SPOT-‐4 BURN AREA MAPPING

Transcription

SPOT-‐4 BURN AREA MAPPING
SPOT-­‐4 BURN AREA MAPPING Parwati Sofan,
Any Zubaidah, Yenni Vetrita, Fajar Yulianto, Kusumaning Ayu DS
Remote Sensing Applica1on Center Na1onal Ins1tute of Aeronau1cs and Space (LAPAN) www.lapan.go.id APAN MEETING, 21 JANUARY 2014 Outline Introduc1on LAPAN Ac1vi1es Related to Forest Fire State of the Art: Satellite Data & Methods Research Objec1ve Methodology Results and Disscussions Conclusions Introduc;on The forest fire, including peatland fires, have become a more regular feature during dry season in Indonesia. Peatland fires occur mainly on peatland with unclear ownership and responsibility. They are mostly triggered by human ac1vi1es in or around peatland (Joosten et al. 2012). The economic cost of the peatlands fire impact: q  CO2 emission around 30% (27 % from peat fires) of the annual global average emissions, during El Nino in 1997 (IPCC, 2000) q  smoke haze pollu1on and carbon emissions in 1997/98 would range between $5.1 billion and $6 billion (Taconi, 2003) The informa1on of peatlands fire is important to manage and restore the peatlands condi1on. Remote sensing technology has been developed and commontly used to monitor the global informa1on on forest fire LAPAN ac;vi;es One of task and func1ons of Remote Sensing Applica1ons Center is to do R & D of remote sensing applica1ons for environment monitoring, disaster mi1ga1on, and dissemina1on of disaster informa1on. Forest/Land fires researchs in LAPAN are: q  Fire hotspot monitoring q  Burn area detec1on Facili1es à Ground Sta1on à SPOT-­‐4. It could be used for forest fire monitoring in local scale Satellite data Satellite data for burn area mapping : –  MODIS (Maren et al. 2002, Roy et al. 2002, 2005, Sá et al. 2003, Chuvieco et al. 2005, Miednen, J. 2007, Suwarsono et al. 2010), –  NOAA-­‐AVHRR (Barbosa et al. 1998, Frazel et al. 1999, Roy et al. 1999, Fuller and Fulk 2001, Nielsen et al. 2002), –  SPOT VEGETATION (Stroppiana et al. 2002, 2003, Silva et al. 2004) –  Landsat (Conese & Bonora. 1999, Wagtendonk et al. 2004, Cocke, A.E., et.al. 2005 ). Satellite data with low spa1al resolu1on (250 m – 1000 m) have limita1on to detect small burn scars that oben exist in Indonesia.
(Miendnen, 2007). Burn area index (Ep1nga et al, 2005; Cocke, 2005; Escuin, 2007): –  Single band : NIR(band4), Mid IR (band 7), BT (band 6) Landsat –  Ra1o band : band (7/5), band (7/4), band (4/5) Landsat –  Vegeta1on Index : Normalized Difference VegetaCon Index (NDVI), Normalized Burn RaCo (NBR), Soil-­‐Adjusted VegetaCon Index (SAVI), Modified Soil-­‐Adjusted VegetaCon Index (MSAVI), –  Mul1variate component: Principle Component 2, Tasseled Cap-­‐greenness, Tasseled Cap-­‐wetness. Compara;on of methods Methods Weakness/Advantages Band combina1ons: Red and SWIR (5 or 7) Can be used to es1mate the water content (Avery and Berlin 1992; of vegeta1on Eva and Lambin 1998). Normalized Difference NDVI was less accurate if the fires Vegeta1on Index (NDVI) occurred aber the vegeta1on died or in the region which had less vegeta1on before the fires occurred Single band dan ra1o Any methods that use visible band will be band with the visible and effected with atmospheric interference infrared channels (cloud , smoke haze) High accuracy for burn area mapping SWIR NBR (NIR, SWIR) Reference (Salvador et al. 2000); (Eva and Lambin 1998) (Rogan and Yool 2001) (Avery andBerlin 1992; Eva and Lambin 1998). White et al. 1996; Roy et al. 1999; Rogan andYool 2001). •  has high correla1on compared with Ep1nga et al, 2005; other methods. Cocke, 2005; Escuin, •  The most suitable index related to the 2007 sensi1vity of SWIR reflectance and NIR (Key, 2006) reflectance for fires in vegeta1on area Spectral reflectance for vegeta;on ↓ water content. ∼ ↑ SWIR SWIR can be used to detect the water stress in vegeta;on and burn vegeta;on Sumber:
Univ. of British Columbia Canada
www.forestry.ubc.ca
NIR
SWIR
Research objec;ve •  To analyze the characteris;c of SPOT-­‐4 channels in burn area •  To verify the use of burn area indices (NDVI and NBR) for burn area mapping in Riau Province Methodology Data : SPOT-­‐4 from LAPAN Ground Sta1on in 2009-­‐2011. ü  mul1spectral band XS1:GREEN (0.50 – 0.59 µm), XS2:RED (0.61 – 0.68 µm) XS3: NIR (0.78-­‐0.89 µm), XS4:SWIR(1.58 – 1.75 µm) ü  resolusi spasial 1nggi (20 m) ü  Path/row : 270/347, 271/350, 272/350 Others data: MODIS Hotspot in 2009-­‐2011, Digital Eleva1on Model (DEM) SRTM Flowchart SPOT -­‐4 Data HOTSPOT-­‐MODIS 2009-­‐2011 Data selec;on Pre-­‐fire Post-­‐fire Pre-­‐processing: GEOMETRIC , ORTHO, RADIOMETRIC Correc;ons, Cloud masking NDVI, NBR dNDVI, dNBR VERIFICATION using Ground survey data Burn area mapping Burn area index •  Cloud masking: thresholding •  Normalized Difference Vegeta1on Index NDVI = (ρnir -­‐ ρred)/(ρnir + ρred) •  Normalized Burn Ra1o (NBR) NBR = (ρnir -­‐ ρswir)/(ρnir + ρswir) •  Burn area mapping : dNBR = NBR prefire – NBR posqire dNDVI = NDVIprefire – NDVI posqire criteria: Vegeta1on index in burn area will have higher value before fires occurred rather than aber fires, so the difference between pre and post fire will be posi;ve value Study area DUMAI
study area
PELALAWAN
RGB SPOT-­‐4 in Dumai Burn area RGB SPOT-­‐4 and MODIS-­‐Hotspot Okt 2010
Nov 2010
Des 2010
AQUA 8 May 2011 - 13.05 WIB
Jan 2011
TERRA 9 Mei 2011 - 10.55 WIB
Mei 2011
Feb 2011
Jun 2011
Mar 2011
Apr 2011
Jul 2011
Ags 2011
Black dots refer to Hotspot
Feb, May – Jun 2011
Reflectance value of SPOT-­‐4 in post and pre fire NIR and SWIR
are capable to
separate
vegetation in prefire and post-fire
condition
post-­‐fire: §  Damaged vegeta;on detected by low of NIR reflectance §  burn area has darker colour and the increasing temperature can increase SWIR reflectance Burn area in Pelalawan using dNBR RGB SPOT-4
07/05/2011
RGB SPOT-4
24/07/2011
dNBR
location:
Langgam
Pelalawan
Verifica;on in Pelalawan RGB SPOT-4
dNBR 07/05/2011 – 24/07/2011
19-08-2011
19-08-2011
Ground location
in Langgam Pelalawan
Verifica;on in Pelalawan Sample area: P-1
Burn area in Dumai using dNBR RGB SPOT-4
16/10/2010
RGB SPOT-4
27/06/2011
dNBR
Bukit Kapur
Dumai
Verifica;on in Dumai RGB SPOT-4
dNBR 16/10/2010 – 27/06/2011
21-08-2011
21-08-2011
Ground location
in Bukit Kapur Dumai
dNDVI and dNBR in Burn area (Dumai) dNDVI and dNBR in opening land without burning (Pelalawan) Sample Area: P-2
NBR and NDVI Extrac;on value ü 
Neither NDVI or NBR in pre-fire has higher value than in post-fire
ü 
in burn area: dNBR > dNDVI.
ü 
P-2 Sample (opening land without biomass burning) : dNBR < dNDVI.
Comparison of dNBR dan dNDVI in burn area and no biomass burning Burn area No burning ü  dNBR is more sensitive to identify burn area
ü  dNDVI is more sensitive to detect opening land from vegetation to non
vegetation without biomass burning.
Conclusion ü  The SWIR and NIR reflectance of SPOT-­‐4 is a good indicator to separate the pre-­‐fire and post-­‐fire vegeta1on. ü  The dNBR is suitable for mapping the burn area in peatland of Riau Province, while the dNDVI is more sensi1ve to detect opening land from vegeta1on to non vegeta1on without biomass burning. . 

Similar documents