Central Asia (CA) is one of the most arid regions with serious water shortages. To understand the impacts of climate change on the regional water storage in CA, we introduce the terrestrial water storage (TWS) as a comprehensive water resource indicator and evaluate its spatial-temporal variation by regional coupled simulation. In this study, the performance of the Weather Research Forecasting (WRF) model in TWS simulation is evaluated. To consider the uncertainties and study the sensitivity of TWS to model schemes, two microphysics, two planetary boundary layers, and three cumulus schemes are combined in the numerical experiments at 25 km horizontal resolution. The results show that the modeled TWS by WRF agrees well with the satellite-based TWS in high correlation and consistent trends. The simulated TWS has larger variability than that of the satellite-based TWS, but a weaker decreasing trend due to lack of human activities consideration. Overall, the modeled TWS is mainly influenced by the precipitation and soil moisture, and insensitive to the physical parametrizations on a monthly scale in CA. This can serve as a basic tool for water resource assessment in CA by the stable performance of WRF, especially after the consideration of human activities in the future.