In an effort to get a better understanding of the extent and patterns of burning in South America, geostationary satellite data have been used to monitor active fires. Previous work demonstrated the ability to manually detect subpixel fire activity in selected areas of the selva and cerrado regions in South America with shortwave and longwave infrared data available from the Geostationary Operational Environmental Satellite (GOES) visible infrared spin scan radiometer atmospheric sounder (VAS). This paper presents the GOES VAS automated biomass-burning algorithm (ABBA) which was developed to determine basin-wide trends in fire activity in South America utilizing the GOES VAS archive. Comparisons between the manual and the automated techniques revealed tht overall, the ABBA proved to be more consistent in identifying fires and better suited for trend analysis. The automated algorithm was applied daily to a study area extending from 5¿S to 15¿S and from 45¿W to 70¿W for 2 weeks at the peak of the burning seasons in South America in 1983, 1988, 1989, and 1991 in an effort to measure the areal extent of burning in South America during the past decade and to provide additional insight into the diurnal signature in satellite detection of biomass-burning activities.

The expansion of the regions of burning are readily detected in a comparison of these 4 years. From 1983 to 1991 the amount of burning detected by the GOES VAS ABBA during these 2-week periods nearly doubled in the selva and mixed regions and tripled in the cerrado. Diurnal analyses confirmed earlier results indicating that the optimum time to monitor biomass burning is around 1530 UTC. ¿ American Geophysical Union 1994