Comparing many-objective robust decision making and multi-objective robust optimization for the lake problem

Abstract

Methods for decision making under deep uncertainty (DMDU) have attracted increasing interest in the context of problems such as climate adaptation, and more generally for the management of environmental systems. In contrast to “predict-then-act” management, DMDU methods aim to identify policies which are robust to uncertain future conditions. In addition, the management of complex systems typically involves meeting multiple performance objectives. DMDU methods which have been used for this purpose include many-objective robust decision making (MORDM) (Kasprzyk, Nataraj, Reed, & Lempert, 2013), and many-objective robust optimization (RO) (Watkins & McKinney, 1997; Kwakkel et al., 2015). MORDM is used to generate multiple policy alternatives and assesses their robustness a posteriori across multiple states of the world, while many-objective RO directly optimises the robustness of alternatives.