Uncertainty analysis in integrated assessment: the users' perspective

Integrated Assessment (IA) models aim at providing information- and decision-support to complex problems. This paper argues that uncertainty analysis in IA models should be user-driven in order to strengthen science-policy interaction. We suggest an approach to uncertainty analysis that starts with investigating model users' demands for uncertainty information. These demands are called uncertainty information needs. Identifying model users' uncertainty information needs allows focusing the analysis on those uncertainties which users consider relevant and meaningful. As an illustrative example, we discuss the case of examining users' uncertainty information needs in the SEAMLESS Integrated Framework (SEAMLESS-IF), an IA model chain for assessing and comparing alternative agricultural and environmental policy options. The most important user group of SEAMLESS-IF are policy experts at the European and national level. Uncertainty information needs of this user group were examined in an interactive process during the development of SEAMLESS-IF and by using a questionnaire. Results indicate that users' information requirements differed from the uncertainty categories considered most relevant by model developers. In particular, policy experts called for addressing a broader set of uncertainty sources (e.g. model structure and technical model setup). The findings highlight that investigating users' uncertainty information needs is an essential step towards creating confidence in an IA model and its outcomes. This alone, however, may not be sufficient for effectively implementing a user-oriented uncertainty analysis in such models. As the case study illustrates, it requires to include uncertainty analysis into user participation from the outset of the IA modelling process.