Dalinor Lake, the second-largest endorheic salt lake in Inner Mongolia, has shown a shrinking trend given the lack of a significant decrease in precipitation (PRE). Based on high-spatial-resolution datasets, we employed a linear regression model, Theil–Sen median trend analysis, the Mann–Kendall test, and a land use transfer matrix to identify the spatio-temporal distribution and trends of PRE and actual evapotranspiration (AET) at the watershed scale during 2001–2019; then, the water deficit (WD) caused by land use changes in different surface lithology zones was analyzed. The results showed that the annual PRE and WD of the Dalinor Lake watershed showed insignificant upward trends, while the annual AET showed a significant upward trend. Spatially, about 89% of the watershed showed a significant upward trend for AET, while 12% showed a weak significant upward trend for PRE. The WDs of the aeolian sand zone and the sand, gravel, and silt accumulation zone were most heavily affected by the new increased land use from 2001 to 2019, accounting for 43.14% and 25.56% of the total WD of the watershed, respectively. Specifically, the WD of the aeolian sand zone caused by the new increased grassland and farmland in 2019 accounted for 41.92% and 18.52% of the total WD of the zone, respectively. The WD of the sand, gravel, and silt accumulation zone caused by the new increased grassland and farmland in 2019 accounted for 37.07% and 35.59% of the total WD of the zone, respectively. The WD caused by the new increased land use was increased by 7.78 million m³ in 2019 compared with the corresponding land use type in 2001, which would decrease the water yield. It is necessary to strengthen the protection of regional forest ecosystems in the granite and terrigenous clastic rock zone; standardize pasture management and reduce farmland reclamation in the sand, gravel, and silt accumulation zone, the aeolian sand zone, and the basalt platform zone; and reduce unnecessary impervious land construction in the aeolian sand zone.

Clarifying the spatiotemporal variations of terrestrial water storage anomaly (TWSA) and the water scarcity is crucial for maintaining water resources, ecosystem and food security. This study adopts the Bayesian triple collocation method (BTCH) method to fuse three GRACE Mascon product datasets to generate the TWSA_BTCH, covering the period from April 2002 to December 2023. The reliability of TWSA_BTCH is further verified using the water balance approach. The spatiotemporal variations of TWSA and the water scarcity index based on terrestrial water storage (WS_{TWS_based}) at the national and primary water resource regional level are then analyzed. The results indicate that: (1) The spatially averaged uncertainty of the TWSA_BTCH over China is 8.13 mm, which is lower than those of TWSA_average (12.51 mm), TWSA_CSR (23.43 mm), TWSA_GSFC (29.79 mm), and TWSA_JPL (31.76 mm). (2) From 2003 to 2023, the TWSA exhibited an overall significant declining trend with an annual depletion rate of 1.65 mm. The national average value of the WS_{TWS_based} index decreased from 0.32 during 2003–2012 to 0.28 over 2013–2023, which indicates that the pressure of nationwide water scarcity has been alleviated to a certain extent, while the whole country still faces a moderate water shortage status. Meanwhile, the mean annual TWSA across China during 2003–2023 was −10.28 mm, indicating a nationwide terrestrial water storage deficit. (3) Regional water resource risks require joint characterization from the dual perspectives of hydrological fluxes and storage. Northern arid and semi-arid regions feature inherent imbalances between water supply and demand. The latent risks of terrestrial water storage depletion within the SWRB cannot be ignored. In addition, insufficient water regulation capacity in the SRB deserves adequate attention.