The unique anticline geological structure in the central region of Yingjisha County results in significant spatial variations in groundwater quality. The study shows that the recoverable groundwater reserves account for 13.5% of the natural groundwater supply, and the development potential is considerable. Therefore, this study conducts an in-depth analysis of the spatial distribution characteristics of multiple water sources, integrates agricultural cropping patterns, and delineates irrigation districts accordingly. A water quality-based optimized allocation model for water resources is established. After optimization, the total irrigation water demand is reduced from 3685.8 million m³ to 3229.9 million m³, with total groundwater extraction controlled at 694.0 million m³. The total water shortage rate is 12%, and the decline in groundwater levels has been effectively controlled. Additionally, 116.4 million m³ of saline water is utilized, achieving an 83% utilization rate, which accounts for 16.8% of total groundwater extraction. Consequently, the utilization rate of freshwater decreases from 127% to 64%, while the overall water supply reliability reaches 87.6%. The sequence of water supply and consumption in the model remains consistent with the existing supply structure, demonstrating the rationality of the model parameter settings. This study proposes an optimal freshwater–saline water allocation model, which mixes saline water with reservoir water for dilution and subsequent agricultural irrigation. The approach aims to exploit the potential of saline groundwater and enhance the utilization efficiency of groundwater systems, thereby providing an innovative solution to alleviate water supply-demand conflicts in arid regions.

To characterize spatial differences in mine-water inflow conditions across the Baiyanghe coal mining area in Xinjiang, eight indicators were selected to represent water-bearing media, external recharge, the potential for water transmission, and roof-rock response to water. Gini and independence weights, derived from variation among evaluation units and multiple correlations among indicators, respectively, were combined using equal coefficients (0.5:0.5) to form Gini–Independence Weighting (GIW) for evaluation and zoning using a geographic information system (GIS). The individual methods assigned markedly different weights to some indicators. The medium-high and high zones delineated by GIW occupied 40.81% of the area, forming discontinuous patches mainly in its western and central parts. Comparison with 43 sites with historical records of abnormal inflow showed that the medium-high and high zones delineated by Gini, independence, and GIW covered 33, 31, and 36 sites, respectively. GIW achieved 83.72% coverage and included every site covered by either individual scheme. Covering 80% of the recorded sites required 37.04% of the area under GIW, less than under either individual scheme. The areas under the coverage curves (AUCs) were similar for Gini, independence, and GIW (0.7851, 0.7659, and 0.7816, respectively). GIW integrated complementary spatial patterns from the two weighting criteria while retaining their main priority areas. The resulting zones represent relative differences in inflow conditions across the mining area; individual abnormal inflows also depend on local water sources, flow pathways, and mining conditions.