Adequacy of satellite derived data for streamflow simulation in three Hexi inland river basins, Northwest China
Chunlin Wang,Jianhua Si,Chunyan Zhao,Bing Jia,Sindikubwabo Celestin,Duan Li,Xiaohui He,Dongmeng Zhou,Jie Qin,Xinglin Zhu,Chunlin Wang,Jianhua Si,Chunyan Zhao,Bing Jia,Sindikubwabo Celestin,Duan Li,Xiaohui He,Dongmeng Zhou,Jie Qin,Xinglin Zhu
DOI: https://doi.org/10.1016/j.atmosres.2022.106203
IF: 5.965
2022-08-01
Atmospheric Research
Abstract:The limited gauge measurements usually constrain the accurate hydrological modeling of runoff, especially in the ungauged or gauge-sparsely areas with complex terrain conditions. Satellite derived data provide alternative forcing data for hydrological modeling in ungauged areas, while quantitative evaluation on their performance is still needed. This study explores the adequacy and accuracy of satellite derived data based on the gauge observations over three inland river basins in the arid Hexi Corridor, Northwest China, where the satellite-based products are adopted to drive the HBV-Light model, including six satellite precipitation datasets (CHIRPS, CMORPH, GPM, GSMaP, MSWEB, PERSIANN) and the widely used evaporation product (GLEAM) combing with MODIS land surface temperature product (MOD11A1). Seven statistical metrics (CC, ME, RMSE, BIAS, POD, FAR, CSI) are adopted to assess satellite-based precipitation estimates, and three indices (NSE, R2, PBIAS) are used to evaluate the hydrological performance. Results indicate that: (1) For precipitation evaluation at point-to-pixel scale, all products have relatively poor agreement with gauged observations with the CC ranging from 0.16 to 0.56. Both MSWEB and GPM outperformed other products at daily scale with higher CC values, and showed the capability of detecting the precipitation events with higher POD and CSI. PERSIANN, GSMaP, GPM, CMORPH tend to underestimate the precipitation amount in three stations. (2) For the hydrological evaluation, the statistical assessment showed that the GPM presents the best results at daily scale, while underestimates streamflow amount in three watersheds. All the grided precipitation datasets show higher NSE and R2 in the relatively humid HHRB and XYRB than SLRB. (3) The combination of MOD11A1, GLEAM and satellite-based precipitation as forcing data of HBV-Light can achieve good performance in the three inland river basins. This study aims to provide additional reference for the hydrological application of satellite derived data in gauge-sparsely areas located in high altitude and arid regions.
meteorology & atmospheric sciences