An application of dynamic programming to local adaptation decision-making

Abstract

Adaptation decision-making in mountain regions necessitates dealing with uncertainties which are driven by the complex topography and the potential interconnections of stochastic events. Such events can lead to amplifying consequences for the exposed communities located at different elevations. In this study, we present a stylized application of stochastic dynamic programming for local adaptation decision-making for a small alpine community exposed to debris flows and floods. We assume that local decision-makers and planners aim at maximizing specific objectives by choosing from a feasible set of adaptation measures and under given constraints on these actions. Our results show that stochastic dynamic programming is a promising tool to address the underlying problem faced by local planners when evaluating the feasibility and effectiveness of adaptation measures. Furthermore, stochastic dynamic programming has some advantages compared to deterministic approaches which assume full knowledge of the system of interest in a world dominated by randomness. We provide an estimation of a best option and an appropriate metric to benchmark adaptation effectiveness for long time horizons. We show how multiple constraints, risk preferences, time horizons and decision periods all influence the decision-making and the overall success of adaptation responses over time.