Spatiotemporal evolution, determinants, and forecasting of the water use structure in Guangdong Province, China

The examination of urban water use characteristics and the factors that influence these characteristics is crucial for forecasting the regional water demand and achieving effective water management planning. Despite being a major contributor to China's GDP, Guangdong Province faces significant challenges in achieving sustainable development , primarily due to the uneven spatiotemporal distribution of water resources and ongoing water quality concerns. This study addresses research gaps by providing an in-depth analysis of urban water use in Guangdong Province from 2001 to 2020. The results reveal that the total water use of Guangdong Province experienced an initial increase followed by an average annual decline of 1.45 % after 2011, indicating a crucial shift in consumption patterns. Notably, increasing information entropy values (from less than 1 to 1.23) and a rise in the equilibrium degree value (to 0.77) indicate increased complexity in the water use structure. Gini coefficients indicate a relatively balanced distribution for agricultural and domestic water use (approximately 0.2), whereas ecological environmental water use remains less balanced (with Gini coefficients higher than 0.4). The key driving factors influencing regional water dynamics include industrial adjustments and technological advancements. Furthermore, an optimized long short-term memory (LSTM) model is used to predict future trends with high accuracy, achieving a maximum R 2 of 0.99. Projected trends indicate decreases in agricultural and industrial water use, coupled with increases in residential and urban public water consumption, whereas ecological demands are expected to fluctuate and remain complex. Overall, this research provides vital insights that can enhance water resource management and inform sustainable development strategies in Guangdong Province. • Water consumption patterns and features in Guangzhou over two decades are depicted. • The natural and social drivers of the dynamics of different water use types are summarized. • The LSTM approach effectively predicted changes in different water consumption patterns.