A comparative analysis of the precipitation extremes obtained from tropical rainfall-measuring mission satellite and rain gauges datasets over a semiarid region

The objectives of this research were to compare precipitation extremes obtained from Tropical Rainfall-Measuring Mission (TRMM) satellite and those of rain gauges over a semiarid area in Iran. Extreme precipitation indices (EPIs) (i.e., the number of days with a precipitation value over 10 mm, the maximum duration of wet and dry days, the number of days with precipitation over the 95th percentile, total precipitation higher than the 95th percentile, and maximum daily precipitation) were calculated across Fars province, Iran, 2000-2014 on seasonal time scales. The gauges data were interpolated at a spatial resolution of 0.25 x 0.25 degrees to match the 3B42 data using Inverse Distance Weighting (IDW). Then, EPIs from the two datasets were compared with each other. The findings showed that mean values computed from gauges and satellite data did not present any significant differences among all of the extreme indices. Furthermore, their variances presented a good level of congruence. Finally, the majority of indices presented a satisfactory correlation between the two dataset. To evaluate the prediction of extreme events in different temporal and tolerated distances, a fuzzy method was used. The results showed that the percentage of grid cells with useful predictions tripled with spatial tolerance extending by just one pixel. To evaluate methods of eliminating the uncertainty of probable missing rainfall data and the seasonal changes in rainfall averages, probabilistic methods based on Weibull distribution and truncated geometric distribution (TGD) were employed to eliminate uncertainties in estimation of extreme precipitation amounts and extreme wet periods (WPs). The results showed that as to extreme precipitation amounts, a satisfactory method could not be drawn for arid southern regions of Fars, Iran. Similarly, as to extreme WPs, the consistency between gauges and satellite data could not be improved significantly.