Development of a constructed wetland network for mitigating nonpoint source pollution through a GIS-based watershed-scale inexact optimization approach

•A GIS-based watershed-scale inexact optimization approach is developed.•CTI and ECM are used to identify wetland sites and estimate NPS loads, respectively.•FSTP is applied to identify the optimal wetland network for reducing NPS loads.•It can handle uncertainties and offer a bridge of policies and economic implications.•Results identify optimal wetland sites and nutrient permit allocation schemes. In this research, a hybrid Geographic Information System (GIS)-based watershed-scale inexact optimization approach was developed to identify the optimal spatially distributed network of constructed wetlands for NPS pollution control. Such an approach combined the GIS technology and fuzzy-stochastic two-stage programming (FSTP) into a general optimization framework. A GIS-based compound topographic index (CTI) was used to identify potential wetland sites at the watershed-scale, a GIS-based export coefficient model (ECM) was employed to estimate the nutrient load from NPS pollution, and the FSTP was used to map the optimal spatially distributed network of constructed wetlands for reducing the NPS pollution with the minimization of system costs under uncertainty. As an integration of two-stage stochastic programming (TSP) and fuzzy credibility chance-constrained programming (FCCP), FSTP can not only address dual uncertainties expressed as probability density functions and fuzzy membership functions but also provide an effective linkage between the pre-regulated management policies and the associated economic implications. The developed modeling system was applied to the planning of wetland network installation in the Songhuaba watershed, China. The optimization results can help watershed managers to obtain desired plans of wetland siting, and nutrient permit allocation for sub-basins under different confidence levels.