Study region: The Northwest inland basins of China (NWC).Study focus: Terrestrial water resources, especially groundwater resources, are the main source of water for human activities and for maintaining the stability of the ecological environment in NWC. Excessive consumption of water resources will seriously affect the sustainable utilization of water resources and ecological security in this region. Therefore, it is urgent to clarify the long-term changes in water storage in this area in order to handle the pressure of future water re-sources and the natural environment. Using GRACE satellite datasets and global hydrological models (GHMs) products, this study analyzed spatiotemporal variations in terrestrial water storage anomalies (TWSA), groundwater storage anomalies (GWSA), soil moisture, snow water equivalent, and canopy interception combined anomalies (SSCA) in NWC through the application of the water balance, trend decomposition, and empirical orthogonal decomposition methods. Furthermore, the driving factors of water storage change and feasible water resource manage-ment strategies were discussed. New hydrological insights for the region: TWSA in the NWC has experienced a continuous decline over the past nearly 40 years, while SSCA has shown a weak increasing trend (0.03 cm yr-1). Since the availability of glacial retreat data (2003-2016), glacial water storage in the NWC has decreased by 0.09 cm per year, while TWSA, SSCA, and GWSA have changed at rates of -0.25, 0.02, and -0.18 cm yr-1, respectively. The North Tianshan Rivers Basin has become one of the areas with the most severe groundwater depletion in China. 2005-2010 was a turning period in the changes of TWSA, followed by widespread water loss across the NWC. Glacier and snow melt are the most important factors for the decline of TWSA in the Tianshan mountains area, and over -exploitation of groundwater by human activities is a secondary factor. For other regions, Groundwater losses remain the most significant contributor to TWSA losses. The massive loss of water storage in the Tianshan Mountains area, especially the accelerated retreat of glaciers, will affect the stable water supply to the middle and lower reaches of the oasis region, perhaps leading to increased groundwater extraction, which will threaten regional water security and sustainable development. Developing a water-saving society and implementing inter-basin water transfer arefeasible ways to alleviate the water resource crisis. Conducting a comprehensive analysis of all inland rivers in China helps to facilitate horizontal comparisons between various basins, thereby providing more comprehensive insights of water storage fluctuations. The data on water storage changes, extending back to 1980, provide a longer-term perspective on water resource changes in the region, which can contribute to enhancing water resource security and ecological environ-mental protection.

Arid regions of Central Asia have sensitive ecosystems that rely heavily on terrestrial water storage which is composed of surface water storage, soil moisture storage and groundwater storage. Therefore, we employed three Gravity Recovery and Climate Experiment (GRACE) satellite datasets and five global hydrological models (GHMs) to explore the terrestrial water storage (TWS) changes over arid regions of Central Asia from 2003 to 2014. We observed significantly decreasing water storage trends in the GRACE data, which were underestimated by the GHMs. After averaging the three GRACE satellite datasets, we found that the water storage was decreasing at a rate of -4.74 mm/year. Contrary to the prevailing declining water storage trends, northeastern Kazakhstan (KAZ), and southern Xinjiang increased their water storage over the same period. The GRACE data showed that Turkmenistan (TKM), Uzbekistan (UZB) and KAZ experienced the most severe water depletions, while Tajikistan (TJK) and northwest China (NW) experienced the least significant depletions. With respect to the major river and lake basins, the Aral Sea Basin exhibited the most serious water loss (-0.60 mm/month to -0.38 mm/month). The water storage positively correlates with the precipitation; and negatively correlates, with a three-month lag, with temperature and potential evapotranspiration (PET). Partial least square regression (PLSR) had the high capability in simulating and predicting the TWS. These results provide scientific evidence and guidance for local policy makers working toward sustainable water resource management, and the resolution of international water resource disputes among Central Asian countries.