Decision Support and Stakeholder Engagement to Better Site Green Stormwater Infrastructure in Uncertain Urban Environments
Abstract
Green infrastructure (GI) has become a common solution to mitigate stormwater-related problems in urban areas. Despite wide acknowledgement of GI benefits, most decision support tools don't allow practitioners and stakeholders to interactively evaluate the performance of small GI practices using hydrologic models under uncertainty. Moreover, identifying preferable locations in a watershed, given the uncertainty in modelling parameters, is another challenge for GI planning and design. To address these needs, an online Cloud-based interactive tool — called Interactive DEsign and Assessment System for Green Infrastructure (IDEAS_GI)— has been developed. This study demonstrates the application of the tool, using hydrological (SWMM) and empirical models, to estimate life cycle cost, stormwater volume reduction and treatment, and changes in air pollutant deposition. The models are then merged with a multi-objective probabilistic noisy genetic algorithm (MPNGA) to identify preferable locations for GI placement. The MPNGA uses a probabilistic selection method that, over the course of the evolutionary process, requires numerous sampling realizations to consider uncertainties in the fitness (objective function) values, which are cumulative stormwater flow and GI life cycle cost. To overcome the computational challenge and to identify significant features for optimal locations, the MPNGA is merged with Artificial Neural Networks (ANNs), which act as surrogates for the numerical models. Once Pareto optimal solutions are identified, decision trees are created to determine the effects of subwatershed-related parameters on GI coverage classes for four budgetary scenarios. These methods are then applied in a small watershed in the Baltimore metropolitan area. The optimization results show that the addition of ANNs decreases average computational time to reach the optimality condition by 50%± 20%, while it doesn't change the objective function values with statistical significance. Classifying the preferable GI coverages across the subwatersheds via decision trees shows that, for the budgetary scenarios with the highest and lowest budgets, the highest, or even the entire, investments should be allocated to the subwatersheds closest to the watershed outlet.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H21O1983H
- Keywords:
-
- 1880 Water management;
- HYDROLOGY;
- 6344 System operation and management;
- POLICY SCIENCES;
- 6309 Decision making under uncertainty;
- POLICY SCIENCES & PUBLIC ISSUES;
- 6620 Science policy;
- PUBLIC ISSUES