The University of Massachusetts Amherst
University of Massachusetts Amherst

Search Google Appliance

Links

Nonstationary decision model for flood risk decision scaling

TitleNonstationary decision model for flood risk decision scaling
Publication TypeJournal Article
Year of Publication2016
AuthorsSpence CM, Brown C
JournalWater Resources Research
Volume52
Issue11
Start Page8650
Pagination8650–8667
Date Published11/2016
Abstract

Hydroclimatic stationarity is increasingly questioned as a default assumption in flood risk management (FRM), but successor methods are not yet established. Some potential successors depend on estimates of future flood quantiles, but methods for estimating future design storms are subject to high levels of uncertainty. Here we apply a Nonstationary Decision Model (NDM) to flood risk planning within the decision scaling framework. The NDM combines a nonstationary probability distribution of annual peak flow with optimal selection of flood management alternatives using robustness measures. The NDM incorporates structural and nonstructural FRM interventions and valuation of flows supporting ecosystem services to calculate expected cost of a given FRM strategy. A search for the minimum-cost strategy under incrementally varied representative scenarios extending across the plausible range of flood trend and value of the natural flow regime discovers candidate FRM strategies that are evaluated and compared through a decision scaling analysis (DSA). The DSA selects a management strategy that is optimal or close to optimal across the broadest range of scenarios or across the set of scenarios deemed most likely to occur according to estimates of future flood hazard. We illustrate the decision framework using a stylized example flood management decision based on the Iowa City flood management system, which has experienced recent unprecedented high flow episodes. The DSA indicates a preference for combining infrastructural and nonstructural adaptation measures to manage flood risk and makes clear that options-based approaches cannot be assumed to be ‘‘no’’ or ‘‘low regret.’’ 

DOI10.1002/2016WR018981