Framework for Many - Objective Robust Decision Making 1

19 This study introduces a new open source software framework to support bottom-up environmental 20 systems planning under deep uncertainty with a focus on many-objective robust decision making 21 (MORDM), called OpenMORDM. OpenMORDM contains two complementary components: (1) a 22 software application programming interface (API) for connecting planning models to computational 23 exploration tools for many-objective optimization and sensitivity-based discovery of critical deeply 24 uncertain factors; and (2) a web-based visualization toolkit for exploring high-dimensional datasets to 25 better understand system trade-offs, vulnerabilities, and dependencies. We demonstrate the 26 OpenMORDM framework on a challenging environmental management test case termed the “lake 27 problem”. The lake problem has been used extensively in the prior environmental decision science 28 literature and, in this study, captures the challenges posed by conflicting economic and environmental 29 objectives, a water quality “tipping point” beyond which the lake may become irreversibly polluted, and 30 multiple deeply uncertain factors that may undermine the robustness of pollution management policies. 31 The OpenMORDM software framework enables decision makers to identify policy-relevant scenarios, 32 quantify the trade-offs between alternative strategies in different scenarios, flexibly explore alternative 33 definitions of robustness, and identify key system factors that should be monitored as triggers for future 34 actions or additional planning. The web-based OpenMORDM visualization toolkit allows decision 35 makers to easily share and visualize their datasets, with the option for analysts to extend the framework 36 with customized scripts in the R programming language. OpenMORDM provides a platform for 37 constructive decision support, allowing analysts and decision makers to interactively discover promising 38 alternatives and potential vulnerabilities while balancing conflicting objectives. 39