城市降雨径流污染是城市污染物排放的重要来源之一。随着经济社会的发展，国内外对这类污染源的重视程度越来越高。同时不透水地表面积增加导致的城市洪涝灾害频发，传统的城市降雨径流“快排“模式难以有效解决这些问题，而城市降雨径流污染控制最佳管理实践（BMPs）提供了很好的解决思路。因此，在实践中如何设计并选择费用-效益最优的污染控制方案，就显得尤为重要。本研究以昆明市官渡区部分城区为研究区域，通过对Google Earth的0.3 m高分辨率遥感影像进行目视解译的方式，获得了区域详细的下垫面类型分布数据，并证明了该方法是一种准确性较高的可靠的数据获取方法。同时多途径提取获得其他区域基础特征数据。研究中选择TN和TP两类污染物，以2012年降雨和设计的单场3 h降雨量为25 mm的降雨两种情形下，利用城市雨洪管理模型（SWMM）对环内已建成地区域的降雨径流污染负荷进行核算并分析识别其污染特征，在此基础上，通过类比法和面积法估算得到环内未建成利用地区域未来情景下的降雨径流污染负荷。结果表明，在设计的单场降雨情况下，屋面、人行道、道路和绿地4类下垫面单位面积的降雨径流污染负荷产生量TN分别为89.80、101.59、111.49、9.99 mg/m2，TP分别为7.57、19.52、23.42、1.35 mg/m2，环内区域污染负荷总产生量为：径流总量36.46万m3、TN负荷1462.36 kg、TP负荷228.97 kg；城市不渗透地表占比与区域降雨径流污染负荷产生量存在显著的正相关关系。同时证明了SWMM模型是一个较为实用的、可靠的可以在不同时间和空间尺度上完成城市降雨径流污染负荷计算的工具。下一步，在对部分单体式BMPs措施的适用性和效益等特征进行分析对比，并参考相关规划等资料的基础上，初步设计住宅小区（约11.9 ha）尺度上的降雨径流污染控制BMPs方案。之后，利用城市降雨处理和分析整合系统（SUSTAIN），以初设方案中不同BMPs措施的尺寸或数量为决策变量和约束条件，以TN或TP的年降雨径流污染负荷控制率80%-90%为目标，优选小区整体上费用-效益最优的污染控制BMPs方案。通过模拟分析，在初选方案的前提下，系统共优选产生56组最优解；以控制率约为85%为例，通过在住宅小区内设置38个雨水桶、7个透水铺装单元、304个生物滞留单元和400 m2的湿塘，共花费211.6万元，即可实现对TN控制率为84.97%的目标；设置19个雨水桶、1个透水铺装单元、153个生物滞留单元和200 m2的湿塘，共花费103.5万元，即可实现对TP控制率为84.97%的目标。控制方案中费用的主要去向是下沉式绿地，分别占到总费用的86.2%和88.7%。另外，对于单体式BMPs措施，相比于其他措施，绿色屋顶在对降雨径流污染负荷控制方面不占优势，大面积推广下沉式绿地是控制降雨径流污染优先考虑的方案。本研究的相关成果可以为城市降雨径流污染负荷核算及其控制的BMPs方案设计与优化提供方法依据，是节约型社会和海绵城市建设形势下的有益尝试。

Stormwater runoff pollution is a major source in urban watersheds. With the economic and social development, people at home and abroad put more and more emphasis on it. With frequent flood disasters resulting from increased impervious land areas, the traditional Rapid Discharge Model of urban runoff emmission is difficult to solve these problems effectively. Best Management Practices (BMPs) in urban watersheds for stormwater pollutants' control provides a good resolution. In a real project, it is particularly important to design and select the optimal cost-effectiveness of pollution control schemes.The study areas are in Guandu District, Kunming City. Detailed distribution data of the types of underlying surface is obtained through Visual Interpretation Method of Remote Sensing Images with 0.3m resolution from Google Earth. It is demonstrated that the method is a accuracy and reliable data acquisition method. Other basic data of the areas are collected by a variety of ways. The targeted pollutants are TN and TP. In the scenarios of 2012 all year rainfall events and single designed rainfall event with total volume 25mm lasting 3h, the pollutants' loads in constructed sections inner the ring road are calculated with using the Storme Water Management Model (SWMM). The characteristics of the pollutions are identified with the results. On the basis, the pollutants' loads in under construction sections in future are calculated by the Analogy Method and the Area Method. The results show that the 4 types of underlying suface rooftop, sidewalk, carroad and grassland can emmiss different pollutants' loads on per unit of area in the designed rainfall event, respectively, TN 89.80, 101.59, 111.49, 9.99mg/m2, TP7.57, 19.52, 23.42, 1.35mg/m2, and total loads inner the ring road are runoff volume 0.36 million m3, TN 1.46t, TP 0.23t. It also shows that there is a significant positive correlation between the percent of impervious land areas and the polutants' loads. Meanwhile, it is demonstrated that the SWMM model is practical and reliable tool to complete the calculation of urban stormwater runoff pollutants' loads in different temporal and spatial scales.Next step, the analysis contrasts some types BMPs technologies' features including suitability, cost and effectiveness. With the local plans and the features, the preliminar BMPs scheme is designed for a residence community (almost 11.9ha). And then, the tool named System for Urban Stormwater Treatment and Analysis INtegration (SUSTAIN) is used for scheme optimization. It takes BMPs' dimensions or number of units as the decision variables and constraints, and takes the annual runoff loads' control rate 80%-90% as the target, and select the the optimal cost-effectiveness resolutions. As a result, 56 groups of reslutions work out. Take the appropriate rate 85% for example, the placement with 38 rainbarries, 7 units of porous pavement, 304 cells of bioretention and 400m2 wet ponds, and the total cost 2.116million yuan can achieve the TN control rate 84.97%, and to TP, the values are 19, 1, 153, 200, 1.035million and 84.97%. The costs are largely spent on units of bioretention, while the rates reach 86.2% and 88.7% for TN and TP respectively. In addition, green roof is not dominant compared with other types of BMPs for loads' control, and the spread of sunken green space in large areas should be the preferred solution to control the runoff loads.The study supplies methods' guide and support for urban stormwater runoff pollutants' loads estimation and design and optimization of placement of BMPs. These are beneficial attempts at the construction of the Conservation-Oriented Society and the Sponge City.