Effects of climate variability on savannah fire regimes in West Africa

Abstract. The main objective of this work is to investigate at regional scale the variability in burned areas over the savannahs of West Africa and their links with the rainfall and the large-scale climatic indexes such as the Southern Oscillation Index (SOI), Multivariate ENSO Index (MEI), North Atlantic Oscillation (NAO) and sea surface temperature gradient (SSTG). Daily satellite products (L3JRC) of burned areas from the SPOT Vegetation sensor at a moderate spatial resolution of 1 km x 1 km between 2000 and 2007 were analyzed over the West African savannah in this paper. Results from seasonal analysis revealed a large increase in burned areas from November to February, with consistent peaks in December at the regional scale. In addition, about 30% of the pixels are burned at least four times within the 7-year period. Positive correlations were found between burned areas and rainfall values obtained from the TRMM satellite over savannahs located above 8° N, meaning that a wet rainfall season over these regions was favorable to biomass availability in the next dry season and therefore may induce an increase in burned areas in this region. Moreover, our results showed a nonlinear relationship between the large-scale climatic indexes SOI, MEI, NAO and SSTG and burned-area anomalies. Positive (negative) correlations between burned areas and SOI (MEI) were consistent over the Sahel and Sudano-Sahelian areas. Negative correlations with Atlantic SSTG were significant over the Guinea subregion. Correlations between burned areas over Sudano-Guinean subregion and all the large-scale indexes were weak and may be explained by the fact that this subregion had a mean rainfall greater than 800 mm yr⁻¹ with permanent biomass availability and an optimal amount of soil moisture favorable to fire practice irrespective of the climate conditions. The teleconnection with NAO was not clear and needed to be investigated further.