Prof. Dr. Nikki Vercauteren – Unresolved scales in weather and climate models: learning uncertainty from data

This semester, we are excited to announce that the seminar series will be hosted by our collaborative partner ECMWF at their buildings in Bonn and it will be streamed online (via zoom).CESOC continues the seminar series “My Research” this Summer term 2024 with
Prof. Dr. Nikki Vercauteren
from the Institute Geophysics and Meteorology at the University of Cologne, talking on their work
“Unresolved scales in weather and climate models: learning uncertainty from data”
when: on Tuesday, 07 May 2024 at 11:00 am (CEST)
It is open to any interested person within the CESOC research disciplines (any Earth system sciences, mathematics or computer science).
Please contact info[@]cesoc.net, if you would like to participate.
Full Schedule could be found here!
Abstract:
Limited computer resources lead to a simplified representation of unresolved small-scale processes in weather and climate models, through parameterisation schemes. These approximate models of the unresolved scales express unresolved processes using the resolved variables of the prediction model and are often deterministic, but should sometimes be stochastic.
A systematic data-driven approach can help quantifying the uncertainty of parameterisations and inform us on how and when to incorporate uncertainty in the modelling, through stochastic parameterisation schemes. To enable such a systematic data-driven approach, methods from machine learning and uncertainty quantification were combined in a model-based clustering framework. As a result, stochastic parameterisation can be learned from observations. The method is able to retrieve a hidden functional relationship between the parameters of a stochastic model and the resolved variables.
Among the parameterised processes in weather and climate models, turbulent fluxes exert a critical impact on the exchange of heat, water and carbon between the land and the atmosphere. Turbulence theory was, however, developed for homogeneous and flat terrain, with stationary conditions. The theory fails in unsteady flow contexts or with heterogeneous landscapes, leading to important uncertainty in the parameterization of turbulence. Using field measurements of turbulence, the stochastic modelling framework is able to uncover a stochastic parameterisation that represent unsteady mixing in difficult conditions.