The Colorado River Basin (CRB) is a critical resource for drinking water, irrigation, and hydropower. Managed by the US Bureau of Reclamation, Lakes Powell and Mead are the largest storage reservoirs in the system. Unprecedented drought since 2000 has caused sharp declines in storage in Lakes Powell and Mead, prompting the 2007 Interim Guidelines and additional shortage agreements. These operational policies are set to expire in 2026. To support negotiation of new operating policies for post-2026, Reclamation has employed Many Objective Robust Decision Making (MORDM). MORDM creates policy alternatives using evolutionary algorithm optimization then evaluates policy alternatives in large ensembles of model runs that can explore many plausible scenarios of hydrology and demand. While MORDM provides insights, it creates many possible operational strategies, characterized by an intractable number of decision variable, objective, and robustness values. In this presentation, we present a novel framework called post-MORDM, which helps process the large amount of data generated by MORDM to facilitate compromise among decision makers who have differing preferences. In the framework, the Self-Organizing Map (SOM) synthesizes MORDM data as layers organized in a map-like coordinate system. SOM organizes the policy alternatives in relevant clusters, such that decision makers can discover salient characteristics and assess cause-effect relationships between policies and performance (objective and robustness values). In the presentation, an illustrative example will show how post-MORDM can help hypothetical CRB decision makers with different preferences discover commonalities and come to a compromise policy.
