K. Javan

Satellite rainfall estimate systems can produce data for areas for which other sources are unavailable. Without referring to ground measurements, satellite-based estimates cannot be used directly before bias correction. This study aims to evaluate the performance of the TRMM-3B43 V7 rainfall products in the Northwest of Iran. The evaluation was carried out using monthly data obtained from 21 meteorological stations during 1998-2015. The monthly and annual spatial distributions of the Pearson correlation coefficient between the station and satellite-based observations as well as the statistical error measure were calculated. The results revealed that the correlations between TRMM 3B43 and rain gauge data were high (R> 0.80) in October and small (R < 0.55) in August. Moreover, mean annual spatial distribution of R showed that the small values of R occurred in the Northeast and the southern mountainous regions. We found that the TRMM-3B43 V7 overestimated rainfall in the Northeast, in September, October and November (>30 mm) and in December to May (>20 mm). On the contrary, an underestimation was found in the Southwest regions where summer season (June, July & August) is generally characterized by small anomalies in terms of R values. Generally, the satellite products applied in this study underestimate higher rainfall values while showing overestimation for lower rainfall records. The measure of Root Mean Square Error (RMSE) showed that a large spatial variability takes place in September, October and November in most of the stations, particularly when rainfall records is less than 50 mm. Meanwhile, the lowest variability occurred in June, July and August with a slight increase in the Northeast. Our findings imply that satellite products have poor performance for estimating higher rainfalls in the Northwest of Iran on a monthly scale.