In semiarid regions, characterized by low annual rainfall but with occurrence of high intensities, variations in precipitation patterns may increase local runoff and soil erosion. For this reason, spatial and temporal rainfall characterization is important to determine its effects on land surfaces by for example rainfall erosivity. The limitations of rain gauges, especially in developing countries, where they are sparse or data is not always collected, make remote sensing techniques relevant for rainfall data acquisition. Diverse precipitation products from different satellite sensors are available. The precision of these estimates depends on the algorithms employed and the ground data used for calibration. The aim of this study is to compare rainfall estimates from different satellite sensors i.e., the TRMM Microwave imager (TMI) 2A12 and precipitation radar (PR) 2A25 algorithms, the precipitation estimates from the TRMM 3B42 merged HQ/Infrared algorithm and the Multi Sensor Precipitation Estimate (MPE) from the Meteosat SEVIRI and SSM/I sensors. Comparison is made with rain gauge data of Cape Verde, a group of small islands of the west coast of Africa. Cape Verde is a dry semi-arid country subject to very high rainfall variability. Data from Santiago, the largest island of the archipelago, was used for the purpose. A time series comparison for 9 years period was done between 3B42 and ground data, and a single storm was studied for comparing the different satellite estimates, using the ground data as reference. It was found that 3B42 underestimates the amount of rainfall, while for the single storm analysis 2A25 and MPE showed the best similarity when compared to ground data. In conclusion, rainfall satellite products, when complementary to gauge data, can be combined to produce an improved estimate of spatial and temporal rainfall fields, useful for improving agricultural forecasts, water balance and soil erosion evaluations.
展开▼