The precipitation echoes collected by non-Doppler meteorological radar are identified and tracked in the covered area. For that a sequence of images, recorded every 5 min by S-band radar in Bordeaux and previously filtered to remove the ground clutter, is considered. In these images, the radar echoes are labeled as precipitation cells and processed using the method of sum and difference histograms of gray levels. Textural parameters are extracted from these images by slicing an analysis window of . Energy and homogeneity are found to be the best discriminating parameters because each of them clearly assigns the radar echoes to either stratiform or cumuliform clouds. The convective cells mainly differ from the stratiform ones by their texture and the high values of their reflectivity factor. To account for the downpour development, the time variations of barycenter, surface area, and reflectivity factor have been analyzed for the precipitation cells in the sequence of radar images under consideration. In the case of cumuliform cells having reflectivity factor higher than 40 dBZ, the expansion of their surface area and their progress in the observed region constitute important information about the clouds leading to weather extremes.