The proposed IMITATE project aims to address the following questions:
- How well can machine learning methods emulate physical process-based land surface models, focused over Europe?
- Can explainable AI techniques provide new insights into process understanding when combining land surface models and Earth Observation data
- Are the learnt relationships between the modeled inputs and outputs consistent with those from Earth Observation data?
To do this the project will:
- produce land surface model simulations from the JULES ESM (Earth System Model) over Europe focusing on the carbon cycle
- develop, train and evaluate machine learning models (emulators) against the simulated land surface parameters
- use these emulators to investigate the complex emergent relationships and feedback to gain an increased understanding of the underlying Earth System processes and to test whether data from satellite-based essential climate variables (e.g. ESA-CCI) are consistent with the relationships learnt from the land surface models.
- produce an Emulated-GPP (gross primary productivity) data product based on EO data, using the relationships learnt from the land surface model.