disaster monitoring with remote sensing at crisp, nus · fires in 2006 zthe yearly vegetation fires...
TRANSCRIPT
Disaster Monitoring with Remote Sensing at CRISP, NUS
KWOH, Leong KeongDirector, CRISP
The 3rd Sentinel Asia Joint Project Team Meeting (JPTM)Le Meridien Hotel, Singapore
13-15 March 2007
13 metre Antenna
6 metre Antenna
1.5 metre Antenna
Data Processing Facilities
CRISP Ground CRISP Ground Station CoverageStation Coverage
CRISP Ground StationSatellite Data CRISP received since 1995:
1. SPOT 1,2,4 (France, Sep 1995) – 1,110,000 scenes
2. ERS 1,2 (Europe, Mar 1996) – 26,500 scenes
3. RADARSAT 1 (Canada, Apr 1997 till 1999) – 12,000 scenes
4. SeaWiFS (USA, Sep 1999)
5. TERRA MODIS (USA, March 2001)
6. IKONOS (USA, August 2001)
7. EROS (Israel, October 2001) – suspended due to ground equipment failure
8. AQUA MODIS (USA, July 2002)
9. SPOT 5 (October 2002) – 360,000 scenes
Forest Fire Monitoring
Regional fire monitoring operation
Planning SPOT passes aided by MODIS or AVHRR hot spot maps
SPOT data reception
Annoted fire image and fire report
Detection of fire and smoke plumes
Planning of SPOT Pass
SPOT 2 Image
MODIS Hotspot
Fires in 2006
The yearly vegetation fires in Sumatra and Kalimantan were made worse in 2006 by the onset of a mild El-Nino event.This brought drier weather at the tail end of the fire season (September and October) to South Sumatra and Kalimantan.The fires in South Sumatra and Kalimantan in this period were the most severe in recent years. The PSI in Singapore (on Oct 7) also hit the highest level since 1997.
High resolution satellite imagery: South Sumatra
SPOT4 Image, Sep 28, 2006
Red regions – recent burning activities (ground still very hot)
IKONOS Image, September 29, 2006
Systems of Canals, indicating intention of converting to plantations
Some Observations of the spatial and temporal patterns
of fires detected by high resolution SPOT images
Comparison of hotspots in September
Hotspots, 2nd half of September2005
Hotspots, 2nd half of September2006
Comparison of hotspots in October
Hotspots, 1st half of October2005
Hotspots, 1st half of October2006
Monthly Fire Count &. Monthly Rainfall (Sumatra)
0
20
40
60
80
100
120
Jan Feb Mar Apr May Jun Jul Aug Sep Oct NovDec
Aver
age
Mon
thly
Fire
Cou
nt
(Sum
atra
, 199
9 - 2
004)
0
50
100
150
200
250
300
350
Clim
atol
ogic
al M
ean
Mon
thly
R
ainf
all (
Peka
n B
aru)
Monthly Fire Count (Sumatra)2006 vs. 1999 – 2005 mean
0
20
40
60
80
100
120
140
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Mon
thly
Fire
Cou
nt (S
umat
ra)
1999 - 2005 Mean
2006
Validation of MODIS Hotspots Using High
Resolution SPOT Images
SC Liew et al (2003), Proc. 24th Asian Conference on Remote Sensing & 2003 International Symposium on Remote Sensing Vol. I, pp. 671-673.
Validation of MODIS Hotspots
Use high resolution SPOT multispectral images as the reference dataThe reference SPOT data are acquired almost simultaneously as the MODIS dataDetection of fires in SPOT images» Use visible smoke plumes to locate fires» The locations of the active fires are determined from
the points of origin of the individual smoke plumes.» Intense fires with efficient combustion may not
produce sufficient smoke plumes and would not be detected. Such fires are not very common in the humid tropics, due to the high humidity.
Detection of MODIS hotspots» Use standard NASA v4 algorithm
An example:
19 Aug 2002 Central Kalimantan
SPOT acquisition: 02:58 UTC
MODIS acquisition 03:15 UTC
Yellow dots are locations of fires detected in SPOT
Green dots are locations of MODIS hotspots
Results
MODIS Hotspots
SPOT Fires
Validated by SPOT
180 Detected by MODIS
181
False Alarms 66 Not Detected 94
Total 246 Total 275
Commission Error Rate
26.8% Omission Error Rate
34.2%
Comparison with a study conducted in Southern Africa
MODIS Hotspots ASTER fires
Validated by ASTER
33 Detected by MODIS
33
False alarms 3 Undetected fires
98
Total 36 Total 131
Commission Error Rate
0.8% Omission Error Rate
74.8%
Source: C. O. Justice, L. Giglio, S. Korontzi, J. Owens, J. T. Morisette, D. Roy, J. Descloitres, S. Alleaume, F. Petitcolin, and Y. Kaufman, 2002. The MODIS fire products. Remote Sensing of Environment, 83:244-262.
South Asia TsunamiSouth Asia Tsunami
26 December 2004, 00:58 am (UTC), 07:58 am (local time)
Subduction of India plate under Sundaplate
10 Jan 2003 29 Dec 2004
Banda Aceh
19 km x 22 km
Banda Aceh
Banda Aceh
Lhoknga
Lhoknga
Java MudflowJava Mudflow(August 2006)(August 2006)
2005-10-06
2006-08-29
2005-09-25
2006-10-31
Other Natural DisastersOther Natural Disasters
09 May 2006 28 May 2006
060124 before Tsunami
060722 after Tsunami
Village of Guinsaugon on 2 Sept 2004
Village of Guinsaugon on 1 Mar 2006
IKONOS Image of PADANG (8 Mar 07)
IKONOS Image of SOLOK (8 Mar 07)
IKONOS Image of Bukit Tinggi (16 May 02)
IKONOS Image of Bukit Tinggi (08 Mar 07)
(Showing Landslides possibly results of Earthquake)
Sentinel Asia
(Fire Hotspots)
Implementation Progress
● CRISP Hotspot server to host our hotspots is ready; updated after every MODIS pass.
● CRISP server implemented with opensourceMapserver software originally developed by the University of Minnosota in 1994(http://mapserver.gis.umn.edu/)
● URL to access CRISP Webgis: ● http://crisp.nus.edu.sg
TERRA MODIS 12 Jan 07
TERRA MODIS 12 Jan 07
AQUA MODIS 12 Jan 07
AQUA MODIS 12 Jan 07