Application of a decision sensitivity measure for the cost-benefit analysis of flood polders along the Bavarian Danube

Abstract

The use of sensitivity measures to quantify the relative importance of model inputs and to assess the effect of uncertainties is an important part of risk analyses in engineering. In risk analysis, models serve to support decisions, hence it seems natural that the sensitivity of model outputs to input uncertainty is measured in the context of these decisions. Decision sensitivity measures have been proposed, however, they have seldomly been used in engineering. In this contribution, we utilize value of information for sensitivity analysis on a specific application.

We apply the decision sensitivity measure to evaluate the influence of uncertainties in model inputs to a flood risk model for the evaluation of flood protection measures. This flood risk model supports the decision on constructing controlled flood detention basins, so-called flood polders, along the Danube. Through the reduction of maximum discharges during extreme flood events and a consequent load decrease on downstream river dikes, flood polders can prevent the occurrence of dike failure. To support decisions, the flood risk model outputs the benefit-cost-ratio of these flood polders.

A cost-benefit analysis of potential flood polder locations requires the estimation of the expected monetarized risk reduction as well as the quantification of the project costs, which are both subject to significant uncertainty, e.g., concerning the occurrence and characteristic of future extreme flood events, the performance of present and planned river dikes, population and asset developments and the discount rate. This highlights the importance of conducting a proper uncertainty and sensitivity analysis accompanied by the necessity of choosing a well-fitting sensitivity metric that is related to the specific decision.