Validation of satellite precipitation product at an arid-semiarid basin with complex terrain properties

Satellite based High Resolution Precipitation Product (HRPP) (generally with a temporal resolution of 3h and spatial resolution of 0.25°×0.25°) is a valuable data source in global and regional hydroclimatologic applications. However uncertainty information for most of the satellite-based HRPPs is unavailable up to present. This paper presents a statistical evaluation study of daily TMPA research time precipitation products at an arid-semiarid basin with complex terrain properties in the northwest of China. Both local scale and regional scale evaluation studies are conducted utilizing gauge based datasets. Total 15 evaluation indices are selected for this study including 5 traditional quantitative statistical indices (CC, RMSE, ME, MAE and BIAS) and 5 categorical statistical indices (POD, FAR, PFB, CSI, and ETS) as well as 5 indices for bias/error decomposition evaluation(HB, MB, FB, RMSEs and RMSEr). New findings about the uncertainty characteristics of TMPA research time precipitation products can be concluded into the following three aspects: (1) both local and regional evaluation results demonstrate that the performance of daily TMPA products is climatology-dependent. For the middle-upper reaches of Heihe river basin, better performance is observed at the southern mountainous area where the climatology is wetter even with consideration of significant terrain effect which is caused by complex topographical variation there. Different bias/error characteristics would appear in different climatology. For the study basin, TMPA products overestimate precipitation in wetter region while make underestimation estimation in drier area. This implies nonuniform bias correction schemes for TMPA products may lead to accuracy improvement. (2) Performance of TMPA products is much worse in snowfall estimation than rainfall estimation and thus one should be cautious to use TMPA daily products in cold region or winter time. (3) little improvement in the per- ormance of TMPA products on daily scale has been made by transition from version 6 to version 7. This study can serve as a reference of the uncertainty in TMPA research time precipitation products in hydrometeorogical applications on watershed scale.