This five-day course focuses on describing Data Assimilation (DA) methods and general aspects of assimilating observations. Aspects of the practical implementation of the assimilation techniques for real-size numerical weather prediction (NWP) systems will also be described.
A new focus of the course will be to describe how Machine Learning techniques are being incorporated in the traditional DA workflow, and discuss what advantages they can bring in terms of both performance and efficiency.
As well as lectures, there will be discussion and hands-on sessions.
Main topics
- The fundamental data assimilation concepts
- Optimal Interpolation, 3D-Var, 4D-Var and the Kalman filter
- Ensemble Kalman Filter methods; Ensemble of Data Assimilations and uncertainty estimation; Hybrid variational/ensemble based methods
- Modelling of error covariances; handling of non-Gaussian errors
- Machine Learning for DA: model error estimation and correction, generative AI applications, hybrid modelling
- The global observing system, with emphasis on how to use satellite observations
- Bias correction, quality control and diagnostics
- Applications of data assimilation methods for the land surface, ocean, atmospheric composition and reanalysis
Requirements
Participants should have a good meteorological and mathematical background, and in particular a good understanding of linear algebra. They are expected to be familiar with the contents of standard meteorological and mathematical textbooks.
Introductory material not covered by the course can be found in our lecture note series.
Some practical experience in numerical weather prediction is an advantage.
All lectures are given in English.