*3.5. Evaluation of SM2RAIN-CCI*/*ASCAT against TMPA*

Previous studies [47,59,67] reported the better performance of TMPA across Pakistan than other PPs, including TRMM-3B42RT, PERSIANN-CDR, PERSIANN-CCS, and CMORPH. Therefore, TMPA was selected to comprehensively assess the performance of SM2RAIN-CCI and SM2RAIN-ASCAT. Three statistical metrics, i.e., Bias, ubRMSE, and Theil's U are used to assess the performance of SM2RAIN-CCI/ASCAT against TMPA during the common period of 2007–2015. Figure 5 shows a comparison of performances of three precipitation products using box-plots. SM2RAIN-CCI data was not available for the glacial region, therefore this region is neglected.

In the humid region, SM2RAIN-CCI underestimated precipitation (negative Biases) and SM2RAIN-ASCAT under/overestimated precipitation (slightly positive Biases), however, TMPA had completely overestimated (positive Biases) precipitation. Moreover, the SM2RAIN-based product showed over/underestimation in the arid region while TMPA overestimated precipitation. A contrasting scenario is observed in the hyper-arid region where TMPA underestimated, and SM2RAIN-CCI/ASCAT overestimated precipitation. Overall, smaller biases are observed when TMPA is compared with RGs, in comparison with SM2RAIN-CCI/ASCAT.

In the case of ubRMSE, TMPA outperformed the SM2RAIN-ASCAT in all three climate regions, while both of them were superior to the SM2RAIN-CCI. The ubRMSE of SM2RAIN-ASCAT and TMPA were almost comparable in humid and hyper-arid regions. Moreover, ubRMSE decreased significantly from humid region to hyper-arid region, which might be associated with the precipitation magnitude, intensity and any other hydrological (hydraulic) activities in these climate regions.

Among these three PPs, Theil's U coefficient depicts the best performance of TMPA, followed by SM2RAIN-ASCAT and SM2RAIN-CCI. Theil's U shows higher variation (as well as best forecasting accuracy) in the arid region. Overall, TMPA dominated the SM2RAIN-based products across all Pakistan. SM2RAIN-ASCAT and SM2RAIN-CCI need proper bias adjustments before integrating into any hydrological applications.

**Figure 5.** Daily box-plots for Bias (%), ubRMSE (mm/day) and Theil's U for SM2RAIN-CCI (black box) and SM2RAIN-ASCAT (blue box) against TMPA (red box) during the period 2007–2015. The line in each box indicates the median values and each box (top and bottom lines) depicts the 25th and 75th percentile, while the lines perpendicular to boxes (whiskers) extend to the extreme data points.
