The Role of Urban Development Patterns in Mitigating the Effects of Tsunami Run-up: Final Report

Size: px

Start display at page:

Download "The Role of Urban Development Patterns in Mitigating the Effects of Tsunami Run-up: Final Report"

Brett Miller
5 years ago
Views:

1 J-RAPID Final Symposium Sendai, Japan The Role of Urban Development Patterns in Mitigating the Effects of Tsunami Run-up: Final Report March 6, 2013 Fumio Yamazaki, Chiba University, Japan and Ronald T. Eguchi, ImageCat, Inc., USA 1

2 Research Team Fumio Yamazaki, Chiba University Shunichi Koshimura, Tohoku University Masashi Matsuoka, Tokyo Institute of Technology Yoshhisa Maruyama, Chiba University Ronald T. Eguchi, ImageCat, Inc. John Bevington, ImageCat Ltd., UK Michael T. Eguchi, ImageCat, Inc. Albert Lin, UCSD Andrew Huynh, UCSD James D. Goltz, Consultant 2

3 Scope of NSF Rapid Grant Perform field studies to collect perishable data on coastal community performance following the Tohoku earthquake Develop an understanding of the data landscape in post-earthquake Japan Develop a preliminary understanding of the role that urban development patterns played in either mitigating or exacerbating tsunami-induced impacts 3

4 Communities Visited and/or Studied by US team City Predominant Development Types * Area km.) (sq. Population Max. Tsunami Ht ** (m) Manmade Tsunami Protection Coastal Configurati on Mountaino us (Yes/No) Area Description of Damage Asahi RL/A , Coastal Strand N Limited damage Ishinomaki RH/I , Breakwater and Coastal N Wide range of Seawalls Strand damage Kamaishi RH/I ,022 9 Breakwater and Bay Y Wide range of Seawalls damage Kashima RL/I 93 66,249 - Seawall Coastal N Limited Damage Strand Kesennuma No pre-event , Bay Y Wide range of data damage Minamisanriku RL/C , Breakwater and Bay Y Destroyed Seawall (3.5m) Miyako RH , Seawall (10m) Bay Y Wide range of damage Ofunato RH/I , Breakwater Bay Y Wide range of damage Onagawa RH 65 11, Breakwater Bay Y Destroyed Otsuchi RH , Breakwater and Bay Y Destroyed Seawall (10m) Rikuzentakata RH/A , Mitigation Bay Y Destroyed Forest and Seawall (6.5m) Sendai RH/C/A 788 1,031, Coastal Strand N Wide range of damage Yamada RH , Seawall (6.6) Bay Y Wide range of damage 4

5 Tsunami Inundation Heights Ishinomaki Source: Point data from 2011 Tohoku Earthquake Tsunami Joint Survey Group 5

6 Tsunami Inundation Heights Ishinomaki Source: Point data from 2011 Tohoku Earthquake Tsunami Joint Survey Group 6

7 Ishinomaki 3-4 m 4-5 m 5-6 m 6-7 m 7-8 m 7

8 8

9 Damage ratios by Landuse Ishinomaki Land Use % Commercial Industrial Residential Tsunami Height (m) 9

10 Kamaishi 10

11 11

12 Kamaishi 12

13 13

14 Goal of US side Influence the loss modeling done for large and small communities incorporating tsunami hazard effects Current effort in the US to augment HAZUS for tsunami Sendai is a case study area for calibrating damage and loss models for HAZUS-Tsunami 14

25m) Radarsat 1, 2 (C-band, 8m) TerraSAR-X (X-band, 3m) COSMO-SkyMed (X-band, 3m) Optical,

15 Satellites images of the 2011 Tohoku earthquake Optical, Medium Resolution ALOS AVNIR-2 (10m) Terra ASTER (15m) Landsat 7 (30m) SAR ALOS PALSAR (L-band, 6.25m) Radarsat 1, 2 (C-band, 8m) TerraSAR-X (X-band, 3m) COSMO-SkyMed (X-band, 3m) Optical, High Resolution FORMOSAT-2 (2.0m) THEOS (2.0m) RapidEye (2.5m) WorldView-1,2 (0.5m) QuickBird (0.6m) Ikonos (1.0m) GeoEye-1 (0.5m) 15

16 Acquisition condition of various sensors and platforms in disaster response Platform /Sensor Satellite Large coverage Airborne Mod. coverage Ground Based Low coverage Optical Sensor Day, Fixed time No cloud Day, Any time No low cloud Day, Any time Lidar Day, Any time No low cloud Thermal Infrared All day, Fixed time No cloud Low resolution All day, Any time No low cloud Mod. resolution Day, Any time All day, Any time High resolution SAR All day, Fixed time All weather All day, Any time All weather R & D stage 16

17 Crustal Movement Crustal movements observed by GSI GPS stations in March 11 and 13, 2011 InSAR results from PALSAR images Horizontal Vertical okuchi40010.html 17

TerraSAR-X images Pre-event Post-event * Rifu * Yamoto * Natori * Watari a b c d a b c d Date 2010.10.21 2011.03.13 2011.03.24 2011.04.04 View angle 37.316 37.301 37.319 37.317 Heading 190.027 190.

18 TerraSAR-X images Pre-event Post-event * Rifu * Yamoto * Natori * Watari a b c d a b c d Date View angle Heading Mode StripMap (3.01 m x 3.04 m) Polarization HH Data EEC (1.25 m/pixel) Transformed to Sigma Naught (σ 0 ) Enhanced Lee filter (3 x 3 pixels) The 9th CUEE and 4th ACEE Joint Conference 18

Movement of a building 19 115 x 115 pixels I:

x y Pre-event Post-event Color composite of SAR

Maximum Center { }{ } { } { } = = = = = = = 1 0

), ( ), ( T T T T T T M i N j M i N j b a M i N

= 1 0 1 0 ), ( 1 T T M i N j T T j i T N M T = =

N M I Area-based correlation 3.75 m to east, 1.

19 Movement of a building x 115 pixels I: Post-event SAR 101 x 101 pixels T: Pre-event SAR x y Pre-event Post-event Color composite of SAR Correlation coefficient Correlation Matrix Maximum Center { }{ } { } { } = = = = = = = ), ( ), ( ), ( ), ( ), ( ), ( ), ( T T T T T T M i N j M i N j b a M i N j b a T j i T I j i I T j i T I j i I b a R = = = ), ( 1 T T M i N j T T j i T N M T = = = ), ( ), ( 1 T T M i N j b a T T j i I N M I Area-based correlation 3.75 m to east, 1.25 m to south (1.25m/pixel) R: G&B:

20 Building extraction (Yamoto) Yamoto * Building objects (pre-event) 2.5 km Building objects (post-event) Segmentation σ 0 > -2.0 db Size > 100 pixels (about 150 m 2 ) Color composite of building objects R: G&B:

21 Movement detection (Yamoto) Color composite of building objects Building object (pre-event) Surrounding area (post-event) 5 pixels Building exist Non-change buildings SAR intensity images T (pre-event SAR) 3 pixels I (post-event SAR) 5 pixels Resized to 0.25m by cubic convolution r > m to east 1.00 m to south The 9th CUEE and 4th ACEE Joint Conference 21

Crustal movement (Yamoto) * GPS 67 buildings 3 m 20

49 m σ= 0.62 2.00 2.50 3.00 Movements /m 3.50 4.00 4.

50 North -1.00 30 25 20 15 10-0.50 5 0 0.

22 Crustal movement (Yamoto) * GPS 67 buildings 3 m East μ= 3.49 m σ= Movements /m μ= m σ= North Movements /m GPS station Google Earth Survey photo

23 Crustal movement The study area was divided into a square mesh containing 2.5 x2.5 km 2. Only the sub-area including more than 5 building displacements was valid. The movement of the sub-area was the mean value of all building displacements. R: post-event G&B: pre-event 23

Crustal movement and SAR image Horizontal Vertical M M M

Incident angle M: Movement in SAR image r, a : Range and

24 Crustal movement and SAR image Horizontal Vertical M M M M r a E N D cosα sinα 1/ tanθ = D sinα cosα 0 D sinα cosα M r 1 = = cosα sinα M a 0 E N Z 0 1 D cosα / tanθ D sinα / tanθ D E N Z D : Actual movement α : Heading angle θ : Incident angle M: Movement in SAR image r, a : Range and azimuth ffffffdirection E,N : East and north direction 24

25 Comparison of GPS observed data (1) * * * GPS observed data Detected results East Movement /m North * *GPS Date Yamoto 1-Mar 6-Mar 11-Mar 16-Mar 21-Mar 26-Mar 31-Mar 5-Apr 10-Apr 15-Apr 20-Apr 25-Apr Rifu East West Survey photo Rifu 25

Comparison of GPS observed data (2) Watari Natori Survey photo 2012.

00 Movement /m Date 5.00 4.00 3.00 2.00 1.00 0.00-1.

16-Mar 21-Mar 26-Mar 31-Mar 5-Apr 10-Apr 15-Apr 20-Apr 25-Apr East

20-Apr 25-Apr Movement /m 0.40 0.00-0.40-0.80-1.20-1.

26 Comparison of GPS observed data (2) Watari Natori Survey photo Movement /m Movement /m Date Date GPS observed data Detected results East 1-Mar 6-Mar 11-Mar 16-Mar 21-Mar 26-Mar 31-Mar 5-Apr 10-Apr 15-Apr 20-Apr 25-Apr East 1-Mar 6-Mar 11-Mar 16-Mar 21-Mar 26-Mar 31-Mar 5-Apr 10-Apr 15-Apr 20-Apr 25-Apr Movement /m Date Natori Watari North 1-Mar 6-Mar 11-Mar 16-Mar 21-Mar 26-Mar 31-Mar 5-Apr 10-Apr 15-Apr 20-Apr 25-Apr North 26

27 27 Flooded Area and Damage Extraction from SAR Data TerraSAR-X 14 scenes

13 2010.09.11 2011.03.16 Date 2010.10.21 2011.

28 SAR data for flooded area detection TerraSAR-X ALOS/PALSAR Date Incident angle 37.3º 41.5º 43.1º Mode SM FBS Resolution 3.0 x 3.0 m (R x A) 6.25 x 7.0 m (R x A) Pixel size 1.25 m 6.25 m 28

Flooded area was extraction by the difference (d) Result of TSX images 1.

29 Extraction of flooded areas The displacements caused by crustal movement were removed by shifting the post-event image. Flooded area was extraction by the difference (d) Result of TSX images 1.25 m/pixel 15 x 15 pixels window d = I b I a Result of PALSAR images 6.25 m/pixel 7 x 7 pixels window Run-up lines (PASCO) Extracted inundation 29

30 Extracted result from TSX Extracted result from TSX Verification of extracted results Error matrix for the result of TerraSAR-X images [%] Inundation map from PASCO Non-flooded area Flooded area Total User accuracy Non-flooded area Flooded area Total Producer accuracy Error matrix for the result of PALSAR images [%] Inundation map from PASCO Non-flooded area Flooded area Total User accuracy Non-flooded area Flooded area Total Producer accuracy It is more difficult to distinguish flooded areas with paddy fields from L- band than X-band SAR images. The different incident angles of PALSAR images were one reason for low accuracy. Most extracted flooded areas from TSX images were within the run-up boundaries. 30

31 Image data TerraSAR-X images

obtained from Zenrin was used to detect damaged buildings.

32 Method of damaged building extraction 32 Building shapes Range Building shapes (about 10 million) A GIS date of building outlines obtained from Zenrin was used to detect damaged buildings. The building heights were detected and the outlines were shifted to match with the image. 12 m

33 Results of building height detection 1F: 0-6 m 2F: 6-9 m 3F: 9-12 m 4F: 12 m- To save calculation time, the building heights were detected by 1m unit. It s result was transformed into story numbers. 33

34 Results of damaged building detection The SAR layover length was used to detect damaged buildings. If the average of (z1+z2)/2 within a building outline is larger than 0, than that building was damaged. Change factors 1: ー : ー Run-up lines (PASCO) Damaged Non-damage Result of damage classification 34

35 Verification of detected result The Building Damage Map made by visual interpretation from aerial photos was used to verify the accuracy. The reasons for errors Wrong outline after shifting Omission caused by debris The influences by flooded area Detected result Building Damage Map Washed Survived Total away U. A. Damaged % No damage % Total P. A. 63.5% 95.0% 91.1% Building unit-based Error matrix by building unit-based and pixel-based Kappa coefficient = 0.61 Building Damage Map (%) Pixel-based Washed Survived Total away U. A. Damaged No damage Total P. A Detected result 35

36 Major Outcomes Tsunami run-up boundary maps were produced from field survey and remote sensing data. Flooded areas and building damage in the target areas were accessed by field surveys and image interpretation. Crustal movements were estimated based on the shifts of buildings in high-resolution SAR intensity images. 36

37 Major Publications W. Liu, F. Yamazaki, Detection of Crustal Movement from TerraSAR-X intensity images for the 2011 Tohoku, Japan Earthquake, Geoscience and Remote Sensing Letters, IEEE, Vol. 10, No. 1, pp , W. Liu, F. Yamazaki, H. Gokon, S. Koshimura, Extraction of Tsunami- Flooded Areas and Damaged Buildings in the 2011 Tohoku-Oki Earthquake from TerraSAR-X Intensity Images, Earthquake Spectra, EERI, Vol. 29, S1- S18, Y. Maruyama, K. Kitamura, F. Yamazaki, Estimation of Tsunami-inundated Areas in Asahi City, Chiba Prefecture, after the 2011 off the Pacific Coast of Tohoku Earthquake, Earthquake Spectra, EERI, Vol. 29, Acknowledgement The TerraSAR-X images were provided to the present authors from Pasco Corporation, Tokyo, Japan, as one of the granted projects of the SAR data application research committee. 37

38 Thank you very much! Inc. 38

DETECTION OF BUILDING SIDE-WALL DAMAGE CAUSED BY THE 2011 TOHOKU, JAPAN EARTHQUAKE TSUNAMIS USING HIGH-RESOLUTION SAR IMAGERY

10NCEE Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 Anchorage, Alaska DETECTION OF BUILDING SIDE-WALL DAMAGE CAUSED BY THE 2011 TOHOKU,