南疆丰收灌区水资源多目标优化配置方案优选

针对灌区水资源多目标优化配置模型求解算法的不确定性和最优方案选择问题,以新疆阿克苏地区阿瓦提县丰收灌区为研究对象,以灌区可持续发展为目标,建立基于优选算法的灌区水资源多目标优化配置及方案优选模型。对比改进的NSGA-II算法(Non-dominated Sorting Genetic Algorithm-II,带精英策略的非支配排序遗传算法)、NSGA-III算法(Non-dominated Sorting Genetic Algorithm-III,基于参考点的非支配排序遗传算法)、MOEA/D算法(AMulti-Objective Evolutionary Algorithm based on Decomposition,基于分解的多目标进化算法),最终选取性能最好的CNSGA-III算法作为优化算法对水资源配置模型进行求解;建立水资源配置方案评价体系,利用熵权-TOPSIS(Technique for Order Preference by Similarity to an Ideal Solution,优劣解距离法)综合评价模型对符合决策制定者期望的配置方案进行对比筛选,选出灌区水资源最佳配置方案。结果表明:优化后的方案相较于传统配置方案,在"三条红线"用水限额下,经济净效益增加1.0%,灌区总缺水量减少8.4%,碳吸收总量增加4.5%。建立的灌区水资源多目标优化配置及方案优选模型可为管理者制定灌区水资源配置方案提供参考,也可适用于其他干旱灌区,有一定的理论意义。

Optimized selection of the solution for multi-objective optimal allocation of water resources in Fengshou Irrigation Areas of South Xinjiang

Abstract: Severe global water shortages have posed a great challenge on modern agriculture as ever-increasing demand for water due to the population and economic growth. Therefore, it is necessary to scientifically allocate the limited water resources in main irrigation areas, further to improve the utilization rate of agricultural water for the local economy and ecosystem. Taking Fengshou Irrigation Area in Awati County, Aksu Prefecture, Xinjiang of China as the research area, a multi-objective optimal allocation of water resources was established to obtain an optimal selection model (Multi-Objective optimal allocation model of water resources in irrigation area based on Algorithm Selection and Plan Optimization, MOASPO). The economic, social and ecological development indicators were considered, and the evaluation system of water resource allocation scheme in Fengshou Irrigation Area was established, in order to obtain better net economic benefits in the irrigation area with smaller agricultural water consumption. Two procedures were included in the model: 1) To select the optimal solution for the multi-objective optimal allocation model of water resources in the irrigation area, where the NSGA-II, NSGA-III, and MOEA/D were utilized for each Pareto solution, and then the HV algorithm evaluation index was selected to evaluate each Pareto solution under a unified reference point, and finally to select an optimal solution. 2) To select the best allocation plan of water resources suitable for the specific irrigation area. The candidates were taken from the Pareto solution set, thereby constructing an optimal evaluation system for water resources allocation plans in irrigation areas, combining the local actual conditions from economic, social, and environmental benefits. Specifically, the subsystem of economic benefit included two indicators: "per capita net income from planting industry" and "economic crop occupancy rate", while the subsystem of social benefit included two indicators: "per capita food occupation" and "agricultural water use ratio". Meanwhile, the eco-environmental subsystem included two indicators: "fertilizer use per unit of irrigation area" and "total carbon absorption". An entropy weight-TOPSIS comprehensive model was used to evaluate the options to be selected, and finally to determine the optimal allocation plan of water resources. The results show that the cotton planting area increased by 900 hm2 under the appropriate grain output, whereas, the planting area of maize and wheat reduced by 200 and 700 hm2, respectively, indicating an optimal proportion of crops planting area. After optimization, the net economic benefit was 255.18 million Yuan, while the total shortage of agricultural water was reduced by 13.19 million m3, and the carbon sequestration of crops was 118.74 million kg. The net economic benefit increased by 1.0%, while the total shortage of agricultural water was reduced by 8.4%, and the amount of carbon sequestered by crops increased by 4.5%, compared with the traditional allocation plan. The proposed model and optimized plan can provide a potential reference to formulate an allocation plan of water resources in the arid irrigation areas.