Compositing of modis images for burnt area detection in Southeast Asia

Abstract

This paper presents preliminary results on the usability of different compositing algorithms for burnt area detection in Southeast Asian conditions. A cloud removal method using only reflective bands is created and six monthly composites of MODIS surface reflectance product (MOD09) images are built using four different algorithms. According to the results, minimum NIR and minimum NBR (i.e. (band 2-band 7)/(band 2+band 7)) composites produce higher separability between burnt and unburnt areas than maximum NDVI (i.e. (band 2-band 1)/(band 2+band 1)) and maximum band 7 composites and are thus considered more suitable for burnt area detection purposes. In addition it is worth noticing that even though monthly compositing period is used, the average number of available clear observations per pixel is very low varying between 2.5 and 6.2 for the six composites.