Assessing the fidelity of gridded datasets for characterizing precipitation over the Tibetan plateau using multi-source observations and water balance

Abstract There is a significant interest in how much water enters the Tibetan Plateau (TP), as it is the source of many rivers in Asia. However, owing to scarce station observations and large uncertainties in multi-source precipitation datasets, this question still remains obscure. Here, we assess a high-resolution merged precipitation dataset (TPHiPr) for characterizing annual precipitation and its changes over the TP based on station observations, water balance in six river basins, and snow water equivalent of 10 glaciers. The results show that TPHiPr is superior to CRA-LAND, ERA5, HARv2, and GPCC when they are assessed using independent precipitation observations, water balance of basins and snow water equivalent of glaciers. TPHiPr shows that average annual precipitation is 596 mm over the TP during 1981–2020, i.e., approximately 1824 billion tons of water entering the TP (regions above 2500 m) in terms of precipitation every year. Annual precipitation increases by 11.3 mm 10a − 1 from 1981 to 2020. From a spatial perspective, annual precipitation significantly increases in the Qiangtang Plateau (Inner TP) as well as the eastern and western regions of the Qaidam Basin, whereas it decreases in the Tajikistan glacier area in western TP, partial areas in the western Himalayas, and southeastern TP. On the basin scale, annual precipitation significantly increases in the upper Yellow and Yangtze basins but remains relatively stable in the upper Mekong, Salween, Indus and Brahmaputra. These results provide important references for evaluation of water resource over the TP.