The Development of COMCOT Storm Surge Model and the Applica on to 2013 Typhoon Haiyan
2014/03/25 (二 )
Prof. Tso-Ren Wu (吳祚任) 蔡育霖、莊美惠、林君蔚 [email protected] Na onal Central University (Taiwan) 國立中央大學(台灣) Introduc on of COMCOT (Cornell Mul -grid Coupled Tsunami Model)
• Solve shallow water equa ons on both spherical and Cartesian coordinate systems • Explicit leapfrog Finite Difference Method for stable and high speed calcula on system • Mul /Nested-grid system for mul ple wave scales • Moving Boundary Scheme for inunda on (2). COMCOT has been used on many scien fic papers At least 26 SCI papers were published during 2001~2011 (Including Science)
1. Title: Long waves through emergent coastal vegeta on 7. Title: An insitu borescopic quan ta ve imaging profiler for the 12. Title: Analy cal and numerical simula on of tsunami Author(s): Mei Chiang C.; Chan I-Chi; Liu Philip L. -F.; et al. measurement of high concentra on sediment velocity mi ga on by mangroves in Penang, Malaysia Source: JOURNAL OF FLUID MECHANICS Volume: 687 Pages: Author(s): Cowen Edwin A.; Dudley Russell D.; Liao Qian; et al. Author(s): Teh Su Yean; Koh Hock Lye; Liu Philip Li-Fan; et al. 461-491 DOI: 10.1017/jfm.2011.373 Published: NOV 2011 Source: EXPERIMENTS IN FLUIDS Volume: 49 Issue: 1 Special Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 Issue: SI Pages: 77-88 DOI: 10.1007/s00348-009-0801-8 Issue: 1 Pages: 38-46 DOI: 10.1016/j.jseaes.2008.09.007 2. Title: Insights on the 2009 South Pacific tsunami in Samoa Published: JUL 2010 Published: SEP 4 2009 and Tonga from field surveys and numerical simula ons Author(s): Fritz Hermann M.; Borrero Jose C.; Synolakis 8. Title: Tsunami hazard from the subduc on megathrust of the 13. Title: Simula on of Andaman 2004 tsunami for assessing Costas E.; et al. South China Sea: Part I. Source characteriza on and the resul ng impact on Malaysia Source: EARTH-SCIENCE REVIEWS Volume: 107 Issue: 1-2 tsunami Author(s): Koh Hock Lye; Teh Su Yean; Liu Philip Li-Fan; et al. Special Issue: SI Pages: 66-75 DOI: 10.1016/j.earscirev. Author(s): Megawa Kusnowidjaja; Shaw Felicia; Sieh Kerry; et al. Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 2011.03.004 Published: JUL 2011 Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 Issue: 1 Issue: 1 Pages: 74-83 DOI: 10.1016/j.jseaes.2008.09.008 Pages: 13-20 DOI: 10.1016/j.jseaes.2008.11.012 Published: SEP 4 Published: SEP 4 2009 3. Title: Solid landslide generated waves 2009 Author(s): Wang Yang; Liu Philip L. -F.; Mei Chiang C. 14. Title: SPECIAL ISSUE Tsunamis in Asia Preface Source: JOURNAL OF FLUID MECHANICS Volume: 675 Pages: 9. Title: Simula on of Andaman 2004 tsunami for assessing impact Author(s): Liu Philip L. -F.; Huang Bor-Shouh 529-539 DOI: 10.1017/S0022112011000681 Published: on Malaysia Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 MAY 2011 Author(s): Koh Hock Lye; Teh Su Yean; Liu Philip Li-Fan; et al. Issue: 1 Pages: 1-1 DOI: 10.1016/j.jseaes.2009.05.001 Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 Issue: 1 Published: SEP 4 2009 4. Title: An explicit finite difference model for simula ng Pages: 74-83 DOI: 10.1016/j.jseaes.2008.09.008 Published: SEP 4 weakly nonlinear and weakly dispersive waves over slowly 2009 15. Title: INDIAN OCEAN TSUNAMI ON 26 DECEMBER 2004: varying water depth Times Cited: 0 (from Web of Science) NUMERICAL MODELING OF INUNDATION IN THREE CITIES ON Author(s): Wang Xiaoming; Liu Philip L-F THE SOUTH COAST OF SRI LANKA Source: COASTAL ENGINEERING Volume: 58 Issue: 2 Pages: 10. Title: Modeling tsunami hazards from Manila trench to Taiwan Author(s): Wijetunge J. J.; Wang Xiaoming; Liu Philip L. -F. 173-183 DOI: 10.1016/j.coastaleng.2010.09.008 Published: Author(s): Wu Tso-Ren; Huang Hui-Chuan Source: JOURNAL OF EARTHQUAKE AND TSUNAMI Volume: 2 FEB 2011 Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 Issue: 1 Issue: 2 Pages: 133-155 Published: JUN 2008 Pages: 21-28 DOI: 10.1016/j.jseaes.2008.12.006 Published: SEP 4 5. Title: Field Survey of the Samoa Tsunami of 29 September 2009 16. Title: TSUNAMI SOURCE REGION PARAMETER 2009 Times Cited: 0 (from Web of Science) IDENTIFICATION AND TSUNAMI FORECASTING Author(s): Okal Emile A.; Fritz Hermann M.; Synolakis Costas Author(s): Liu Philip L. -F.; Wang Xiaoming E.; et al. 11. Title: Tsunami hazard and early warning system in South China Source: JOURNAL OF EARTHQUAKE AND TSUNAMI Volume: 2 Source: SEISMOLOGICAL RESEARCH LETTERS Volume: 81 Sea Issue: 2 Pages: 87-106 Published: JUN 2008 Issue: 4 Pages: 577-591 DOI: 10.1785/gssrl.81.4.577 Author(s): Liu Philip L. -F.; Wang Xiaoming; Salisbury Andrew J. Published: JUL-AUG 2010 Source: JOURNAL OF ASIAN EARTH SCIENCES Volume: 36 Issue: 1 17. Title: Bo om fric on and its effects on periodic long wave Pages: 2-12 DOI: 10.1016/j.jseaes.2008.12.010 Published: SEP 4 propaga on 6. Title: Impact of a 1755-like tsunami in Huelva, Spain 2009 Author(s): Orfila A.; Simarro G.; Liu P. L. F. Author(s): Lima V. V.; Miranda J. M.; Bap sta M. A.; et al. Source: COASTAL ENGINEERING Volume: 54 Issue: 11 Source: NATURAL HAZARDS AND EARTH SYSTEM SCIENCES Pages: 856-864 DOI: 10.1016/j.coastalene.2007.05.013 Volume: 10 Issue: 1 Pages: 139-148 Published: 2010 Published: NOV 2007 ( To be con nued) (3). COMCOT has been widely validated: Synolakis solitary wave runup (1986, 1987). soliton
Simulated by COMCOT
(Wu, 2012) (NOAA) (4). Widely Used on Prac cal Cases
(4A). Simula on Results for Sri Lanka • The model has been used to inves gate several historical tsunami events, such as the 1960 Chilean tsunami, the 1992 Flores Islands (Indonesia) tsunami (Liu et al., 1994; Liu et al., 1995), the 2003 Algeria Tsunami (Wang and Liu, 2005) and more recently the 2004 Indian Ocean tsunami (Wang and Liu, 2006).
Galle, Sri Lanka
Inunda on
(Products from COMCOT: Inunda on and flux direc on) (4B). 311 Japan Tsunami ( Revealing Stability and High-speed Calcula on) Good comparison with Japan and American data.
Good comparison with CWB gauge data
1. COMCOT only spends one (without astronomical de). minute in computa on.
2. Good comparison with observa on data like 12 cm predic on wave gauge height. (4C). COMCOT was applied in Na onal Disaster Preven on
Distribu on of 22 poten al tsunami sources
T1, Mw=8.1 花蓮外海
T2, Mw=8.2 馬尼拉海溝 1
T3, Mw=8.4 馬尼拉海溝 2
T8,Mw=8.7 亞普海溝 Five Nested-grid were Applied to Simulate the Tsunami propaga on. (Bathymetry and eleva on database in Taiwan were already build up.) (Nonlinear equa ons and spherical coordinate were chosen in all simula on.)
Layer 1: 2 min (~3500m); Layer1 Layer2 Layer 2: ½ min (~900m); Layer 3: 1/8 min (~200m); Layer3 Layer 4: 1/32 min (~50m); Layer 5: 1/128 min (~20m);
Inunda on range and maximum wave height in Nanwan.
南灣 核三廠
後壁湖
These pictures indicate the tsunami energy transport from Yap Trench to Taiwan. Cartesian coordinate is not suitable in big domain because of error predica on. • (5). Stable and Fast。 Parallelized by ASGC, COMCOT now is able to use all the mutli-core CPU resources • Thanks a lot to ASGC, especially Dr. Simon Lin (林誠謙 ) and his excellent team!!
( We tested COMCOT on a new 32-core server in NTU in 2010, Singapore. A case used to be done in 30 minutes can be finished in 2 minutes on the new machine.) 9 iCOMCOT: a grid/cloud-based Tsunami system
COMCOT is now upgraded to a cloud system by ASGC. Countries worldwide are able to perform the simula on on a mobile device. (1) Given a keynote speech invited by UNESCO (2) Interviewed by isgtw, London, UK
h p://www.isgtw.org/feature/forecas ng-wrath-tsunami Transfer COMCOT into a Storm Surge Model Pressure gradient and wind shear stress were added to the shallow water equa on.
(1). Governing Equa ons in Spherical Coordinate ��/�� +1/�cos� {��/�� +�/�� (cos�·�)} =0 Pressure gradient
��/�� +�ℎ/�cos� ��/�� −�⋅�=−ℎ/��cos� ���/�� + 1/� ⋅�↓�↑� Wind shear stress ��/�� +�ℎ/� ��/�� +�⋅�=−ℎ/�� ���/�� +1/� ⋅�↓�↑�
� : free surface eleva on �� : atmospheric pressure ℎ : s ll water depth � : Coriolis parameter �: earth radius � : density of water �, � : longitude and la tude of the earth �↓�↑� , �↓�↑� :surface wind shear stress P, Q :volume luxes in longitude and latitude (2). Verifica on by the analy cal solu on in the Cartesian Coordinate System
With pressure gradient With wind shear stress
∂∂η 1 Pa ∂η g = gH= F s ∂∂xxρ ∂x x uuv uuv ∂Pxa =10000sin(2π / 5000) s FCVVxadww= ρ 1 ∂ηπ=∂Pxa =−1.02sin(2 / 5000) −5 gρw ∂η =2.1×× 10 5000= 0.105m
P Vw:26 m/s
10m 10m 400m 5000m 400m 5000m
Black: Analy c Solu on; Blue: Numerical Solu on. Black: Analy c Solu on; Blue: Numerical Solu on. (3). Verifica on in the Spherical Coordinate System
Only Pressure Gradient Only Wind Shear Stress ��/�� =−1/�� ��↓� /�� ��/�� =�cos�/�� ⋅1/ �↓� ⋅�↓�↑� ��↓� =200000cos(2��/� ) uuv uuv FCVVs Vmsw = 20 / ψ = ρad w w
Water Depth: 100 m. Black: Analy c Solu on; Blue :Numerical Solu on. (4). Flow Chart of the Storm Surge Predic on
Atmospheric Model Input Data: ex: TWRF 1. CWB Warning Data 2. Typhoon Informa on
Wave Model ex: WaveWatch III
Parametric Wind Model: Holland Model (1980) Current Model
Product
Gauge Data: 2-D Output Data: 36 hrs wave height Surge Devia on 36 hrs wind velocity Pressure Field 36 hrs pressure Wind Field
地形資料
水位計位置 (5). Parametric Wind Model ���� : maximum wind velocity • Holland Model � (1980) max: radius of maximum wind R max B PPac=+()exp[()] PP nc − − �� r : atmospheric pressure
Pc − 900 � : Coriolis parameter B =2 − 160 �↓� : density of water BP()− P R R rf rf V =nc()exp[()]()maxBB− max +2 −�↓� : ambient pressure w ρ rr22 a � : distance to storm center �� : wind velocity
• Current Model ⎧42.6−⋅− 0.86 (Pc 990) R ⎪51.0 0.84 (P 980) max B ⎪ −⋅−c PPac=+( PP nc − )exp[ − ( ) ] r ⎪58.4−⋅− 0.74 (Pc 970) ⎪ 1 Vmax 2 RP63.0 0.46 ( 960) B = () max =⎨ −⋅−c (km) PP5.375 nc− ⎪70.0−⋅− 0.234 (P 930) ⎪ c Rrmax ⋅ ⎪80.0−⋅− 0.167 (Pc 870) VVw =2⋅max ⋅ 22 ⎪ Rrmax + ⎩80.0 (6). Model Coupling (6A). Coupled with Atmospheric Model
• TWRF model was one of atmospheric model used by CWB, which projec on system is Lambert.
• Run four mes every day at 00 UTC , 06 UTC , 12 UTC and 18 UTC.
• The ini al condi on of first nested-grid data was form American NCEP GFS, and the second and third nested-grid were interpolated form first layer. Nested-grid Set Domain 1 Domain 2 Domain 3 Projec on Lambert Resolu on(km) 45 km 15 km 5 km X Mesh Number 221 181 151 Y Mesh Number 127 193 181 Center of Longitude 118.59 122.27 121.07 Center of La tude 27.07 22.87 24.28 Lambert Projec on Spherical Coordinate
interpolate
2011 Typhoon Nanmadol 08.27 00 – 08.29 00 Case Study (1). Introduc on Typhoon Nanmadol (interna onal designa on: 1111, JTWC designa on: 14W, PAGASA Name: Mina) was the strongest tropical cyclone in 2011 to hit the Philippines and also the first of the year to directly impact Taiwan and the rest of the Republic of China (ROC).
Seawater inwelling was caused by storm surge and spring de. Because of its low eleva on, flood was inundated in the streets. (TVBS) (Wiki) (2). Numerical Domain
• Big numerical domain was chosen to cover the whole life-cycle of typhoon.
• The bathymetry is obtained from ETOPO2.
Domain Holland Model (3). Parametric Model
Current Model (3A). Surge Devia on
Holland Model Current Model
Dura on: 2011.08.23 12 – 2011.08.30 18 (UTC+8) Resolu on: 10 min Time Step: 10 s Total Time: 174 hours (~ 7 days) (4). TWRF Model
Dura on: 2011.08.23 12 – 2011.08.30 18 (UTC+8) Resolu on: 10 min Time Step: 10 s Total Time: 174 hours (~ 7 days) (5). Posi ons of Numerical Gauges
Name Longitude La tude SHIHTI 121.50 23.49 CHENGKUNG 121.37 23.08 TAITUNG 121.19 22.79 DAWU 120.89 22.33 LYUDAO 121.46 22.66 LANYU 121.49 22.06 SHIHMEN 121.51 25.28 DANSHUEI 121.40 25.18 TAOYUAN 121.23 25.12 HSINCHU 120.91 24.85 HOULONG 120.77 24.65 WUCHI 120.53 24.28 FANGYUAN 120.30 23.91 MAILIAO 120.17 23.77 TAIHSI 120.13 23.61 TUNGSHI 120.14 23.44 JIANGJYUN 120.07 23.21 ANPING 120.16 22.96 YONG-AN 120.19 22.81 KAOHSIUNG 120.28 22.61 TUNGKANG 120.43 22.45 JIAHE 120.63 22.32 SUNGUANGZUEI 120.71 21.99 NANWAN 120.73 21.94 SIAOLIOUCIOU 120.38 22.35 34 numerical gauges were set up in numerical model PENGHU 119.56 23.56 KINMEN 118.43 24.41 according to CWB informa on. MATSU 119.94 26.16 (6). Numerical Gauge Data Dura on: 2011.08.23 12 – 2011.08.30 18 (UTC+8) Total Time: 174 hours (~ 7 days)
Surge Devia on 基隆 梧棲
蘇澳 成功 (6B). Numerical Gauge Data Dura on: 2011.08.23 12 – 2011.08.30 18 (UTC+8) Total Time: 174 hours (~ 7 days)
馬祖 蘭嶼
南灣 安平
Surge Devia on (6C). Comparison with Observa on
馬祖 蘭嶼
(3A) Introduc on
• Typhoon Haiyan, known as Typhoon Yolanda in the Philippines, was a powerful tropical cyclone that devastated por ons of Southeast Asia, par cularly the Philippines, on November 8, 2013.
• Haiyan is also the strongest storm recorded at landfall, and unofficially the strongest typhoon ever recorded in terms of wind speed.
• The cyclone caused catastrophic destruc on in the Visayas, par cularly on Samar and Leyte. According to UN officials, about 11 million people have been affected and many have been le 如海嘯過後 海燕颱風重創菲律賓 homeless. h p://world.yam.com/post.php?id=1316 (Wiki) (3B) Numerical Set l Time: 2013/11/07 – 2013/11/09 (UTC+0) l Dura on: 60 hrs (21600 sec) l Domain: 114.80 °– 134.63°, 7.80 ° – 17.63 ° l Resolu on: 3 arc-min l Bathymetry and Eleva on Data: ETOPO 1
Computa onal Domain WRF DATA
PRESSURE FIELD
INPUT
COMCOT WRF DATA
WIND FIELD
INPUT
COMCOT RESULT-SURGE DEVIATION Performance Test
Simula on Informa on Total Run Time 216000 s (~60 hrs) Total Step 108000 Mesh Number 400*200 Time Spend 900 sec + 600 sec Thanks for Listening.