Seminar: Sophia Walther

Combining in-situ measurements of terrestrial CO2 fluxes with relevant information on environmental conditions from in-situ and satellite observations using machine learning is a key method for mapping terrestrial fluxes to regional or global domains. However, data-driven upscaling is inherently dependent on data quantity, quality, the representativeness of flux data, and the relevance of predictors. Observational conditions, both in-situ and spaceborn, are particularly challenging in the Arctic-Boreal zone (ABZ). Modeling these highly vulnerable and heterogeneous regions can therefore potentially benefit from domain specific data streams.

We systematically explore data-driven terrestrial CO2 flux estimates in the ABZ originating from products with both a focus on ABZ and the global domain, aiming at understanding the robustness of flux magnitudes and their spatiotemporal dynamics.