Spatial and temporal variation of water stress in China and its driving factors: A multi-scale analysis

Water resources are fundamental for sustaining natural ecosystems and human activities, playing a critical role in the sustainable development of the regional environment. Under the dual pressures of human activities and climate change, however, the stress on water resources has become increasingly evident, emerging as one of the greatest global risks for the next decade. In this study, by applying the water stress index, Lorenz curve, and Theil index, we explored the spatiotemporal patterns and inequality distribution characteristics of water resource stress across two scales: catchment and basin. Additionally, we used partial least squares regression to identify the key factors influencing water resource stress. The results indicated significant regional variations in water stress across China during 2002 to 2020. At the catchment scale, areas with a water stress index greater than 0.4 were distributed in the eastern, northeastern and northwestern regions. While at the basin scale, a north–south pattern emerged with lower stress in the south and higher stress in the north. The Haihe and Huaihe river basins exhibited the highest water stress. The Lorenz curve deviated significantly from the line of absolute equality, indicating a high degree of inequality in regional water resource stress. The Theil index increased from 1.26 to 1.50, showing a slight upward trend in inequality. Analysis of the driving factors revealed that the Yellow River Basin was primarily influenced by GDP and population, the Songhua River Basin was affected by population and urban land use, and the Southwest River Basin is driven mainly by vegetation cover. Overall, precipitation was the most critical driver affecting water stress, predominantly exerting a negative influence. This study provides a theoretical basis for alleviating regional water stress and offers valuable insights for the scientific planning and management of water resources.