Analyzing water footprint and water resources sustainability in China’s arid Northwest with Bayesian network

Abstract

Water scarcity in arid regions severely restricts regional economic development. To promote the sustainable development of both the economy and water resources, this study integrates the water footprint (WF) theory with a Bayesian network (BN) model to probabilistically characterize the uncertainties of influencing factors in water sustainability assessment, thus providing a flexible and effective decision-support tool. The BN model, constructed by calculating sectoral WFs and incorporating water sustainability indicators, was rigorously validated and applied to the arid region of Southern Xinjiang (SX). Results indicate that the WF in SX exhibited a fluctuating upward trend from 2004 to 2020, dominated by blue WF primarily driven by the expansion of agricultural irrigation. The model predicts the current probability of water sustainability is 41.9% in SX. Sensitivity analysis reveals that anthropogenic factors significantly influence the WF. Scenario analysis based on four distinct conditions demonstrates that changes in WF and pollutant emissions substantially affect water sustainability. Under future climate scenarios projected to 2030, despite potential increases in precipitation, rising population and economic growth will continue to intensify water demand pressure. Under policy constraint scenarios, regional water sustainability shows improvement. To achieve sustainable development in SX, it is essential to promote efficient irrigation and optimize crop structure, restrict industrial water consumption while enhancing wastewater reuse, improve residential water pricing and conservation measures, develop green infrastructure, and advance virtual water management to balance regional water resources. The BN model effectively elucidates the interactions among factors within the water resource system, providing a robust scientific basis for formulating sound water management strategies.