A high-resolution near-surface meteorological forcing dataset for arid Xinjiang

We present a hybrid framework that combines regional modelling with data assimilation and station-informed bias correction to produce a physically coherent, observation-constrained near-surface meteorological dataset for Xinjiang for 1960–2020. First, a regional modelling system is used to obtain spatially continuous and dynamically consistent fields. Station observations are then employed to estimate and apply gridded bias corrections on a common spatial framework, thereby reducing systematic and locally varying errors while preserving cross-variable physical relationships. The resulting dataset provides gridded fields of six key variables, including 2 m air temperature, relative humidity, 10 m wind speed, surface pressure, downward shortwave radiation and precipitation, at multiple temporal resolutions. It is designed as a unified, well-documented forcing dataset to support dry–wet indices, extreme-event analyses, moisture-transport diagnostics and a wide range of hydrological, ecological, agricultural and human–environment applications over Xinjiang’s complex terrain.