LAUSR

Abstract In this study, we proposed a multi-scale modelling framework for Low Impact Development (LID) by integrating (i) local-scale LID cost-effectiveness analysis, (ii) landscape-scale Management Category (MC) classification and LID selection, and (iii) macro-scale integrative decision making. The framework was used for sitting LID to reduce Total Phosphorus (TP) loss in the City Kunming regional watershed, which discharges massive urban Non-Point Source (NPS) pollution to Lake Dianchi, one of the three most eutrophic large lakes in China. Local-scale simulation-optimization was performed to optimize the sizes and quantities of the candidate LID practices with the System for Urban Stormwater Treatment and Analysis Integration (SUSTAIN) tool. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) was used to solve the optimization problem. The local-scale LID strategies were ultimately integrated at the macro scale in compliance with the water quality standards. Our analysis suggests that the marginal benefits of LID implementation decline dramatically with increasing TP abatement attributed to the law of Diminishing Marginal Benefits. The areal cost-effectiveness curves of different scenarios resemble each other for the same MC since the Coefficients of Variation (CVs) of unit costs are mostly below 10%. Spatial variability of LID implementation is more prominently affected by disparities of landscapes than drainage area sizes because the coverage ratios among different MCs vary from 2.3% to 24.7% whereas they resemble each other among different area sizes. The multi-scale decision-making framework could dramatically improve computational efficiency by reducing the model runs from 923 to 23 compared to the conventional full-watershed simulation-optimization.