Land surface model parametrization using machine learning - Ph.D. project outline
By Lasse Keetz
In this seminar, Lasse Keetz, who commenced his Ph.D. fellowship in the group in October 2019, will present his project outline to discuss the planned research and schedule. The proposed projects seek to improve land surface model parametrization with a focus on the model representation of vegetation in northern latitudes. More specifically, it will be explored if machine learning (ML) techniques that have been shown to generate state-of-the-art results in numerous ecological applications (e.g. Species Distribution Modelling) can be used to create a high-accuracy vegetation distribution map for Norway based on a national area frame survey of vegetation types. Moreover, a subsequent project aims at applying ML to efficiently constrain a dynamic global vegetation model (DGVM) with observational data. The seminar will include a brief introduction to relevant terms and principles of machine learning.