The limitations and potential of AVHRR GAC data for continental scale fire studies

Abstract

Global Area Coverage (GAC) data from the Advanced Very High Resolution Radiometer (AVHRR) are available on a daily basis, dating back to July 1981. The AVHRR's 3·55–3·93 μm channel is suitable for detection of terrestrial hot spots, such as bushfires. The long-term archives and global cover make the GAC a potentially valuable data source for large scale fire studies. However, these data are sampled spatially through a combination of line skipping and averaging. This study shows that the sampling affects the sensitivity of GAC for fire detection in relation to ecosystem and season. The GAC are found to provide a reasonable measure of fire activity in grassland and open b'ush savannah, but to perform poorly in the forest margins. Overall at least 79 per cent of fires detected with non-sampled AVHRR data are missed by the GAC. This severely limits the use of GAC data for quantitative fire studies. The GAC does appear to provide a reasonable measure of fire calendar (i.e., variations in fire activity with time) and on a continental scale successfully identifies the main regions of fire activity. The potential of these data for continental scale fire studies is illustrated through the preliminary analysis of 277 GAC mosaics of Africa for the period September 1988 to August 1989.