Decision Making Methodology to Mitigate Damage From Glacial Lake Outburst Floods From Imja Lake in Nepal

Abstract:

Climate change has accelerated glacial retreat in high altitude glaciated regions of Nepal leading to the growth and formation of glacier lakes. Glacial lake outburst floods (GLOF) are sudden events triggered by an earthquake, moraine failure or other shock that causes a sudden outflow of water. These floods are catastrophic because of their sudden onset, the difficulty predicting them, and enormous quantity of water and debris rapidly flooding downstream areas. Imja Lake in the Himalaya of Nepal has experienced accelerated growth since it first appeared in the 1960s.

Communities threatened by a flood from Imja Lake have advocated for projects to adapt to the increasing threat of a GLOF. Nonetheless, discussions surrounding projects for Imja have not included a rigorous analysis of the potential consequences of a flood, probability of an event, or costs of mitigation projects in part because this information is unknown or uncertain. This work presents a demonstration of a decision making methodology developed to rationally analyze the risks posed by Imja Lake and the various adaptation projects proposed using available information. In this work the authors use decision analysis, data envelopement analysis (DEA), and sensitivity analysis to assess proposed adaptation measures that would mitigate damage in downstream communities from a GLOF. We use an existing hydrodynamic model of the at-risk area to determine how adaptation projects will affect downstream flooding and estimate fatalities using an empirical method developed for dam failures. The DEA methodology allows us to estimate the value of a statistical life implied by each project given the cost of the project and number of lives saved to determine which project is the most efficient. In contrast the decision analysis methodology requires fatalities to be assigned a cost but allows the inclusion of uncertainty in the decision making process. We compare the output of these two methodologies and determine the sensitivity of the conclusions to changes in uncertain input parameters including project cost, value of a statistical life, and time to a GLOF event.