Fire is a global phenomenon, acting as an important disturbance process. Africa is one of the continents that has higher fire density, particularly in savanna regions, making it the subject of innumerous studies about fire regime and behavior. Here, a new method of time series analysis called Empirical Mode Decomposition (EMD) was applied to monthly fire counts time series from MODIS Terra/Aqua sensors. The goals were to analyze the differences between the time series from the two instruments (MODIS Terra and Aqua), the differences in the behavior of the active fire time series from the north and south parts of Africa and they‟re relationships with climatic modes (ENSO and IOD). For most of the time series, the application of the EMD resulted in four IMF‟s and a residue. Although there is always an IMF related with seasonality, the physical meaning of the other isn‟t clear. This may be due to various reasons, some related with intrinsic problems of the method, other with the applicability of the method to this type of series.

Mestrado em Engenharia Florestal e dos Recursos Naturais - Instituto Superior de Agronomia