Remote Sensing Information System for Disaster Management

Transcription

1 Remote Sensing Information System for Disaster Management KWOH Leong Keong, LIEW Soo Chin, CHIA Aik Song, John LOW* National University of Singapore (*Currently at Meteorological Division, National Environment Agency, Singapore) 22 nd APAN Meeting, Earth Monitoring Working Group, Singapore 19 July 2006

2 Introduction Remote sensing applications for disaster management Forest fires Floods Algal blooms Tsunami damage assessment Information support for disaster management Distribution of images of disaster affected areas through WWW (jpeg compressed for quick download). In-house development of an environmental information system Sentinel Asia

3 Regional Forest Fire Monitoring

4 Forest Fire in Southeast Asia Vegetation fire is an annual occurrence in Southeast Asia. During periods of draught, fires get out of control causing smoke haze pollution problem June 1995 : ASEAN Co-operation Plan on Transboundary Pollution July - Nov 1997 : Severe fires and haze in Southeast Asia ASEAN Regional Haze Action Plan Sets out co-operative measures amongst ASEAN member countries to address the problem of smoke haze in the region arising from land and forest fires Singapore takes up the responsibility of coordinating the regional land and forest fire monitoring mechanisms.

5 Fire Monitoring Operation and Research Activities at CRISP Daily fire monitoring operation using high resolution SPOT satellite imagery Collaboration with the National Environmental Agency, Singapore Determination of fire locations and types of land cover burnt Operational production of near-real time active fire products using MODIS direct broadcast data Validating MODIS fire products with high resolution satellite imagery Adapting global algorithms to local and regional environmental conditions optimized for fires in this region

6 IKONOS SPOT CRISP Ground Station Coverage

7 CRISP ground station reception circle for MODIS

8 Fire monitoring Procedure Automatic hotspot detection is performed on MODIS data immediately after acquisition and L1 processing. Hotspots data are transmitted by to users. Detected hotspots are examined for fire severity. Locations of significant hotspots are programmed for SPOT image acquisition the next day if available. SPOT images are acquired and processed. High resolution SPOT images are examined by human operators for presence of smoke plumes. Annotated satellite image maps and fire report are transmitted to user within the same day.

9 Fire Monitoring Operation Programming SPOT Passes Fire detected by SPOT on 9 June 2003 MODIS detected hotspots on 8 June 2003

10 Annotated Fire Image

11 Burned Area in 1997 Fire Derived from SPOT Quicklook Images Sumatra : 2.08 Mha Kalimantan : 3.40 Mha

12 N Sumatra Perak, Selangor Fires detected by High Resolution SPOT Images Riau Jambi Sumatra and Peninsular Malaysia January to 2005 July

13 Central Kalimantan West Kalimantan Fires detected by High Resolution SPOT Images Kalimantan and East Malaysia January to 2005 July

14 Fires detected by high resolution SPOT images by region Total Fire Count from 1999 January to 2005 July W Malaysia Sarawak Sabah E Kalimantan S Kalimantan C Kalimantan W Kalimantan Lampung S Sumatra Jambi Riau W Sumatra N Sumatra Aceh

15 Yearly Fire Count, , Kalimantan & E Malaysia (Jan-Jul) W Kalimantan C Kalimantan S Kalimantan E Kalimantan Sabah Sarawak

16 Yearly Fire Count, , Sumatra & W Malaysia (Jan-Jul) Aceh N Sumatra W Sumatra Riau Jambi S Sumatra W Malaysia

17 Monthly Fire Count and Monthly Rainfall (Sumatra) Average Monthly Fire Count (Sumatra, ) Climatological Mean Monthly Rainfall (Pekan Baru) Jan Feb Mar Apr May Jun Jul Aug Sep Oct NovDec 0

18 Fire and Land Use/Cover Change Left: Locations of active fires detected in SPOT images, 1998 to 2002, superimposed on 1998 land cover map Right: Land cover types observed in 2002, which were primary vegetation in 1998 Miettinen and Liew, Int. J. Remote Sensing 26, 1109, 2005.

19 Flood Monitoring

20 Flood Monitoring Procedure MODIS DB data is geolocated and calibrated to L1B A simple atmospheric correction (rayleigh correction) is performed on the L1B. The L2 (namely the shortwave bands 1-7) data is now warped to appropriate region at 250m Cloud and shadow test (used in NASA cloud mask). Perform a maximum likelihood supervised classification to 3 simple classes (water, forest and 'other' class)

21 Flood (cont'd) Perform a pixel change detection of the water class pixels against a monthly classified image during the normal season to detect water anamoly (possible flood). Change detection may also be performed on a composite of a chosen number of days (e.g. 7 days) on cloudfree data. Composite scheme based on minimum SWIR (Band 6) to preferentially choose water pixels.

22 Comparison of June 2004 and Feb 2004 composite. False colour using Bands 7(2.2micron), 2(0.8micron) and 1(0.6micron)

23 MODIS Cloud Free Composite of Flood in Lower Mekong Basin, 2002 Dry season Flood season (September 2002)

24 Sea Surface Chlorophyll Anomaly (Algal Bloom Monitoring)

25 A prototype near-real time sea water chlorophyll anomaly monitoring system MODIS Raw data Baseline Chlorophyll Distribution MODIS Level 1B data MODIS Level 2 Chlorophyll Product Chlorophyll Anomaly Detection S. C. LIEW and Alice W. C. HENG (2003), Proc. 24th Asian Conference on Remote Sensing Vol. II, pp Chlorophyll Anomaly Product

26 10 1 Average Chlorophyll Distribution (August 2002)

27 Standard deviation, August 2002 (blue:0 0.1, green: , yellow: , cyan : , red: >1.0)

28 Chlorophyll distribution and deviation above 2002 August monthly average 2003 August 11 Magenta: 1 sd Blue: 2 sd Green: 3 sd Yellow: 4 sd Red: >= 5 sd Chlorophyll Anomaly Product

29 Tsunami Damage Assessment

30 NW Sumatra MODIS 29 Dec 2004

31 SW Coast of Aceh SPOT 5, 8 Dec 2004 CNES, 2004

32 SW Coast of Aceh SPOT 5, 29 Dec 2004 CNES 2004

33 Banda Aceh (Ikonos Images) 10 Jan Dec km x 22 km

34 Banda Aceh

35 Banda Aceh

36 Tsunami affected areas in west coast of Aceh, Sumatra

37 Legends for (c) (a) Pre-tsunami Landuse map (b) Post-tsunami Damage map (c) SRTM DEM overlayed on the inundation map Legends for (a) and (b) P. Chen and S. C. Liew, Proc. IGARSS 2006

38 Pre-tsunami SPOT scene

39 Post-tsunami SPOT scene

40 Elevation of Tsunami Affected Area (DEM from SRTM)

41 Land cover affected by tsunami

42 Mapping Tsunami-Affected Coastal Aquaculture Areas in Northern Sumatra Using High Resolution Satellite Imagery S. C. Liew 1, S. P. Kam 2, P. Chen 1, Muchlisin Z. A. 3 1 CRISP, National University of Singapore, Singapore 2 World Fish Centre, Penang, Malaysia 3 Marine Science Centre, Syiah Kuala University, Banda Aceh, Indonesia Supported by United Nations Office of Outer Space Affairs

43 Main objectives To produce base maps of coastal areas of Aceh province using high resolution satellite images to map the damaged and intact aquaculture ponds after tsunami to be used for village-level targeting and planning of rehabilitation of aquacultural activities as part of reconstructing the livelihoods of fishing communities affected by the tsunami. Maps to be validated by Field checks

44 Study Area Northern Coast of Aceh SPOT-5 Quicklook Mosaic Kab. Aceh Besar Kab. Aceh Timur

45 SPOT 5 image: Before tsunami 10 July 2004 Continuous sand bars

46 SPOT 5 image: After tsunami 03 Feb 2005 Delineated polygons: Red : damaged ponds Magenta : intact ponds Yellow: uncertain Broken sand bars

47 Internet-Based Environmental Information System

48 MODIS Direct Broadcast Processing System at CRISP Workstation with GEOREF to generate browse image and Level 0 format. LINUX Workstation with GSFC DAAC software to generate L1 data- Calibration Geolocation LINUX Workstation with Brownie storage to archive L1 data and L2, 3 processing Web Server to distribute near realtime products to users

49 6m X-band antenna MODIS Data Reception and Processing System Terra/Aqua Demodulator MODIS Web-based Browse catalog PDS Generation Alert Near-Real Time Products Level 1 Processing Web-based Environmental Information System Composite Products Level 2 and higher level processing

50 CRISP as Data Provider Reception of Terra and Aqua MODIS Direct Broadcast MODIS Level 1B product generated using SEADAS software Near-real time processing of MODIS Hotspot product MOD14 Hotspot product subsequently generated Hotspot data pushed to Sentinel FTP server alert sent to subscribers Other environmental products will be added High resolution images (SPOT, IKONOS) of disaster affected areas (JPEG compressed for fast download) posted at CRISP web pages for public access. Under development: Southeast Asia Web-Mapping System, as part of Sentinel Asia

51 Southeast Asia Web Mapping System Linux OS: Fedora Core 2 Apache Web Server MapServer Static Layers: MODIS Land Cover GTOPO30 Country boundaries Layers pulled from MODIS processing system: MODIS Level 1B MOD14 Hotspots Server Client Any recent browser with Javascript enabled

52 Terra MODIS Fire Hotspots overlaid on 250m true color image 09 March 2005, UTC 03:27

53 (CRISP) National University of Singapore Near Realtime Fire Hotspots Alert System Satellite : TERRA Sensor : MODIS Algorithm: NASA ver Date : Start Time : UTC Location : Sumatra Number of hotspots detected : 164 No Lon Lat T4(K) T11(K) Bkg_T4(K) Bkg_T11(K) Sample Hotspot report automatically sent to subscribers by when fires detected within their prescribed monitoring area. Near Real-Time Alert

54 MODIS Hotspots provided by CRISP superimposed on a forest map of South Sumatra AQUA :29:31 UTC Forest map by Forestry Department, South Sumatra Province, Indonesia

55

56 In-house development of a regional environmental information system RISE Web Portal Regional Information System for the Environment Fire/Hotspot Menu Page: Java Applet allows selection of ROI

57

58 Thank You