Evaluating the hydrological utility of four gridded precipitation datasets for streamflow prediction in a mountainous basin

Abstract

<p>Precipitation with high spatio-temporal resolution is one of the critical components of meteorological forcing for hydrological modeling. Its accurate measurement required a large number of rain gauges that are limited for many regions, especially highly elevated domains with a complex and mountainous topography, such as the eastern part of Turkey. On the other hand, open access Gridded Precipitation Datasets (GPDs) varying in spatial and temporal resolutions deliver alternative sources in data-scarce regions. However, their hydrological utilities are to be assessed in different basins to make adequate knowledge for both the developers and end-users. Hence, this study was carried out to investigate the spatio-temporal stability and hydrological utility of four GPDs (MSWEPv2.8, CHIRPSv2.0, ERA5, and IMERGHHFv06) over the upper Euphrates (Karasu) River Basin in the eastern part of Turkey. The accuracy of selected GPDs compared to observed precipitation is expressed in the form of Kling–Gupta Efficiency (KGE), while Hanssen–Kuiper (HK) skill score was utilized to address the detectability strength of GPDs for five different precipitation intensities. Moreover, the hydrological utility of GPDs is evaluated by employing a conceptual hydrologic model under KGE and Nash–Sutcliffe Efficiency (NSE) statistical metrics. Overall, MSWEPv2.8 shows the highest performance (median KGE of 0.34) for the direct comparison with observed precipitation followed by CHIRPSv2.0 (median KGE of 0.34) and ERA5 (median KGE of 0.08) where IMERGHHFv06 shows low performance (median KGE of 0.02) comparatively. Furthermore, CHIRPSv2.0 shows a stable performance for streamflow prediction compared to other Gridded precipitation datasets for the entire period (2015 – 2019), considering two different scenarios. These findings provide guidance for selecting appropriate GPD for the particular region of interest.</p>