Exploring scenario and model uncertainty in cross-sectoral integrated assessment approaches to climate change impacts

Abstract

In this paper we present an uncertainty analysis of a cross-sectoral, regional-scale, Integrated Assessment Platform (IAP) for the assessment of climate change impacts, vulnerability and adaptation. The IAP couples simplified meta-models for a number of sectors (agriculture, forestry, urban development, biodiversity, flood and water resources management) within a user-friendly interface. Cross-sectoral interactions and feedbacks can be evaluated for a range of future scenarios with the aim of supporting a stakeholder dialogue and mutual learning. We present a method to address uncertainty in: i) future climate and socio-economic scenarios and ii) the interlinked network of meta-models that make up the IAP. A mixed-method approach is taken: formal numerical approaches, modeller interviews and network analysis are combined to provide a holistic uncertainty assessment that considers both quantifiable and un-quantifiable uncertainty. Results demonstrate that the combined quantitative-qualitative approach provides considerable advantages over traditional, validation-based uncertainty assessments. Combined fuzzy-set methods and network analysis methods allow maps of modeller certainty to be explored. The results indicate that validation statistics are not the only factors driving modeller certainty; a large range of other factors including the quality and availability of validation data, the meta-modelling process, inter-modeller trust, derivation methods, and pragmatic factors such as time, resources, skills and experience influence modeller certainty. We conclude that by identifying, classifying and exploring uncertainty in conjunction with the model developers, we can ensure not only that the modelling system itself improves, but that the decisions based on it can draw on the best available information: the projection itself, and a holistic understanding of the uncertainty associated with it.