Validation of an infrared-based satellite algorithm to estimate accumulated rainfall over the Mediterranean basin

The potential of an infrared-based satellite rainfall algorithm, the well-known Convective-Stratiform technique (CST), to estimate accumulated rainfall in the Mediterranean basin is tested. The CST, calibrated by coincident, physically retrieved rainfall rates from the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR), is applied over the central-eastern Mediterranean region for the twelve-month period September 2004-August 2005. Estimates from this technique are verified over a 1° × 1° gridded precipitation dataset, based on rain gauge data only, for different time scales (monthly, seasonal and annual). The comparisons between satellite-derived precipitation estimates and validation data provide a high correlation coefficient (0.88) and low biases only for the summer season. In contrast, the comparison statistics for winter demonstrate the shortcomings of the CST algorithm in reproducing adequately the precipitation field in the mid-latitudes during this season. Although the correlations for spring and annual precipitation are relatively high (0.76 and 0.73, respectively), a strong positive bias exists. Rainfall variability is less adequately reproduced for the autumn, but the errors are within an acceptable range. A comparison test conducted in the different climate zones of the study area indicated that the calibrated CST performs better in the sub-tropical deserts and steppes of northern Africa and in humid, continental climates. Mediterranean climates produce higher correlations for autumn, summer and spring precipitation, whereas humid sub-tropical climates present the lowest correlation coefficients. Finally, the potential of the CST technique in climatic studies was demonstrated by studying the diurnal variability of precipitation at high spatial and temporal resolutions.