Decision Analytic Model for Nonstationary Flood Risk Management

Abstract

Lack of stationarity in long-term climate statistics is an emerging concern in flood-control system design. Potential trends in future flood peaks could invalidate historical records as planning guidelines, but future flood trends are uncertain. While infrastructure-based flood-management measures such as dams and levees are traditionally employed, options-based measures, in which a small investment is made at the present to secure the option of using and paying for some good or service in the future, could be preferable to expensive flood control infrastructure if an increase in flood peaks is not certain. Furthermore, options-type management strategies are often less ecologically disruptive than infrastructure-based flood control measures. We present a decision analytic model that identifies the combination of infrastructure-based and options-based flood-management strategies that minimizes damage and management costs in a floodplain, incorporating nonstationary flood hazard. The model is used to evaluate whether options-based or infrastructure-based flood management options perform best over a variety of trends in flood risk. The decision analytic model is applied to a case study of the Iowa River at Iowa City, Iowa. Existing flood control structures in the area include levees and Coralville Lake, a reservoir operated by the United States Army Corps of Engineers (USACE). The decision analytic model is used to choose between combinations of further development of flood control infrastructure and risk sharing with farmers in exchange for flooding fields when needed. The sensitivity of the decision to the strength of trend in flood peaks can be used alongside future flood risk information to aid decision making.