基于Logistic回归和DEA的水资源供需月风险评价模型及其应用

Monthly Risk Assesment Model of Water Supply and Demand Based on Logistic Regression DEA and Its Application

我国城市水资源供需研究多集中在供需平衡分析方面,而忽略了供需风险的研究。针对这一问题,论文从危险性、脆弱性和损失的角度构建水资源供需风险指标体系,建立了危险性和损失的函数表达式,并且考虑了随机性对损失的影响。利用Logistic回归模型模拟供水量和需水量系列的概率分布,建立了基于数据包络分析的水资源供需风险评价模型,对2020年北京市水资源供需月风险进行定量评价。结果表明:在1956—2012年的来水条件下,北京市月风险具有明显的季节变化特征,其中春季和夏季的风险系数值比较小,在0.4左右波动,秋季的风险系数值较大,基本在0.4到0.6之间波动,冬季的风险系数值最大,在0.6到1之间波动,如果不采取一些措施进行风险调控,1—10月的水资源供需风险会不断增大;利用再生水和南水北调水后,1—12月的风险均有了大幅度的降低,但是1—10月和12月的风险仍然有增大的趋势。

In the studies of water resources supply and demand, supply and demand balance has received much attention, but the risk of supply and demand is little addressed. In this paper,the index system of risk of water supply and demand is established from the viewpoint of threat, vulnerability and loss. Functions of threat and consequent lost are constructed which take into consideration of random effects. The probability distributions of water supply and water demand are simulated by Logistic regression. Based on DEA, a evaluation model for water supply and demand risk is developed and the risk of 2020 in Beijing is evaluated. The results show that, if the inflow condition is as that in 1956-2012, the monthly risk has evident seasonal variation features. The risk coefficients of spring and summer are relative low, which are around 0.4. Those of autumn are relative high, basically ranging from 0.4 to 0.6. Those of winter are the highest, which range from 0.6 to 1. If no measures are taken for risk control, the risk from January to October will increase continuously. After using reclaimed water and Southto-North Transferred water, risks from January to December are reduced greatly, however, the risk of each month still has the tendency of increase except for November.