Large-scale forest fires in Siberia analysed by MODIS, MERIS and ASTER multiresolution satellite imagery

Siegert, F. and Huang, S.

in: proceedings of Envisat 2004 Symposium, Salzburg, Austria, September 6-10, 2004

2004

Exceptionally large fires were observed in Southern Siberia in 2003. Persistent cloud cover and orbit parameters did not allow observation of large areas with high resolution imagery shortly after the fires. Therefore we used a combination of high, medium and low resolution satellite imagery in which Moderate Resolution Imaging Spectroradiometer (MODIS) was used to record fire occurrence each day over large areas, while multispectral MEdium Resolution Imaging Spectrometer (MERIS) and Advanced Spaceborne Thermal Emission and Reflection (ASTER) imagery were used to determine the burnt area in cloud-free subsets of our study site. A regression analysis showed that there were linear correlations between burnt areas derived from ASTER and MERIS imagery as well as between MERIS imagery and MODIS hotspot data. We then extrapolated the detailed results on burnt areas obtained from ASTER and MERIS to the full range covered by MODIS hotspots data. The analysis showed that the burnt area derived from MODIS hotspots resulted in an underestimate of approximately 40%. To determine the land cover types affected by fire we intersected the burnt area with the GLC2000 land cover map. 71.4% of the burnt area was forest, 10% was humid grasslands. Large scale wildfires like the one observed in 2003 amplified by climate extremes will diminish the carbon sequestration potential of this largest resource of boreal forest worldwide and thus to a net carbon release to the atmosphere.