Annual and diurnal African biomass burning temporal dynamics

Abstract. Africa is the single largest continental source of biomass burning emissions. Here we conduct the first analysis of one full year of geostationary active fire detections and fire radiative power data recorded over Africa at 15-min temporal resolution and a 3 km sampling distance (at the sub-satellite point) by the SEVIRI imaging radiometer onboard the Meteosat-8 satellite. We use these data to provide new insights into the rates and totals of African open biomass burning, particularly into the extremely strong seasonal and diurnal cycles that exist across the continent. We find peak daily biomass combustion totals are 9 and 6 million tonnes per day in the Northern and Southern Hemispheres respectively, and total fuel consumption between February 2004 and January 2005 is at least 855 million tonnes. Analysis is carried out with regard to fire pixel temporal persistence, and we note that the majority of African fires are detected only once in consecutive 15 min imaging slots, indicating the importance of optimizing the fire pixel detection strategy performance. An investigation of the variability of the diurnal fire cycle is carried out with respect to 20 land cover types, and whilst differences are noted between land covers, the diurnal characteristics for a given land cover type are similar in both African hemispheres. We compare the FRP-derived biomass combustion estimates to burned-areas, both at the scale of individual fires and over the entire continent at a 1-degree spatial scale. Fuel consumption estimates are found to be less than 2 kg/m2 for almost all land cover types, and for savanna grasslands where literature values are commonly reported the FRP-derived median fuel consumption estimate of 309 g/m2 appears reasonable. From mid-2008, geostationary FRP data of the type presented here will become available to interested users continuously and in near real-time from the EUMETSAT Land Surface Analysis Satellite Applications Facility (LandSAF), allowing the types of analysis presented in this paper to be undertaken on multi-year datasets where relationships between climate variables, active fires and fuel consumption can be further examined.