[Objective] In recent years, excessive exploitation of groundwater resources has led to severe depletion of water resources in the Dusitu River Basin, limiting the healthy development of the local ecological environment and economy. Therefore, accurately acquiring the long-term spatiotemporal change characteristics of water storage is crucial for the sustainable utilization of water resources. [Methods] This study employed the Bayesian three-cornered hat method to integrate three GRACE Mascon products, combined with the soil water and snow water components simulated by the GLDAS model, to generate high-precision groundwater storage change results for the Dusitu River Basin. The results were validated for accuracy using measured groundwater level data from 2018 to 2021 and the water body area of Bulong Lake extracted from satellite remote sensing. The cross wavelet transform method was further introduced to analyze the synergistic effects of precipitation, temperature, and evapotranspiration factors on groundwater storage changes in the time-frequency domain. [Results] From 2003 to 2021, terrestrial water storage and groundwater storage in the Dusitu River Basin decreased significantly at rates of -6.71 mm/a and -7.88 mm/a, respectively. Spatially, the groundwater depletion trend intensified from west to east, with the declining rate increasing from -5.71 mm/a to -9.31 mm/a. After 2018, groundwater depletion accelerated, with the decline rate increasing from -7.37 mm/a to -9.52 mm/a. The trends and seasonal characteristics of measured groundwater levels were consistent with GRACE results, with an average correlation coefficient of 0.56. The area of Bulong Lake continuously decreased at a rate of approximately -2 698 m²/a, showing significant seasonal fluctuations, which was largely consistent with the trends of groundwater storage changes in the river basin. Cross wavelet analysis showed that precipitation and groundwater storage were significantly positively correlated at the 1-month scale, while temperature and evapotranspiration were significantly negatively correlated. [Conclusion] This study significantly improves the inversion accuracy of water storage changes through multi-source GRACE data fusion, clarifies the severe reality of continuous and intensifying groundwater over-exploitation in the Dusitu River Basin, and highlights regional water resources and ecological pressures. Furthermore, precipitation is the main source of groundwater recharge, while temperature and evapotranspiration exacerbate its consumption. The research findings provide reliable technical methods and data support for water resource management in the river basin.