Urban non-point source (NPS) pollution management is receiving increased attention because of the environmental and ecological consequences of urbanization. We developed a Storm Water Management Model (SWMM)-based interval double-sided chance-constrained programming (SWMM–IDCCP) model for the optimal design of low impact development (LID) during urban NPS pollution management at a community scale. Pollution control efficiency and economic costs related to LID practices were considered in the SWMM–IDCCP model, simultaneously addressing uncertainties in pollutant output and decision-making processes. Specifically, by combining SWMM, Monte Carlo simulation, interval linear programming, and double-sided chance-constrained programming, the SWMM–IDCCP model can evaluate export loads of urban NPS pollutants and pollution control efficiencies of LID practices and analyze the corresponding uncertainties. Concurrently, the proposed model can provide reliable and optimal implementation schemes for LID practices under different scenarios, dealing with uncertainties such as discrete intervals and double-sided random parameters. The proposed model was applied to the Dongguan University of Technology in Dongguan City, South China. The random distribution characteristics of pollution load were identified. The ranges of pollution load for total suspended solids, chemical oxygen demand, total nitrogen, and total phosphorus were 76,430–165,698, 44,366–74,155, 2118–3,830, and 97–177 kg in 2021, respectively. Probability distribution characteristics of pollution control efficiencies of three LID practices (i.e., bio-retention cell, green roof, and permeable pavement) for the four urban NPS pollutants were obtained. Using the SWMM-IDCCP model, LID configuration schemes were obtained under diverse pollution control targets and confidence levels, taking into account variations in the decision preference of different administrators. This research indicates that the SWMM-IDCCP model can be used to manage urban NPS pollution at a community scale under uncertainty.
Read full abstract