Evaluating satellite‐ and reanalysis‐based precipitation products over the <scp>Qinghai‐Tibetan</scp> Plateau in the perspective of a new error‐index system

Abstract

Abstract The necessity of evaluating satellite‐ and reanalysis‐based precipitation products as optional proxies of precipitation observations is widely understood. Four typical satellite‐ or reanalysis‐based precipitation products from TRMM, CPC, CMFD and ERA5‐Land datasets were evaluated at temporal and spatial scales using a new error‐index system method across the Qinghai‐Tibetan Plateau (QTP) from 1979 to 2018. The error‐index system consisted of 10 indices, which were classified into two categories, that is, the comprehensive error indices and error component indices. A local error decomposition scheme suitable for the QTP was built according to the simulation test. Results showed the datasets except ERA5‐Land performed better at monthly scale than that at daily scale generally, though significant differences were found among the products. Particularly, CMFD achieved the highest accuracy in estimating monthly and daily precipitation, followed by CPC, while TRMM performed similarly to ERA5‐Land. It was estimated that over one fifth of the precipitation were falsely recorded in the four products. Approximately over 15% of daily precipitation were overestimated or underestimated with an accumulated extent reaching up to 60 mm per month in summer. Spatial results indicated that large deviations were found in the southeast QTP, which were derived from a significant underestimation. Besides, a large proportion of false and missing estimation of precipitation events were witnessed in the northeast river basins. Through the comparison, CMFD and CPC had reduced all error components except HNB in both magnitude and the ratio. And the datasets tended to overestimate the slight precipitation and underestimate the heavy rainfall. And the products did not eliminate the large deviations in the southeast and north QTP. The assessment results reported here will provide some valuable feedbacks for the improvement of these products.