a Workflow for the Application of Sensitivity Analysis to Earth System Models

Predictions of any earth system model are affected by unavoidable and potentially large uncertainty. When models are used to support risk management of natural hazard, such uncertainties can undermine the transparency and defensibility of the risk assessment. When models are applied to understand dominant controls or other aspects of the system under study, uncertainties will reduce our ability to chose between competing hypotheses. Sensitivity Analysis (SA) provides quantitative information about the contribution of the different input factors (e.g. parameters, boundary conditions or forcing data) to such uncertainty. SA application thus provides insights into the model behavior and potential for model simplification, indicates where further data collection and research is needed or would be beneficial, and enhances the credibility of our modelling results. The value of such analysis has motivated an increasing research effort in the development, application and comparison of SA techniques. Still, comprehensive understanding to guide choices between available SA methods and practical guidelines for their application in the context of earth system models is still insufficient. In this contribution, we aim at filling this gap by (i) providing a map of the existing SA techniques and their appropriateness in different contexts of earth system modeling; (ii) developing a workflow for the choice and application of SA techniques to environmental models; (iii) presenting a suite of visualization tools that can support the assessment and communication of SA results; (iv) defining challenges and opportunities for future research.