To obtain high-precision precipitation simulation results, different types of rainfall events in the Ili Region are simulated by using the Weather Research and Forecasting (WRF) model with different physical parameterization schemes. According to the spatiotemporal distribution of rainfall evenness, six rainfall events in the Ili Region are divided into four types. Six microphysical parameterization (MP) schemes, five planetary boundary layer (PBL) schemes, and five cumulus (CU) schemes are combined into 14 parameterization members to simulate the rainfall events. It is worth noting that the simulation result sequence of the WRF model (from best to worst) is as follows: type I (events 3 and 5) > type II (events 1 and 6) > type III (event 2) > type IV (event 4). This finding would imply that the WRF model has the best performance for rainfall events with even spatiotemporal distributions, while it is hard to achieve good simulation results for rainfall events with highly uneven spatial and temporal distributions. The results suggest that no single combination of parameterization members provides the best performance for all rainfall events. According to the overall scheme rankings, d, n, and j are the optimal parameterization combination members that accurately describe the spatiotemporal characteristics of the six rainfall events. The study provides guidance for the selection of the physical parameters for the accurate simulation of different types of rainfall events in the arid region of northwestern China.

Water scarcity and uneven spatiotemporal water distribution trigger severe supply–demand conflicts in arid irrigation districts, making rational water allocation critical to safeguarding food security and improving water use efficiency. This study selects the Jiamu Town Irrigation District in the Tailan River Basin of Xinjiang as the research case. By incorporating hydrological year types, ten-day runoff sequences, exploitable groundwater volumes, and crop irrigation priorities, a multi-objective optimization model was constructed. The two optimization objectives are minimizing total groundwater extraction and the weighted irrigation water deficit. Herein the weighted deficit equals the physical water shortage of each crop-growth-stage irrigation task multiplied by its priority coefficient, which drives the optimization iteration to prioritize water-shortage relief for high-priority irrigation demands. The key findings are summarized as follows: (1) Relative to the actual water consumption in 2023, the optimized scheme raises surface water utilization efficiency from 73.1% to 81.3% and cuts groundwater exploitation by 2.05 million m³ (from 11.09 million m³ to 9.04 milion m³). (2) Under a fixed crop planting structure, irrigation water demands across all ten-day intervals can be largely satisfied. (3) Crop irrigation satisfaction rates exceed 97% across all hydrological scenarios, with water deficits predominantly occurring in drip-irrigated pepper and orchards. The proposed optimal allocation model can effectively curb groundwater overexploitation and minimized water use conflicts among diverse crops and their critical growth stages, offering technical references for refined water resource management in arid canal–well irrigation districts.

To reveal the spatiotemporal evolution characteristics and driving mechanisms of ecosystem patterns in the mine ecological restoration project area on the western margin of the Junggar Basin, spatial data of ecosystem types from 2015 to 2024 were utilized, and the evolution characteristics and driving factors were analyzed by comprehensively applying the PNTIL model, landscape pattern indices, and the Geodetector. The results indicated: (1) Approximately 70% of the study area was consistently occupied by bareland; a net increase of 163.52 km² (+61.63%) was recorded for grassland area, while a net decrease of 118.08 km² (−15.95%) was observed for farmland area. (2) For forest, grassland, and wetland ecosystems, the negative transformation rates were all higher than the positive transformation rates, and the highest negative transformation rate was recorded for grassland, at 8.56% (2020–2024). (3) At the landscape level, increases in patch number and patch density were observed; at the class level, fragmentation was increased for bareland and farmland, while grassland patches were found to be more aggregated. (4) Annual mean evapotranspiration (q = 0.746), per capita GDP (q = 0.677), and annual mean temperature (q = 0.518) were identified as the primary driving factors, and the interaction between annual mean evapotranspiration and annual mean precipitation was found to be the strongest interactive driving factor. The findings of this study can be used as a theoretical basis for similar historically abandoned mine ecological restoration project areas in terms of land use planning, ecological restoration practices, and sustainable development.