Evaluation of precipitation products based on rain gauge data and water budget in a mountainous river basin, Eastern Tibetan Plateau

Precipitation in mountainous areas provides abundant water resources for downstream regions, and reliable precipitation data in these areas is of crucial importance for the management of water resources and water-related disasters. Because in-situ precipitation data are usually scarce in mountainous areas, satellite-based precipitation products are expected to play an important role; however, they should be carefully validated before application. This study evaluated the performance of three high-resolution precipitation products in the mountainous Qingyi River basin, by comparison with both rain gauge-based and water budget-based methods. The basin is located at the eastern margin of the Tibetan Plateau, and has high precipitation leading to high runoff (~1100 mm/year). The three precipitation products are CMPA (the China Merged Precipitation Analysis), IMERG (the Integrated Multi-satellitE Retrievals for GPM) and GSMaP (the Global Satellite Mapping of Precipitation). In general, both rain gauge-based and water budget-based methods showed that CMPA has the highest accuracy and IMERG has the poorest accuracy in this region. In two sub-basins with steep terrain and high precipitation, the rain gauge-based evaluation indicated negative or even positive basin-averaged biases of about 1 mm/day or less, but the water budget analysis indicated that all the products had much larger negative biases, of 2.4 ~ 3.8 mm/day. This difference likely arises because the evaluation based on rain gauge data cannot reflect errors in products at the basin-scale, due to the sparse spatial distribution of rain gauges. Finally, observed altitudinal gradients of precipitation were used to correct the precipitation products. Under this approach the water budget can be better closed but is not always satisfactory. Therefore, developing a high-quality precipitation data set for mountainous regions based only on satellite products and sparse ground observations remains challenging and other data sources (e.g. high-resolution meteorological modeling) should be taken into consideration in future.