Development and testing of algorithms for a global burnt area product from ERS ATSR-2

A set of algorithms capable of detecting burnt areas from vegetation fires at a global scale, across the major biomes have been tested in order to adopt a single algorithm. The different algorithm limitations are quantified considering the errors of commission/omission and over/under estimation of burnt areas. The main ecosystems studied were the dense humid tropical forest, the savanna-forest interface, Mediterranean forest and shrub land, boreal forest and tundra. Algorithm evaluation was performed against high resolution satellite imagery (mostly from Landsat TM) as reference data.