Terrestrial water storage anomalies (TWSA) play an important role in the regional and global hydrological cycle. The Gravity Recovery and Climate Experiment (GRACE) (from April 2002 to June 2017) and GRACE-Follow on (GRACE-FO) (from June 2018 to now) provide monthly global TWSA on a large scale, which contributes to the development of hydrology, but there is an 11-month data gap (i.e., from July 2017 to May 2018) between them due to the launch period. Therefore, based on the Center for Space Research (CSR) spherical harmonics (SH) and Mascons solution from April 2002 to June 2017, this study develops a method based on the Variational Mode Decomposition (VMD) with the Long Short-Term Memory (LSTM) model to fill the data gap of GRACE dataset and then tests over Northwest China (NWC). The result shows that integrated VMD–LSTM model achieve significantly higher prediction accuracy as a whole over NWC (NSE = 0.974, CC = 0.989, and RMSE = 0.808 cm) and exhibit higher reliability compared to the single LSTM models. The proposed method outperforms compared to previous similar studies. The current study offers an accurate and effective means of bridging the gap in GRACE-based TWSA data, especially in poor dataset regions like NWC, which may contribute to further understanding of the hydrological cycle and water resources management.