SEASONALITY OF VEGETATION FIRES IN AFRICA FROM REMOTE-SENSING DATA AND APPLICATION TO A GLOBAL CHEMISTRY MODEL

This paper sets out to show the potential use of remote sensing of act ive vegetation fires for continental- to global-scale modeling of biom ass burning studies. It focuses on the analysis of the seasonality of vegetation fires for the African continent, as derived from NOAA-AVHRR -GAC-5km satellite data. These data are ideally suited for savanna fir es, which constitute between 60 and 80% of the biomass burnt in Africa . Monthly counts of fire pixels, within 1 degrees latitude x 1 degrees longitude grid cells, over continental Africa have been calculated fr om November 1984 through October 1989. These 1 degrees grid cells are summated to a 5 degrees x 5 degrees grid to enable comparison with pre vious studies and are analyzed at this resolution to show various feat ures of the fire season. The analysis shows that previous attempts to characterize the seasonality of biomass burning have tended to underes timate the intensity of the peak months of burning or have predicted t oo long a fire season in certain areas. It also shows that there can b e, for a given area, a temporal shift in the timing of the fire season from year to year. Such an interannual variability of fire seasonalit y makes satellite data more appropriate than statistical data for the modeling of atmospheric transport of vegetation fire products and the comparison with experimental measurements. Modeled values of black car bon mass concentration from a global transport model (MOGUNTIA), using the seasonality of biomass burning as an independent variable, are co mpared with measurements taken at Amsterdam Island (38 degrees 30'8, 7 7 degrees 30'E) and Lamto, Ivory Coast (6 degrees N, 5 degrees W). Alt hough 5-year averaged satellite data were used, the seasonality as der ived from satellite data determined in this paper gives modeled values of black carbon mass concentration that are in good agreement with th e measurements.