流域污染负荷分配是实施流域精细化管理的重要措施。在指标选取过程中，为了定量描述评价指标对流域污染负荷分配的重要性，引入随机森林算法对各指标作重要性评价，依据重要性程度筛选涵盖全部影响因素的评价指标，并进一步确定各指标权重。将评价指标体系与环境基尼系数模型相结合，构建污染负荷分配耦[JP]合模型，并以清潩河流域（许昌段）为例，制定基于控制单元的COD负荷分配方案。结果表明，满足重要性评分要求的指标分别是居民人均收入、单位GDP污水产生量、工业企业水消费量、地表水资源量、控制单元面积，评分分别为13.97、9.65、8.77、5.21、1.09。依据耦合模型制定的2020年清潩河流域（许昌段）各控制单元COD负荷分配方案中，控制单元QYH3的削减量和削减率均最大，分别为617.14 t/a和30.44 %。分配方案基本符合流域实际，结果较为科学合理。

Watershed pollution load distribution is an important measure to implement watershed fine management.In order to quantitatively describe the significance of assessment indicators on the distribution of pollution load in river basin,the random forest algorithm was adopted to evaluate the importance of the selected indicators.The evaluation indices,which can reflect all the influence factors,were selected based on their importance,then,the weight coefficient of each indicator was further determined based on its score.The coupling model of pollution load distribution is constructed by combining the evaluation index system with the environmental gini coefficient model.Additionally, allocation plan of COD load was formulated based on control unit in the target Qingyi river basin (Xuchang Section).We found that the indicators,which fulfill the requirements ofimportance score,include per capita income,sewage production per unit of GDP,water [JP]consumption of industrial and enterprises,the amount of surface water resources and area of control-unit, and the scores are 13.97,9.65,8.77,5.21 and 1.09,respectively.The allocation plan of COD load in the river basin on 2020 was further formulated based on the coupling model,and the pollutant reduction and reduction rate of the control-unit QYH-3 are the largest,which are 617.14 t/a and 30.44 %,respectively.The allocation plan is reasonable for reflecting the actual situation of the basin.

国家水体污染控制与治理科技重大专项项目（2015ZX07204-002-005)