Project

Discipline

environmental decisiion support; expected utility; imprecise probabilities; uncertainty; river rehabilitation prioritization; expected expected utility; density ratio class; multi-criteria decision analysis; ambiguity

Lead
Entscheide von Behörden oder politischen Gremien im Umweltmanagement sind schwierig, da unterschiedliche Interessengruppen unterschiedliche Ziele verfolgen und die Konsequenzen der Entscheidungsalternativen oft unsicher sind. Aus diesem Grund ist es besonders wichtig, dass solche Entscheide auf dem besten aktuellen wissenschaftlichen Wissen aufbauen und die Unsicherheiten und Gründe für einen Entscheid transparent kommuniziert werden.
Lay summary
Das Forschungsprojekt hat zwei Ziele. Erstens sollen Methoden erarbeitet werden, um die Unsicherheit in den gesellschaftlichen Präferenzen und in der Prognose der Konsequenzen von Entscheidungsalternativen besser berücksichtigen zu können als das in der Vergangenheit gemacht wurde. Zweitens sollen mit einer Fallstudie zur Priorisierung von Fliessgewässerrevitalisierungsmassnahmen die Anwendbarkeit und der Nutzen dieser Methoden getestet werden. Auf der methodischen Seite besteht der Innovationsbedarf primär in der Berücksichtigung der Unsicherheit der gesellschaftlichen Präferenzen, die im Gegensatz zur Unsicherheit der Prognosen von möglichen Handlungsalternativen bisher kaum berücksichtigt wurden. Was die Prognosen betrifft, besteht Bedarf an der Berücksichtigung der Unsicherheit, die durch unterschiedliche Einschätzungen von verschiedenen Fachpersonen entsteht. Die Revitalisierungsplanung ist ein Anwendungsfall, der diesen Unsicherheiten unterworfen ist und von aktueller Bedeutung ist, da in den nächsten 80 Jahren mit grossem finanziellem Aufwand viele Fliessgewässer rehabilitiert werden sollen. Es ist deshalb von grosser Bedeutung, mit diesem Aufwand einen möglichst grossen Gewinn für die Ökosysteme und die Bevölkerung zu erreichen.

Direct link to Lay Summary

Last update: 05.04.2020

Number	Title	Start	Funding scheme

Environmental Management can profit from formal Decision Analysis (DA) methodologies as they can help structuring the argumentation and negotiation processes, they can support combining scientific predictions with societal preferences, they increase the transparency of argumentation and - after the decision has been taken - the justification of the selected alternative. State of the art of environmental decision support based on DA considers the uncertainty in scientific predictions of the consequences of decision alternatives and a formal, quantitative description of the preferences of decision makers or stakeholders, ideally including their risk attitudes. In Multi-Criteria Decision Analysis (MCDA) based on Expected Utility theory (EU), ranking of alternatives is done by decreasing values of the expected utility across the predicted outcome distributions of the decision alternatives.This state of the art methodology ignores the uncertainty in the preferences and their quantification and the ambiguity in the probability distributions that quantify knowledge and uncertainty. This is a considerable flaw as individual preferences are uncertain, the aggregation to a “societal preference” is problematic, and the parameterization and elicitation processes add additional uncertainties. The overall goals of this project are to address these flaws by1.extending this state of the art in environmental decision support by accounting for the uncertainty in preferences and the ambiguity in probability distributions, and2.investigating the feasibility of the proposed approach and the relevance of the increase in uncertainty for the decision making process in a case study.Suggested methods to address these topics are to do Bayesian inference for a parameterized value or utility function using elicited preference information from stakeholders and to consider the ambiguity about prior information as imprecise probabilities in the form of sets of probability distributions using the so-called “density-ratio class”. The innovative part in the first methodology is to formulate a likelihood function for the elicitation process and to deal with elicitation results as with data in the environmental sciences. Bayesian inference is then straightforward. The second part is innovative as the concept of imprecise probabilities has not yet found its way into application and testing of its practical usefulness.The suggested techniques will be applied to the problem of spatial prioritization of river rehabilitation measures in selected regions of Switzerland. This problem is of high practical relevance as there is a 80 year, 5 billion CHF program for river rehabilitation in place and the societal preferences to be used for spatial prioritization are not yet part of the established river assessment procedures in Switzerland.Expected results on the methodological side are an elicitation and post-processing procedure for the derivation of uncertain value functions from stakeholder responses and further developed methods for considering the ambiguity about priors and posteriors that quantify our knowledge of outcomes of decision alternatives and stakeholder preferences. On the applied side, the project intends to stimulate the discussion of river rehabilitation prioritization criteria in Switzerland and the comprehensive consideration of uncertainty and its communication.More detailed information about the planned methodological approach can be found in the paper Reichert (2020) "Towards a comprehensive uncertainty assessment in environmental research and decision support" (https://doi.org/10.2166/wst.2020.032) and about the framework for the application to river rehabilitation prioritization in Kuemmerlen et al. (2019) "Ecological assessment of river networks: From reach to catchment scale" (https://doi.org/10.1016/j.scitotenv.2018.09.019).