Intercomparison of gauge based, reanalysis and satellite gridded precipitation datasets in High Mountain Asia: insights from observations and runoff data.

The Tibetan Plateau and the adjacent mountain ranges are known as the "Asian Water Tower" (AWT) because they hold the third largest frozen water reserve in the world after the polar regions. This region plays a vital role in supplying water to nearly 2 billion people through rivers like the Indus, Ganges, Brahmaputra, Yangtze, and Yellow River. 
Accurate precipitation data are essential for understanding hydrological processes in high mountain basins. However, in many mountainous areas, precipitation gauges are either sparse or absent due to the challenging environmental conditions. Moreover, the available precipitation gauges are often located in valleys and they are not adequate to represent the diverse topography of the territory. This underlines a significant gap in the existing precipitation datasets, since precipitation at high elevations is likely considerably underestimated.
In this study, we aim to address these challenges by analyzing an extensive area of High Mountain Asia (70°-100°E for longitude and 25°-40° N for latitude). Specifically, we examined two reanalysis datasets (ERA5 and HAR), two gauge-based datasets (GPCC and Aphrodite), and one satellite-derived dataset (PERSIANN) to evaluate their performance in capturing precipitation patterns. 
At first, we compared the different datasets over the common period (1983-2007) evaluating their ability to reproduce the precipitation spatial distribution both at annual and seasonal level.
Then, due to the discrepancies in precipitation values over the area, particularly influenced by the complex orography, we decided to compare the datasets with the observational data available from the Copernicus Data Store (Global Land Surface Atmospheric Variables dataset, 1755–2020) and the runoff data provided by the GRDC (Global Runoff Data Centre) dataset as a reference.
When comparing gauge-based datasets with the observational data, there is consistency, whereas the other datasets tend to exhibit higher precipitation especially in areas with greater topographic complexity.
To compare precipitation values with the measured river flow, the total evaporation from the ERA5-Land dataset was taken into account to improve the estimates. The results indicate that reanalysis datasets are the most effective in simulating the hydrological balance while the gauge-based and the satellite datasets significantly underestimate precipitation.