Rainfall in tropical regions is mostly of convective origin. With the objective of building climatological statistics on propagation of African cloud clusters, and studying their morphological and radiative properties, an automatic method for tracking cloud clusters has been developed. Using Meteosat infrared full resolution images, the method allows for the tracking of systems larger than 5000 km2 during at least one hour. A cloud cluster is defined as adjacent pixels colder than a brightness temperature threshold such as: 213 K, 233 K, 253 K. The method is based on the cluster overlap between two successive images and takes into account splitting and merging of clouds. It allows for an objective determination of parameters such as: size, mean temperature, temperature variance, duration, coordinates of the centre of gravity, speed, trajectory, eccentricity etc. In order to use the full space resolution, the tracking is limited to West Africa during the rainy season (June to September). Nine years have been processed, from 1989 to 1997.
In a first part, we use this nine year satellite dataset to study the interannual variability of the West African convective systems. For instance, we consider the number and the cloud cover corresponding to different types of convective systems such as Mesoscale Convective Complexes (see Maddox (1980)), long lived or fast convective systems. In 1997, we observed an important decrease of the number of fast and long lived convective systems that could be a consequence of the last El Nino event.
In a second part, we focus on the small area of the dense EPSAT- Niger recording raingauge network. Combining satellite and ground based data from 1990 to 1997, we study the relationships between rainfall and convective systems characteristics