Analyses of costs and benefits require the prediction of the effects of restoration measures and the quantification of societal values. Both of these estimates are uncertain. In this report, some of the key issues related to the assessment, description and quantification of uncertainty are discussed and guidelines are provided for considering uncertainty.
This report provides a brief overview on the representation and quantification of uncertainty in scientific prediction followed by examples of typical risks associated with river restoration that could lead to unintended, adverse effects and in more detail, how uncertainty can be considered in CEA/CBA and in MCDA.
There are two important sources of uncertainty to consider in environmental management in general, and in particular for river restoration:
- Uncertainty about scientific predictions of outcomes.
Depending on alternatives, this requires prediction and uncertainty estimation of the behavior of a natural system, natural-technical system, or even of a combined natural-technical-socio-economic system (e.g. in case of measures that include incentives to some of the affected stakeholders). In particular, one has to consider the potential for adverse outcomes as discussed in chapter 3.
- Uncertainty about the preferences of the society elicited from inquiries or stakeholders.
In addition to the difficulties of the stakeholders to be aware of their own preferences and to be able to quantify them, this also includes their risk attitude (how uncertainty about the outcomes affects their preferences).
- Communication of uncertainty is a key element of any communication of scientific predictions. Visualization of uncertainty ranges can support this task. Lack of communication of scientific uncertainty in the past led to a reduction of trust of stakeholders to scientists.
- Clearly separating scientific predictions and societal valuations is an essential element of any decision support procedure. Uncertainties in both elements should be clearly communicated separately. In particular if there are disagreements among experts about scientific predictions and of stakeholder groups about preferences.
- Uncertainty about scientific predictions can be addressed by probability distributions and scenarios; uncertainty about societal preferences are often better addressed by sensitivity analysesof the ranking of the alternatives resulting from combining predictions of the outcomes of decision alternatives with preferences.