随着城市化的快速发展，城市内涝、非点源污染和地表水体水质恶化等水环境问题受到越来越多的关注。作为城市区域中相对独立且具有空间拓扑关系的单元，各种下垫面单元在城市雨水径流和非点源污染物的产生、迁移过程中起着重要作用。弄清城市区域中各种下垫面的雨水径流和非点源污染物贡献是有针对性地实施LID设施的基础。对城市下垫面单元产生的雨水径流和污染物的有效控制是实现区域环境治理目标的前提。低影响开发（Low Impact Development, LID）措施主要针对某块城市下垫面单元起作用。阐明下垫面单元上实施的LID设施的雨水径流和非点源污染物削减机制并量化其削减效果可以为LID设施在更大范围内推广奠定基础。场次降雨事件是城市内涝和非点源污染输出的驱动因素，研究不同等级场次降雨事件中的雨水径流和非点源污染物来源和LID设施效果，可以为城市管理决策提供更加详细的支持。本论文首先结合城市区域水文水质模型和同位素技术量化了各种城市下垫面以及雨水管网沉积物的雨水径流和非点源污染物贡献。在此基础上识别了需要重点控制的下垫面类型。继而利用人工降雨试验，得到了典型LID设施及其组合的水文表现数据，利用这些数据建立了LID设施模型，并探讨了LID设施及其组合的雨水径流削减机制和污染物释放效应。然后针对关键下垫面单元，结合下垫面单元场地模型和LID设施模型，考虑水量和水质削减的关系，研究了下垫面单元上实施的典型LID设施的水量-水质削减机制和效果。在此基础上进一步量化了场地和社区中实施的LID设施组合的效果，探讨了组合中各LID设施的配合效应，确定了成本效益最优的LID设施组合方式。本论文结合了城市区域水文水质模型、同位素技术、人工降雨试验和LID设施场地模型等多种研究方式，建立了场地尺度LID设施及其组合的模拟、评价、优化的理论与方法体系。为LID设施在更大范围内实施的研究提供了坚实的基础。本研究的主要结论包括：

1）市政道路、屋顶和广场是雨水径流的主要来源。不透水率较高的下垫面也是非点源污染物的主要来源。屋顶贡献率较大的污染物是总氮（TN）和NO₃-N，市政道路多种污染物的贡献率都较大，贡献率最大的是有机物和总磷。广场和小区道路贡献率最大的是总悬浮物。在大暴雨中，未利用地成为第二大总磷贡献源。不同等级降雨事件中，各种NO₃-N源的贡献率发生了显著变化。随着降雨事件等级的增加，大气和地表冲刷效应在区域NO₃-N输出中的作用越来越重要。

2）与屋顶面积相等的绿色屋顶和普通绿地可以有效削减大雨以及1年重现期设计暴雨条件下的屋面径流。50%屋顶面积的下沉式绿地和占地2.3%屋顶面积的雨水桶（容积为24 m³）可以有效削减暴雨及10年重现期设计暴雨条件下的屋面径流。雨水桶和下沉式绿地的径流削减能力大于绿色屋顶和普通绿地且不易受前期干旱条件影响。雨水桶的峰值削减效果较差。下沉式绿地各种削减指标均较好。随着事件降雨量的增加绿色屋顶和雨水桶的体积削减量变化不大，绿地的体积削减量显著降低。新建绿色屋顶出流具有较高的COD、TN、NO₃-N、TP、磷酸盐浓度和浊度。生物滞留池出流的各种污染物浓度均显著低于绿色屋顶。

3）20%道路面积的两种生物滞留池对暴雨及5年重现期设计暴雨条件下的道路径流具有显著削减效果，TN总负荷削减率超过57%。生物滞留池本身参数比道路参数对其效果影响更大。底层基质参数的影响整体上大于上层参数。不设置底部出流回收设施的生物滞留池的径流总体积和TN总负荷削减效果显著，蓄水量是影响其径流总体积削减率的最主要因素。TN总负荷的削减主要通过径流总体积削减达到。设置了底部回收设施的生物滞留池的TN总负荷和事件平均浓度削减效果较为显著。TN事件平均浓度削减是径流体积削减和基质吸附共同作用的结果。TN总负荷的削减主要通过浓度削减达到。

4）LID设施组合可以在降低其中各LID设施规模的情况下，保持良好的径流削减效果。组合的削减效果指标值大多介于更大规模的单一LID设施效果值之间，能够发挥各种LID设施的优点，避免其缺点。绿色屋顶、雨水桶和下沉式绿地三种LID设施的组合要达到最高的成本效率，需要增加下沉式绿地的面积，尽量不用绿色屋顶。将绿色屋顶和生物滞留池组合可以降低绿色屋顶出流直接排放的环境影响。

5）在北京师范大学校园内实施针对屋面径流的LID设施可以有效地削减校园排水口出流和管网溢流。单一LID设施中下沉式绿地的调控效果最好，其次是普通绿地。雨水桶的出流体积削减效果最差，但仍具有明显的管网溢流调控效果。LID设施组合具有最佳的调控表现。社区尺度LID设施种类和效果评价指标的选择应针对项目的主要目标来进行。在校园内实施针对屋面径流的设施组合基本可以实现政府提出的年径流削减目标。

With the rapid development of urbanization, urban environmental problems such as urban floods, non-point source pollution, and deterioration of urban surface water quality have attracted more and more attention. The effective control of runoff and pollutants generated by urban underlying surface units is the premise to achieve the goal of regional environmental protection. Recognizing the contribution of various underlying surfaces to runoff and non-point source pollutants in urban areas is the basis for targeted implementation of low impact development (LID) facilities. This paper combines various research methods such as regional hydrological and water quality model, isotope technology, artificial rainfall experiment, and LID model, and establishes the methodology of simulation, evaluation, and optimization of field scale LID facilities and their combination. It provides a solid foundation for the implementation and research of LID facilities in a wider range.

The main conclusions of this study include:

1) Municipal roads, roofs, and squares are the main sources of runoff. The pollutants with a high contribution rate for roofs are total nitrogen (TN) and nitrate-nitrogen (NO₃-N). Municipal roads have high contribution rates for several pollutants, and chemical oxygen demand and total phosphorus have the highest contribution rates. Squares and community roads have the largest contribution rate for total suspended solid. The contribution rate of various NO₃-N sources changes significantly for different grades of rainfall events. With the increase of rainfall event grade, the washoff effect in the atmosphere and on urban surface plays a more and more important role in NO₃-N output.

2) Green roofs and common greenfields with the same area as the roof can effectively control the roof runoff under storm events and design storm rainfall of a return period of 1-year. The sunken greenbelts with 50% of the roof area and the rain barrels with a volume of 24 m³ covering 2.3% of the roof area can effectively reduce the roof runoff under storm events and the design storm rainfall with a return period of 10-year. The runoff reduction capacity of rain barrels and sunken greenbelts is greater than that of green roofs and common greenfields, and it is immune to the influence of antecedent dry conditions. The peak reduction effect of rain barrels is poor. Various reduction indicators of sunken greenbelts are good. With the increase of event rainfall, the volume reduction of green roofs and rain barrels changes little, and the volume reduction of greenfields decreases significantly. The effluent from the new green roofs has high COD, TN, NO₃-N, TP, phosphate concentration, and turbidity. The concentrations of various pollutants in the effluent of the bioretentions are significantly lower than those of the green roofs.

3) Bioretentions with 20% of road area can significantly control the road runoff under storm events and 5-year return period rainfall, and the reduction rate of total TN load is more than 57%. For the bioretention without an underdrain, the total volume of runoff and TN load is reduced most significantly. Water storage volume is the main factor affecting the volume reduction rate. The reduction of total TN load is mainly due to the reduction of total runoff volume. For the bioretention with an underdrain, the reduction effect of total TN load and event mean concentration of TN is more significant. The reduction of TN event mean concentration is mainly the result of volume reduction and medium adsorption. The reduction of total TN load is mainly due to concentration reduction.

4) The combination of LID facilities can maintain a good runoff reduction effect while reducing the size of constituent LID facilities. The values of reduction effect indicators of the combined LID are mostly between that of the effect indicators of its constituent LID facilities. The combination of LID facilities can give full play to the advantages of constituent LID facilities and avoid their disadvantages. To achieve the highest cost efficiency, the area of sunken greenbelts should be increased and green roofs should not be used in the combination of green roofs, rain barrels, and sunken greenbelts. The combination of green roofs and bioretentions can reduce the environmental impact of direct discharge of the effluent of green roofs.

5）The implementation of LID facilities for roof runoff reduction on the campus of Beijing Normal University can effectively reduce the outflow of campus outfall and network overflow. For the single LID facilities, sunken greenbelt has the best regulation effect, followed by common greenfields. The reduction effect of outflow volume of rain barrels is the worst, but it still has an obvious control effect on network overflow. The combination of LID facilities has the best regulatory performance. The selection of LID facility types and effect evaluation indicators in community scale should be based on the main objectives of the project.