Assumptions and Uncertainty in Climate Impact Modeling

Despite the significance of climate impact models outputs for society, thus far no systematic overview of assumptions and uncertainties contained in these models has been made. A detailed discussion of the uncertainties they entail would however be beneficial for sounder decisions under uncertainty. In this thesis, I develop a framework that allows for a structured and rigorous analysis of model assumptions and of the epistemic and ethical uncertainty they lead to. The framework consists of two steps: a reconstruction of the model’s justification scheme with regard to its adequacy-for-purpose assumption, and an evaluation and prioritization of the previously detected assumptions. I present a practical application of this framework on one example of climate impact modeling, namely the model CLIMADA. The practical application stresses the importance of the adequacy-for-purpose assumption and reveals that an overall model evaluation is not judicious. Rather, specific evaluations on a case-by-case basis are needed. While model’s quality and overall uncertainty cannot be generalized, many assumptions found in the San Salvador case study could nonetheless be generalized to the use of CLIMADA in general and to other climate impact models as most of the problematics are recurrent in this kind of models. The framework provides a good way to spark discussions on the most ambiguous assumptions to help decision-makers make sounder decisions under uncertainty, particularly regarding climate risk management options. It helps to go beyond the probabilistic assessment of uncertainties and proves useful in discussing ethical uncertainty inherent to climate impact models while not forgetting about epistemic uncertainty ubiquitous in climate change science and its modelling process. This thesis provides users, developers and decision-makers with questions to be asked and discussions to be conducted in order to foster an adequate use of the model