The increasing water stress projected for China could shift the agriculture and manufacturing industry geographically

Water resources are crucial for societal development, but escalating depletion and climate change are exacerbating water stress globally. This is particularly concerning for water-intensive sectors like agriculture and manufacturing, potentially impacting industrial spatial distribution. China, with per capita water resources only one-fourth of the global average and a geographical mismatch between water availability and demand (northern regions having less water but more industry), faces significant challenges. The existing South-to-North Water Diversion Project (SNWDP) partially alleviates the problem, but a more comprehensive understanding of future spatiotemporal water stress variations and their impacts on industries is crucial. Current water stress indexes (WSIs) often simplify calculations and lack the capacity to handle data uncertainties. This study aims to address these limitations by developing the Fuzzy Technique for Order Preference by Similarity to the Ideal Solution water stress prediction (FTOPWSP) index. This index will analyze spatiotemporal variations in water stress and their effects on China's population, investigating the potential impacts on the migration of agriculture, manufacturing, and people. The FTOPWSP index complements existing methods by alleviating calculation ambiguities and uncertainties, offering a more holistic water risk assessment. The study will select ten water demand and supply indexes (WDSIs), compute their weights, combine their distances to fuzzy positive-ideal and negative-ideal solutions to establish the FTOPWSP index, determine a threshold for calculating the water-stressed population (FWSPOP), and examine the implications of water stress variations on industrial and population migration.

Numerous water stress indexes (WSIs) have been developed, typically defined as the ratio of water withdrawal to availability. However, the definition of water availability often limits itself to river discharge, neglecting factors like environmental flow requirements, upstream consumption, and blue/green water components. Studies have incorporated these factors for more accurate estimations. Liu et al. considered local runoff, environmental flow requirements, and upstream withdrawals to forecast water stress changes in China. Munia et al. analyzed the impact of local runoff, upstream inflows, and consumption globally. He et al. estimated global urban WSIs, showing a marked increase. Liu et al. incorporated blue and green water components to assess agricultural water scarcity worldwide. Despite these advancements, WSI calculations can be simplistic and computationally deficient, especially when data is uncertain. There's a need for methods handling data uncertainties effectively, and further research on the impact of water stress on industrial migration in China is needed.

This study developed the FTOPWSP index to project spatiotemporal water stress variations in China. Ten indexes (four from the demand side: domestic, irrigation, manufacturing water withdrawal, and evapotranspiration; six from the supply side: groundwater runoff, groundwater recharge, surface runoff, subsurface runoff, total soil moisture, and snow water equivalent) were selected based on data availability and representativeness. Principal component analysis (PCA) and the entropy weight method (EWM) were used to determine the combined weight of each index. The fuzzy TOPSIS method was employed to create the FTOPWSP index, which incorporates data from three water resource prediction models (H08, CWatM, and PCR-GLOBWB) driven by three general circulation models (GCMs: GFDL-ESM2M, HadGEM2-ES, and MIROC5) to enhance robustness. A fuzzy decision matrix was constructed using the minimum and maximum values obtained from the three models for each index in each 0.5° × 0.5° grid cell. This matrix was normalized, weighted using the calculated weights from PCA and EWM, and then used in the fuzzy TOPSIS calculation to determine the distances to the fuzzy positive-ideal solution (FPIS) and fuzzy negative-ideal solution (FNIS). The FTOPWSP index was calculated using these distances. The water-stressed population (FWSPOP) was determined using the 60th percentile of the FTOPWSP index as a threshold. The analysis used monthly data from 2020 to 2099 with a spatial resolution of 0.5° × 0.5° grid cells. SSP2 population data from SEDAC was used for FWSPOP calculations. The IPSL-CM5A-LR GCM model was excluded due to a persistent cold temperature bias.

The FTOPWSP index projects increasing water stress in China from 2020 to 2099 under both RCP2.6 and RCP6.0 scenarios. The increase is primarily driven by decreasing water supplies (total soil moisture, surface runoff, and snow water equivalent show statistically significant negative trends). Spatial analysis reveals higher water stress in northern and western regions, particularly the North China Plain (NCP), compared to the southeast. The NCP, a major agricultural region, faces severe water stress. Hot spot analysis using Getis-Ord Gi* confirms these findings, showing high-stress clusters in the NCP and low-stress clusters in the south. Provincial analysis shows the highest water stress in northern provinces (Tianjin, Beijing, Shandong, etc.) and the lowest in southeastern provinces. Seasonal analysis indicates that spring and autumn changes in water stress contribute most to annual variations (RCP2.6), while winter contributes most under RCP6.0. Irrigation water withdrawal (IWW) significantly influences the difference in water stress between RCP2.6 and RCP6.0 scenarios. The FWSPOP analysis, using the 60th percentile of FTOPWSP as the threshold, projects that over 20% of the Chinese population will experience severe water stress annually under both RCP scenarios. The water-stressed population is mainly concentrated in northern China. The consistency between FTOPWSP and the traditional WSI approach further validates the findings. The study also confirms the potential of glacier melt from Asia's water tower to alleviate water stress, though this effect is unsustainable due to global warming.

The findings address the research question by providing a comprehensive spatiotemporal analysis of water stress in China under future climate change scenarios. The increasing water stress, particularly its uneven distribution, strongly suggests potential migration of agriculture, manufacturing, and population from north to south. The high water stress in the NCP, a vital agricultural region, poses a significant threat to food security and necessitates improved water use efficiency, potential relocation of agricultural production to less water-stressed regions, and implementation of effective water resource allocation policies like the SNWDP. The migration of manufacturing industries, particularly the water-intensive semiconductor sector, from water-stressed regions to the south is also a probable outcome. Population migration from water-stressed northern areas to the south is likely, influenced by both economic factors and climate change. While this migration can alleviate stress in northern regions and improve the welfare of migrants, it also necessitates careful consideration of potential negative environmental impacts. The FTOPWSP index provides a novel approach to assessing water stress, offering flexibility and adaptability for various management aims and policy contexts.

This study introduces the FTOPWSP index, offering a more robust and comprehensive method for assessing water stress in China. The findings highlight the increasing and unevenly distributed water stress, pointing towards significant impacts on agriculture, manufacturing, and population distribution. Future research should focus on detailed analysis of the socioeconomic implications of industrial and population migration, refined modeling of water resource management strategies, and exploration of sustainable adaptation measures to mitigate water stress.

The study relies on projected data from multiple models, introducing uncertainties inherent in climate and hydrological modeling. The choice of the 60th percentile as the threshold for defining the water-stressed population is somewhat subjective, although sensitivity analysis with varying thresholds is presented in the supplementary materials. The study focuses primarily on water stress, not incorporating other factors that might influence industrial and population migration decisions. Detailed data on water supply from different sources (aqueducts, reservoirs, desalination) were not incorporated due to inconsistent classification across different models, potentially leading to overestimation/underestimation of water stress in certain regions.