Talk: Mathias Frisch (Hannover)

Title:

Incorporating Machine-Learning Models into the Modeling Hierarchy for Understanding Climate Change: Prospects and Challenges

Abstract:

Recent years have seen an increasing call for need of “process understanding” to supplement projections based on large global climate models. At the same time there is a growing reliance on data-informed machine learning models (ML-models) as a tool in climate change projections and attributions. Indeed, ML-models have become so prominent that some modelers have suggested that they should be given their own place in the climate modeling hierarchy. The aim of this talk is to examine whether there exists a pragmatic tension between these two developments and to what extent ML-models may be able to contribute to our understanding of climate change.