),23 a significant increase of BSi was observed in the area off Vietnam. BSi concentrations in the entire water column at most of the stations in this area were higher than 0.20 µmol L−1 (Fig. 2(b)). Except at station Y93, the maximum of BSi appeared at 25 m along transect Y9, with the highest value of 2.16 µmol L−1 at station Y92. Vertical distributions of BSi at station Y93 displayed a general decreasing trend, with a surface maximum of 2.08 µmol L−1 . The high BSi concentrations in the top 25 m water along this cross-shore transect were caused by the upwelling of coastal jet.23 However, BSi concentrations along the nearshore along-shore transect were relatively lower, with the highest value observed at 25 m of station Y80. The maximum of BSi concentrations at station Y70 (0.64 µmo L−1 ) and at station Y90 (0.61 µmol L−1 ) were both observed at 50 m depth. BSi concentrations below 50 m decreased but they were still higher than or comparable to that at the DCM of the station SEATS (Fig. 2(b)). The Chl-a concentration had a 3-fold increase in the area off Vietnam in the summer, and different from the reference station, nearly all of the maximum value of Chl-a appeared in the upper 50 m layer. At the same time, the proportion of fucoxanthin had an average 5-fold increase (Huang et al., unpublished data), consistent with the increase in BSi. The profiles of BSi at the stations off Vietnam in winter were overall similar to the reference station of SEATS. BSi concentrations in the surface waters ranged from 0.01 to 0.11 µmol L−1 , with an average value of 0.06 µmol L−1 (Fig. 2(a)). At 100 m, the BSi concentrations varied in a range of 0.02–0.13 µmol L−1 with an average value of 0.08 µmol L−1 . 3.2. The inventory of BSi in the SCS BSi concentrations were significantly enhanced during the upwelling seasons, off Luzon in winter and off Vietnam in summer. This is particularly true at the layer of the DCM in the upwelling centers where BSi was at least one order of magnitude higher than that at the reference station of SEATS. It is interesting to note that BSi in the upwelling zones in the SCS were lower than in other typical upwelling zones such as at Californian coast where the BSi concentration is as high as 42 µmol L−1 .24

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch05

Y. Liu et al.

62

Table 2. The inventory of BSi in the upper 100 m water column of upwelling zones off Luzon and Vietnam and at station SEATS in the South China Sea. Area

Stations

Average BSi inventory [mmol m−2 ]

Winter Vietnam Luzon

SEATS Stations Yxx and Transect M Transect Z

6.28 7.48 (n = 13) 97.27 (n = 4)

Summer Luzon Vietnam

SEATS G04 and H16 Transect Yx0 Transect Y9

5.59 7.71 35.15 (n = 3) 65.79 (n = 4)

For the upwelling stations off Luzon in winter, the depth-integrated BSi inventory in the upper 100 m varied between 74.44 and 120.86 mmol m−2 . Its average value of 97.27 mmol m−2 was slightly higher than that of 52.66 mmol m−2 determined for the upwelling stations off Vietnam in summer where the BSi inventory ranged from 24.04 to 88.80 mmol m−2 (Table 2). BSi inventory in the upper 100 m waters displayed remarkable enhancement by ∼13-fold at the upwelling centers off Luzon as compared to surrounding waters in winter, while in the upwelling zone off Vietnam in summer it exhibited a modest increase of up to 9-fold higher than the reference stations off Luzon. 3.3. BSi in the sediment As shown in Fig. 3, BSi contents in the sediment at station Y32 ranged from 2.42% to 4.29% and displayed a general decreasing trend from the surface. Exception occurred to the 10–12 cm layer where BSi were slightly higher. BSi contents were 4.12 ± 0.10% in the surface sediment at station Y90 (Fig. 3), while the distribution patterns of BSi in the sediment at stations Y20 (Fig. 3) and Y10 (Fig. 3) showed minor variations, within the range of 2.40–3.36% and 3.12–3.40%, respectively. Jia et al.25 and Chen26 have also reported similar BSi contents in the sediments at the adjacent area (Table 3). All these data, however, indicate no significant differences of BSi contents in the sediments between the upwelling zone off Vietnam and other non-upwelling areas of SCS. Note that BSi contents in the sediment of the Luzon upwelling zone are not high either.27

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch05

Distribution of BSi in the Upwelling Zones in the South China Sea

63

BSi (wt. SiO2%) 0

1

2

3

4

5

6

0 2 4

Depth (cm)

6 8 10 12 Y32 Y20 Y10 Y90

14 16 18

Fig. 3. The content of BSi (wt. SiO2 %) in the sediment at stations Y32, Y20, Y10, and Y90 in the western South China Sea.

Table 3. Summary of the content of BSi (SiO2 wt. %) in the sediment in different areas of the South China Sea. Area

Lat. [◦ N]

Region off Vietnam

11.5−13.4 ∼110.0 11.0 110.0 10.0−11.0 109.0−111.0 7.2 112 12.0 110.0−113.0 17.0−20.0 120.0 ∼18.0 ∼116.0

Luzon basin

Lon. [◦ E]

BSi [SiO2 wt. %] 2.42−4.29 3.30 3.62−5.28 3.38 3.8 − 5.5 2.5 − 4.8 4.3 − 6.2

References This study Zhou et al.27 Zhang et al.8 Jia et al.25 Chen26 Chen26 Chen26

3.4. Water column versus sediment Although BSi concentrations were nearly one order of magnitude higher in the upper water column of the upwelling zones in the SCS, high BSi content was not observed in the underlying sediments. Reasons for this decoupling might be related to the terrigenous material dilution, lateral transport, dissolution during the particle sinking or perhaps the combination of all these processes.26,28,29 Such decoupling of BSi between the sediment and the

July 13, 2012 9:9

AOGS - AS and OS

64

9in x 6in

b1387-ch05

Y. Liu et al.

upper water column suggest that validation is mandatory before sediment BSi can be used to reconstruct the paleoproductivity in the coastal ocean.

4. Concluding Remarks BSi in the water column was nearly one order of magnitude higher in the Luzon upwelling zone than in the non-upwelling zone in winter, while BSi concentrations in the region off Vietnam upwelling zone in summer were also significantly higher than at the reference station of SEATS. However, the coupling of BSi between sediment and water column in those two upwelling zones was not observed in this study. The linkage of biogeochemical processes in the water column and the record in the underlying sediment merits further examination and validation. Acknowledgments This study was supported by the National Science Foundation of China (Grant No. 40521003). We thank Jian-yu Hu for providing the hydrological data. Bang-qin Huang is thanked for sharing the chlorophyll-a data. The crew of R/V Dongfanghong II provided much help during the cruises. We are grateful to the help during the sampling and analyzing from Jin-Yu Yang, Kuan-Bo Zhou, and Qian Liu.

References 1. D. M. Nelson, P. Tr´eguer, M. A. Brzezinski, A. Leynaert and B. Qu´eguiner, Global Biogeochem. Cycles 9 (1995) 359–372. 2. S. Honjo, Oceanus 40 (1997) 4–7. 3. R. A. Mortlock and P. N. Froelich, Deep-Sea Res. I 36 (1989) 1415–1426. 4. M. Leinen, Marine Micropaleontology 9 (1985) 375–383. 5. P. Tr´eguer, D. M. Nelson, A. J. Van Bennekom, D.J. DeMaster, A. Leynaert and B. Qu´eguiner, Science 268 (1995) 375–379. 6. R. Jyothibabu, C. R. Asha Devi, N. V. Madhu, P. Sabu, K. V. Jayalakshmy, Josia Jacoba, H. Habeebrehman, M. P. Prabhakaran, T. Balasubramaniana and K. K. C. Nair, Con. Shelf Res. 28 (2008) 653–671. 7. J. R. Li, R. J. Wang and B. H. Li, Chinese Sci. Bull. 47 (2002) 235–237. 8. L. L. Zhang, M. H. Chen, R. Xiang, J. Lu and L. L. Zhang, J. Trop. Oceanogr. 26 (2007) 24–29. 9. P. T. Shaw and S. Y. Chao, Deep-Sea Res. I 41 (1994) 1663–1683. 10. K. K. Liu, S. Y. Chao, P. T. Shaw, G. C. Gong, C. C. Chen and T. Y. Tang, Deep-Sea Res. I 49 (2002) 1387–1412.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Distribution of BSi in the Upwelling Zones in the South China Sea

b1387-ch05

65

11. S. Y. Chao, P. T. Shaw and S. Y. Wu, Prog. Oceanogr. 38 (1996) 51–93. 12. P. T. Shaw, S. Y. Chao, K. K. Liu, S. C. Pai and C. T. Liu, J. Geophys. Res. 101 (1996) 16435–16448. 13. M. J. B. Udarbe-Walker and C.L. Villanoy, Deep Sea Res. I 48 (2001) 1499–1518. 14. R. S. Wu and L. Li, J. Oceanogr. Taiwan Strait 22 (2003) 269–277. 15. S. H. Shen, G. G. Leptoukh, J. G. Acker, Z. J. Yu and S. J. Kempler, IEEE Geoscience and Remote Sensing Letters 5 (2008) 315–319. 16. T. H. Zhang, H. G. Zhan and C. Q. Chen, J Trop. Oceanogr. 26 (2007) 9–14. 17. J. J. Wang, D. L. Tang, and Y. Sui, J. Mar. Syst. 83 (2010) 141–149. 18. T. F. G. Wong, T. L. Ku, M. Mulholland, C. M. Tseng and D. P. Wang, Deep-Sea Res. II 54 (2007) 1434–1447. 19. O. Ragueneaua, N. Savoye, Y. D. Amo, J. Cotten, B. Tardiveau and A. Leynaert, Cont. Shelf Res. 25 (2005) 697–710. 20. D. J. Conley, Mar. Chem. 63 (1998) 39–48. 21. D. J. DeMaster, Geochimica et Cosmochimica Acta 45 (1981) 1715–1732. 22. J. D. H. Strickland and T. R. Parsons, Bulletin of Fisheries Research Board Canada 167 (1972) 1–310. 23. J. Y. Hu, J. P. Gan, Z. Y. Sun, J. Zhu and M. H. Dai, J. Geophys. Res. 116 C05016 (2011), doi:10.1029/2010JC006810. 24. G. F. Firme, E. L. Rue, D. A. Weeks, K. W. Bruland and D. A. Hutchins, Global Biogeochem. Cycles 17 (2003) 1016–1028. 25. G. D. Jia, Z. M. Jian, P. A. Peng, P. X. Wang and J. M. Fu, Geochimica 29 (2000) 293–296. 26. J. F. Chen, Biogeochemistry of Settling Particles in the South China Sea and its Significance for Paleo-environment Studies. PhD thesis, Tongji University (2005). 27. P. Zhou, D. M. Li, G. S. Liu, W. Men and L. H. JI, J. Trop. Oceanogr. 29 (2010) 40–47. 28. C. Hwang and S. Chen, J. Geophys. Res. 105 C10 (2000) 23943–23965. 29. X. Ning, F. Chai, H. Xue, Y. Chai, C. Liu and J. Shi. J. Geophys. Res. 109 C10005 (2004), doi: 10.1029/2004JC002365.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

This page intentionally left blank

b1387-ch05

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch06

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


INFLUENCE OF VELOCITY DISTRIBUTION AND DENSITY STRATIFICATION ON GENERATION OR PROPAGATION OF TSUNAMIS∗ TARO KAKINUMA and KEI YAMASHITA Graduate School of Science and Engineering, Kagoshima University, 1-21-40 Korimoto, Kagoshima, Kagoshima 890-0065, Japan KEISUKE NAKAYAMA Department of Civil and Environmental Engineering, Kitami Institute of Technology, 165 Koen-cho, Kitami, Hokkaido 090-8507, Japan

A set of wave equations derived on the basis of a variational principle in consideration of both strong nonlinearity and strong dispersion of surface/internal waves is numerically solved to simulate generation and propagation of tsunamis in the vertical two-dimension. The velocity potential in each fluid layer is expanded into a power series of vertical position, such that the accuracy of vertical distribution of velocity depends on the number of expansion terms. Numerical results of surface displacement are compared with the existing experimental data, where tsunamis are generated by the seabed uplift. When the fundamental equations are reduced to nonlinear shallow water equations, the numerical model cannot represent propagation of a long wave group especially in distant-tsunami propagation, leading to overestimation of both the wave height and wave steepness of the first wave. The wave height becomes larger in the stratified ocean than that in a one-layer case, although the present density distribution hardly affects the tsunami phase over a long-distance travel.

1. Introduction Numerical models based on shallow water equations, where the effects of both vertical distribution of horizontal velocity and vertical velocity are neglected, are usually used for calculation of tsunamis due to submarine earthquakes. Moreover, the effects of stratification in the ocean are not ∗ This work is supported by Grant-in-Aid for Scientific Research (C) (21560544) of The Ministry of Education, Culture, Sports, Science, and Technology, Japan.

67

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

68

considered in tsunami simulation. In this study, nonlinear surface/internal wave equations derived through a variational principle1 are numerically solved to evaluate these effects on generation or propagation of tsunamis. For example, Horrillo et al.2 and Iwase3 applied numerical models based on the Navier–Stokes and Boussinesq-type equations, respectively, to compare numerical results with those through shallow water models especially for distant-tsunami propagation. In the present study, the velocity potential is expanded into a power series on vertical position, which means that the dispersion, as well as the nonlinearity, of waves depends on the number of expansion terms of the velocity potential, i.e., the assumed vertical distribution of velocity. The numerical model tries to find solutions nearest to the true values based on the variational principle, such that the influence of velocity distribution on generation and propagation of tsunamis is studied by changing the number of expansion terms of the velocity potential. Computation of tsunamis is performed in the vertical two-dimension. Surface displacements are compared between numerical results and the existing experimental data,4 where tsunamis are generated due to seabed uplift. Wave height and phase of tsunamis after a travel of a long distance are compared for various numbers of expansion terms of velocity potential. Tsunamis propagating over a continental slope or in two-layer water are also numerically simulated. 2. Governing Equations and Numerical Method Inviscid and incompressible fluids, as shown in Fig. 1, are assumed to be stable in still water, where the fluids are represented as i (i = 1, 2, . . . , I) from top to bottom. The i-layer thickness in still water is denoted by hi (x). None of the fluids mix even in motion and the density ρi (ρ1 < ρ2 < · · · < ρI ) is spatially uniform and temporally constant in each layer. Surface tension and capillary action are neglected. Fluid motion is assumed to be irrotational, resulting in existence of velocity potential φi defined

Fig. 1.

Multilayer-fluid system.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Influence of Velocity Distribution and Density Stratification

b1387-ch06

69

as ui = ∇φi and wi = ∂φi /∂z, where ∇ = (∂/∂x, ∂/∂y), i.e., a partial differential operator in the horizontal plane. The pressure on z = ηi,0 , i.e., the lower interface of the i-layer, is written by pi (x, t). In the i-layer, if both the elevation of one interface, z = ηi,1−j (x, t) (j = 0 or 1), and the pressure on the other interface, pi−j (x, t), are known, then the unknown variables are the velocity potential φi (x, z, t) and interface elevation ηi,j (x, t), such that the functional for the variational problem1 is determined in the i-layer by
t1


ηi,1  ∂φi Fi [φi , ηi,j ] = ∂t t0 A ηi,0   2 ∂φ + P 1 1 p i i−j i + (∇φi )2 + + gz + dz dA dt, 2 2 ∂z ρi (1) i−1

where Pi = k=1 (ρi − ρk )ghk ; the gravitational acceleration g is equal to 9.8 m/s2 . The velocity potential in the i-layer, φi , is expanded into a power series of vertical position above the top face, i.e., the free surface, of still water, z, as φi (x, z, t) =

N −1 

{fi,α (x, t) · z α } ≡ fi,α z α ,

(2)

α=0

where, for instance, f2,3 means the weighting of z 3 in the 2-layer and the sum rule of product is adopted for subscript α in the right-hand side. We substitute Eq. (2) into Eq. (1), after which the functional Fi is integrated vertically; then the variational principle is applied to obtain Euler–Lagrange equations, that is the fully nonlinear equations for surface and internal waves. If the number of water layers is equal to two, the equations are [Upper layer] (i = 1) ζα

1 ∂ζ ∂η − ηα + ∇{(ζ α+β+1 − η α+β+1 )∇f1,β } ∂t ∂t α+β+1 −

ζβ

αβ (ζ α+β−1 − η α+β−1 )f1,β = 0, α+β−1

1 ∂f1,β 1 + ζ β+γ ∇f1,β ∇f1,γ + βγζ β+γ−2 f1,β f1,γ + gζ = 0, ∂t 2 2

(3) (4)

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

70

[Lower layer] (i = 2) ηα

1 ∂η ∂b − bα + ∇{(η α+β+1 − bα+β+1 )∇f2,β } ∂t ∂t α + β + 1 −

ηβ

αβ (η α+β−1 − bα+β−1 )f2,β = 0, α+β−1

1 ∂f2,β 1 + η β+γ ∇f2,β ∇f2,γ + βγη β+γ−2 f2,β f2,γ + g(η + h1 ) ∂t 2 2  ρ1 1 ∂f 1,β − ηβ + η β+γ ∇f1,β ∇f1,γ ρ2 ∂t 2  1 β+γ−2 + βγη f1,β f1,γ + g(η + h1 ) = 0, 2

(5)

(6)

where the surface, interface, and seabed are described by z = η1,1 ≡ ζ(x, t), z = η1,0 = η2,1 ≡ η(x, t), and z = η2,0 ≡ b(x, t), respectively; h1 is the thickness of the top layer in still water. In this paper, the seabed friction, fluid viscosity, and Coriolis force are not considered for simplicity. The governing equations are rewritten to a set of finite difference equations and the time development is carried out by applying implicit schemes similar to that of Nakayama and Kakinuma.5 The one-layer model without internal waves is utilized except in Sec. 6, where the influence of density stratification on propagation of tsunamis is studied. 3. Influence of Velocity Distribution on Generation of Tsunamis Numerical results of water surface displacement are compared with the experimental data4 for verification of the present model. The initial water depth h0 , as well as the width of water basin, is uniform. There is a vertical wall at the place x = 0 m, i.e., one end of the water basin. The bottom through the area 0 m ≤ x ≤ 0.305 m is raised vertically. The time variation of seabed position b(t) is given by b(t) = b0 (1 − e−εt )− h0 when t > 0, where b0 = 0.005 m, ε = 25.84 s−1 , and h0 = 0.025 m; the unit of time is second. As shown in Fig. 2, when the number of expansion terms of velocity potential shown in Eq. (2), i.e., N , is equal to three, the time variation of water surface displacement at the end where x = 0 m is well evaluated in comparison with the experimental data, as well as the three-dimensional calculation results,6 where the number of cells along the vertical direction is 21 and the seabed shift is described by temporal change of porosity

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Influence of Velocity Distribution and Density Stratification

b1387-ch06

71

Fig. 2. Water surface displacement ζ (x = 0 m; ◦: experiment by Hammack,4 •: threedimensional numerical model,6 —: the present model, where N = 3).

Fig. 3. Water surface displacement ζ through the present model (x = 0 m; —: N = 1, : N = 2, •: N = 3).

inside each cell in the neighborhood of the seabed, while the water surface displacement is evaluated by solving an advection equation of a VOF function. Numerical results of water surface displacement in the same condition as that of Fig. 2 are shown in Fig. 3 for the cases where the number of expansion terms of velocity potential, N , is equal to one, two, and three. According to Fig. 3, the shorter waves in the tsunami tail are not reproduced when N = 1, i.e., the set of governing equations is reduced to a set of nonlinear shallow water equations without wave dispersion. When N = 2, the model takes into account both linear and uniform vertical distributions of horizontal velocity ui and vertical velocity wi , respectively, such that the balance between nonlinearity and dispersion is considered, leading to the more accurate result than that when N = 1. When N = 3, the effects of both parabolic vertical distribution of ui and linear vertical distribution of wi are also considered.

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

72

4. Influence of Velocity Distribution on Propagation of Tsunamis In a calculation domain, the still water depth is equal to 4,000 m and a vertical wall exists at one end where x = 0 km. The initial water surface displacement is given by ζ(x, 0) = a0 {1 + cos[2π(x/B)]}(0 km ≤ x ≤ B/2), where a0 = 1.0 m for every case and B means the initial wavelength. Water surface profiles are shown in Fig. 4, where B = 20 km. When N = 1, the front-surface slope and crest height of the first wave are larger without wave-group generation than those when N = 3, respectively, and the peak time, i.e., the arrival time of the first crest, is too early when N = 1. The vertical distribution of horizontal velocity u below the first crest is shown in Fig. 5, where t = 1,000 s. When N = 1, the horizontal velocity u is quite different from that when N = 2 and 3. According to the present results, the number of expansion terms of the velocity potential should be more than or equal to two in consideration of the balance between nonlinearity and dispersion of waves.

Fig. 4.

Fig. 5.

Water surface profile at each time (B = 20 km; N = 1 or 3).

Horizontal velocity u below the first crest (B = 20 km; t = 1,000 s; N = 1, 2, or 3).

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

Influence of Velocity Distribution and Density Stratification

Fig. 6.

73

Water surface profile (B = 30 or 60 km; t = 45,000 s; N = 1 or 2).

Fig. 7. Horizontal distance of position between the first crests when N = 1 and 2 (B = 20, 30, or 60 km).

Water surface profiles are shown in Fig. 6, where B = 30 and 60 km; t = 45,000 s. In the figure, the results only for the cases N = 1 and N = 2 are shown based on Figs. 3 and 5, where the results for the cases N = 2 and N = 3 are not much different. When N = 2, the wave group consists of many waves; it should be noted that the shorter the initial wavelength B is, the longer the total length of the wave group is at this time. The horizontal distance of position, ∆xcrest , and the difference of height, ∆ζcrest , between the first crests when N = 1 and 2 are shown in Figs. 7 and 8, respectively, where B = 20, 30, or 60 km. According to Fig. 7, ∆xcrest is more than 30 km when t = 45,000 s. As shown in Fig. 8, the larger the initial wavelength B is, the larger the difference ∆ζcrest is after t = 13,500 s. The wave height of the first wave, which includes the first crest and the first trough, H, and the horizontal distance between the first crest and

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

74

Fig. 8. Difference of height between the first crests when N = 1 and 2 (B = 20, 30, or 60 km).

Fig. 9.

Wave height of the first wave (B = 20, 30, or 60 km; N = 1 or 2).

the first trough, xcrest –xtrough , are shown in Figs. 9 and 10, respectively, where B = 20, 30, or 60 km and N = 1 or 2. It should be noted that the wavelength of the first wave does not depend on the initial wavelength but the propagating time. According to these figures on the first wave, the wave steepness becomes larger as the initial wavelength is longer, i.e., the total initial potential energy is larger. 5. Tsunamis Traveling over a Continental Slope In a model of seabed configuration, two areas, where the still water depth is 4,000 and 200 m, are connected with a continental slope, the horizontal length of which is 45 km. The tsunamis simulated in Sec. 4 are assumed to

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch06

Influence of Velocity Distribution and Density Stratification

75

Fig. 10. Horizontal distance between the first crest and the first trough (B = 20, 30, or 60 km; N = 1 or 2).

Fig. 11.

Water surface profile at each time (B = 60 km; N = 1 or 2).

continue propagation in the former area to reach the continental slope lying through the area 9,200 km
x
9,245 km, after which they travel over it and then the latter area, i.e., the continental shelf. Computation results of water surface profiles are shown in Fig. 11, where B = 60 km; N = 1 and 2. The first wave, which has inclined backward due to wave dispersion over the flat seabed, begins to lean forward over

July 13, 2012 9:9

76

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

the continental slope, increasing its wave height. In the present case, the horizontal distance between the first crests when N = 1 and 2, ∆xcrest , is around 34 km just before the tsunamis reach the continental slope, after that ∆xcrest decreases to about 6 km over the continental slope.

6. Influence of Density Stratification on Propagation of Tsunamis It is assumed that the ocean where tsunamis propagate has uniform twolayer stratification over uniform seabed configuration to study the influence of density distribution on tsunami propagation. The still water depth is 150 and 3,850 m and the water density is 1,020.26 and 1,023.26 kg/m3 in the upper and lower layers, respectively. On the other hand, in the case of onelayer water for comparison, the still water depth is 4,000 m. In these cases, a vertical wall is set at one end, where x = 0 km, of the calculation domain and a part of the seabed, where 0 km ≤ x ≤ 15 km, is raised at a uniform upward velocity to the uplift height of 2 m within 0 s ≤ t ≤ 20 s to simulate tsunami generation and propagation in one- or two-layer ocean. Numerical results of water surface profiles are shown in Fig. 12, where t = 14,000 s and N = 3. According to the figure, the wave height of the first wave in the two-layer ocean is larger than that in the one-layer water by 13.4%. There is, however, almost no difference between the horizontal positions of the corresponding wave crests, as well as the corresponding wave troughs, for the one- and two-layer cases, which means that even distanttsunamis hardly feel the effect of density stratification on their phases.

Fig. 12.

Water surface profile of tsunamis in one- or two-layer water (t = 14,000 s; N =3).

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Influence of Velocity Distribution and Density Stratification

b1387-ch06

77

7. Conclusions The set of surface/internal wave equations derived on the basis of the variational principle was numerically solved to simulate generation and propagation of tsunamis in the vertical two-dimension. The velocity potential in each fluid layer was expanded into the power series on vertical position, such that the influence of velocity distribution on generation and propagation of tsunamis was studied by changing the number of expansion terms of the velocity potential. According to the present results, the number of expansion terms of the velocity potential should be more than one. When the number of expansion terms of the velocity potential is two, both the horizontal and vertical velocities distributed linearly and uniformly, respectively, in the vertical direction are considered with the balance between nonlinearity and dispersion of waves. If the fundamental equations are reduced to nonlinear shallow water equations, then the numerical model has disadvantage as follows: • • • • •

Shorter waves in a tsunami tail are not reproduced in the case where the velocity or acceleration of seabed uplift is larger. The tsunami height is too large and the wavelength is too short, leading to overestimation of the wave steepness. The peak time is too early, although the error decreases during the tsunami propagation with forward inclination over a continental slope. A long wave group, which consists of many waves, is not represented especially in distant-tsunami propagation. The phenomenon where the wavelength of the first wave is independent of the initial wavelength after a long-distance travel does not appear.

In the stratified ocean, the wave height becomes larger than that in the one-layer water when the friction at the interface can be neglected. The present density distribution hardly affects the distant-tsunami phase, which means that there is some other reason to predict a peak time wrong in actual cases of tsunamis. Acknowledgments Sincere gratitude is extended to Prof. K. Fujima, National Defense Academy of Japan, for beneficial comments especially on propagation of tsunamis in stratified water.

July 13, 2012 9:9

78

AOGS - AS and OS

9in x 6in

b1387-ch06

T. Kakinuma et al.

References 1. T. Kakinuma, Proc. 5th Int. Conf. on Computer Modelling of Seas and Coastal Regions, WIT Press (2001) 225. 2. J. Horrillo, Z. Kowalik and Y. Shigihara, Marine Geodesy 29 (2006) 149. 3. H. Iwase, Doctoral Dissertation, Tohoku Univ. (2005). 4. J. L. Hammack, J. Fluid Mech. 60 (1973) 769. 5. K. Nakayama and T. Kakinuma, Int. J. Numer. Meth. Fluids 62 (2010) 574. 6. T. Kakinuma and M. Akiyama, Proc. 30th Int. Conf. on Coastal Eng., ASCE (2007) 1490.

July 19, 2012 11:30

AOGS - AS and OS

9in x 6in

b1387-ch07

Advances in Geosciences Vol. 28: Atmospheric Science and Ocean Sciences (2011) Eds. Chun-Chieh Wu and Jianping Gan c World Scientific Publishing Company


SEA SURFACE pCO2 IN THE INDIAN SECTOR OF THE SOUTHERN OCEAN DURING AUSTRAL SUMMER OF 2009 SUHAS SHETYE National Centre for Antarctic & Ocean Research, Headland Sada, Goa-403 804, India MARUTHADU SUDHAKAR Ministry of Earth Sciences, Prithvi Bhavan New Delhi-110 003, India RENGASWAMY RAMESH Physical Research Laboratory, Navrangpura Ahmedabad-380 009, India RAHUL MOHAN and SHRAMIK PATIL National Centre for Antarctic & Ocean Research, Headland Sada, Goa-403 804, India AMZAD LASKAR Physical Research Laboratory, Navrangpura Ahmedabad-380 009, India

This Southern Ocean plays a key role in removing carbon dioxide from the Earth’s atmosphere by physical, chemical, and biological processes. The present study attempts to understand: the spatio-temporal variations in pCO2 and its relationship with nutrients and biological production in the Indian sector of the Southern Ocean during the late austral summer of 2009. The partial pressure of carbon dioxide (pCO2 ) showed high spatio-temporal variability in the study area. The highest pCO2 that was recorded along the Polar Front (PF) the two transects is attributed to low productivity in the PF. From 57◦ 30’E(TE ) towards 48◦ E, the average sea surface pCO2 , chlorophyll and Total organic carbon (TOC) increased by 24 µatm, 0.3 mg/m3 , and 3 µM, respectively, suggesting that the physical processes are predominantly active along 48◦ E. Enhanced vertical mixing along 48◦ E supports the corresponding increase in the average NO3 , PO4 , and SiO4 concentrations by 2 µM, 0.4 µM, and 1.7 µM, respectively. pCO2 and chlorophyll a are negative correlated along 57◦ 30’E(TE ), however, positively correlated along 48◦ E’(TW ), which suggests that the biological processes control the pCO2 along 57◦ 30’E. The average air– sea fluxes recorded were about −28 and −33 mmol m−2 d−1 , on TW and TE ,

79

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

b1387-ch07

S. Shetye et al.

80

respectively. A significant finding of this study is that although the Southern Ocean is a known sink of carbon dioxide, the vicinity of the Crozet Island, where oceanic fronts are known to merge, suggests to act as a source of atmospheric CO2 . It is attributed that “The island mass effect” could also be a factor that generates elevated CO2 in the vicinity of the study area. In the last one decade the oceanic pCO2 increased at a rate 0.77 µatm/year in the region south of the Polar front; but is not associated with the Southern Annular Mode effect.

1. Introduction The rapid rise of carbon dioxide (CO2 ) (∼1.9 ppm/year9 in the atmosphere is one of the major environmental concerns because of its impending effects on the global climate.43 Climate change is predicted to impact ocean biology and physics and drive both positive and negative feedbacks, which may strongly influence the global carbon budget.8 Expected global warming and associated environmental changes including sea-level rise would severely affect the socio-economic stability of human society and global terrestrial– marine ecosystems.43 Since the variation of oceanic surface pCO2 is greater than the seasonal variability of atmospheric pCO2 , the direction and magnitude of the CO2 flux through the sea–air interface are regulated primarily by the ocean.17 At present, the Southern Ocean accounts for ∼40% of the oceanic sink of anthropogenic CO2 .37 The carbon uptake capacity of the ocean seems to have reduced recently.37,9 Of particular concern are climate interactions that could reduce the efficiency of the carbon sink, a process that is already occurring in the Southern Ocean,23,27 the sampling of which has been both temporally and spatially sparse owing to its remoteness, difficulty and cost of making such measurements.33 Metzl31 showed that the growth rate of oceanic pCO2 is higher in the subtropical and sub-Antarctic zones of the Southern Indian Ocean compared to other oceans. Areas west of the Crozet Plateau are the key regions where the fronts confluence and split.5,35 The biological productivity of waters surrounding oceanic islands is known to be significantly higher than that of offshore waters,14 a phenomenon commonly referred to as the “Island Mass Effect”. Sander38 discussed the probable causes of this enhanced biological productivity and attributed it principally to inorganic micronutrient enrichment. The sources of nutrients could be land runoff, benthic regeneration, and the breaking of internal waves in stratified waters at the island shelf, bringing about the mixing of nutrient-rich deep waters with the nutrient-deprived surface waters. We report here new data on Partial pressure of carbon dioxide

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Sea Surface pCO2 in the Indian Sector of the Southern Ocean

b1387-ch07

81

(pCO2 ) along with other parameters such as pH and macro-nutrients from the Southern Ocean during the late austral summer of 2009.

2. Methodology 2.1. Sampling details Surface seawater and air sampling was carried out onboard R/V Akademik Boris Petrov along the two transects, TW (30◦ to 65◦ S along 48◦ E) and TE (30◦ S to 66◦ S along 57◦ 30’E) during February–March 2009 in the Indian sector of Southern Ocean (Fig. 1). Samples were collected at standard depths (0, 10, 20, 30, 40, 50, 60, 80, 100, 120, 150, 175, and 200) upto 200 m along the two transects using a Rosette sampler with 5 L Niskin bottles were mounted on the CTD assembly. Sea surface temperature (SST) was also recorded using an onboard bucket thermometer (accuracy: ±0.1◦ C). Salinity was measured with the help of an Autosal. The error between Salinity measured by CTD and Autosal was ±0.1%.

2.2. pCO2 calculation 2.2.1. Partial pressure of CO2 in seawater (pCOsw 2 ) For Total carbon dioxide (TCO2 ) content of seawater samples, a subsample was acidified with 8.5% phosphoric acid and bubbled through with nitrogen. Gaseous CO2 was captured in an ethanol-amine solution with an indicator. The solution was photometrically backtitrated by a coulometer (model 5014 of U.I.C. Inc., USA). The reliability of the coulometric titration was regularly checked with certified referenced materials (CRMs, Batch #92) provided by A. Dickson (SIO, University of California). The accuracy estimated from the CRMs values was 2 µM. The precision estimated from replicate analysis of samples (mean difference) was on average 1.5 µM. The pH was measured at 25◦ C with an Orion pH meter on the total hydrogen ion concentration scale (pHT ), using tris-buffer prepared according to Goyet and Dickson.13 The pHT of the samples was then corrected to the in situ temperature following equation of Gieskes.19 Analytical precision was ∼0.002 for pH. The seawater pCO2 was calculated with TCO2 and pHT using the program of Lewis and Wallace26 with the dissociation constants of Mehrbach et al.32 as refit by Dickson and Millero.12 The precision estimated for pCO2 analysis from replicate analysis of samples was ±5 µatm.

July 13, 2012 9:9

82

AOGS - AS and OS

9in x 6in

b1387-ch07

S. Shetye et al.

Fig. 1. The research area between 30–66◦ S along 48◦ E and 57◦ 30’E in the Indian Sector of Southern Ocean. Black circles indicate the station locations. Different fronts have been marked along with Crozet and Kerguelen Island.

2.2.2. Partial pressure of CO2 in air (pCOatm 2 ) The Air samples were collected by means of 1 L Pyrex flasks. The flasks were evacuated using a rotary pump and then opened for ∼2 minutes on the windward side of the ship, about 15 m above sea level. After the sample collection, the flasks were closed by means of high vacuum stopcocks and stored until the analysis 2–3 months later. The quantitative separation of the CO2 from air was carried out in the laboratory by pumping the flask air samples in a high vacuum line at a rate of about ∼5 mL/min through a digital mass flow controller and a spiral Pyrex trap cooled at liquid nitrogen temperature (−196◦C). After the completion of the air pumping, the Pyrex trap was isolated by means of high vacuum stopcocks and heated to about −80◦ C by means of an ethyl alcohol–liquid nitrogen slash. The evolved CO2

July 13, 2012 9:9

AOGS - AS and OS

9in x 6in

Sea Surface pCO2 in the Indian Sector of the Southern Ocean

b1387-ch07

83

was measured by expanding it in a fixed volume with a calibrated pressure gauge. CO2 standard gas of known concentration (375 ppm) was regularly measured. The instrumental error in the concentration measurement is dataset was then converted to pCOatm (Eq. (1)) 0.1%. The final xCOatm 2 2 considering the atmospheric pressure (patm) and the partial pressure of water vapour (pH2 O), which was calculated from in situ SST readings (Tis) [Cooper et al., 1998] (Eq. (2)). = xCOatm pCOatm 2 2 (patm − pH2 O),

(1)

pH2 O = 0.981 exp(14.32602 − (5306.83/(273.15 + Tis))).

(2)

2.3. Flux calculation The CO2 exchange flux (mmol/m2 /d) across the air–sea interface was calculated using equation given in Wanninkhof.47 F = k × s × ∆pCO2 ,

(3)

where k is the gas transfer velocity, s is the solubility of CO2 gas in seawater,48 and ∆pCO2 is the difference between surface seawater pCO2 atm and atmospheric pCO2 (pCOsw 2 − pCO2 ). The air–sea pCO2 difference is computed using the measured surface water pCO2 values and the atmospheric pCO2 . In situ ship board wind data was used for flux calculation. 2.4. Chlorophyll a For chlorophyll a measurement, 5 L seawater samples were filtered on 10 µm pore size 47 mm polycarbonate filters at low vacuum. The pigments were extracted from the phytoplankton in 90% acetone with the aid of a mechanical tissue grinder and allowed to settle for a minimum of 2 h, but not to exceed 24 h, to ensure thorough extraction of the chlorophyll a. The filter slurry was centrifuged for 15 min to clarify the solution. An aliquot of the supernatant was transferred to a glass cuvette and fluorescence was measured on fluorometer (Model 10-AU, Turner Designs) before and after acidification to 0.003 N HCl with 0.1 N HCl. Sensitivity calibration factors, which have been previously determined on solutions of pure chlorophyll a of known concentration, are used to calculate the concentration of chlorophyll a.

July 13, 2012 9:9

AOGS - AS and OS

84

9in x 6in

b1387-ch07

S. Shetye et al.

2.5. Nutrient Analysis Nutrients (silicate, phosphate, nitrate, and nitrite) were measured with a Skalar Autoanalyzer by standard colorimetric methods. Standards were used to calibrate the auto-analyzer and frequent baseline checks were made. The standard deviation for duplicates was 0.07 µM for silicate, 0.06 µM for nitrate, 0.01 µM for nitrite and phosphate.

2.6. TOC analysis Total Organic Carbon (TOC) which is a sum of Dissolved organic carbon (DOC) and Particulate organic carbon (POC) was measured using the TOC-V-CSH analyzer by high temperature catalytic oxidation method. Standard deviations of triplicate measurements were found to be about 0.2%.

3. Results and Discussion 3.1. Frontal variations Frontal zones are defined by sharp changes in temperature and salinity, and are also areas of enhanced biological production.15 It is necessary to study the different frontal structures within the area since they are known to affect the biological activity and thus the CO2 system.10 We adopted surface temperature gradient and salinity as property indicators for identification of frontal zones.1,28 We identified four fronts, namely: Agulhas Retroflection Front (ARF), Sub Tropical Front (STF), Sub Antarctic Front (SAF) and the Polar Front (PF). The STF divides warmer tropical waters and colder sub-tropical waters and is seen in the present data as a sharp temperature gradient from 41◦ S to 43◦ S at TW and 43◦ S to 46◦ S at TE . The SAF was found from 43◦ S to 48◦ S at TW and from 46◦ S to 48◦ S at TE and is marked by a sharp temperature decrease. The PF, considered to be an important ecological boundary, was found between 49◦ S to 55◦ S at TW and from 50◦ S to 56◦ S at TE . The Polar Frontal Zone (PFZ) is characterized by strong lateral mixing and relatively low biological production.45 3.2. Nutrient dynamics Mixing brings nutrients and CO2 from deeper layers to the surface. Surface NO3 concentrations were