Ewan Pinnington, ECMWF, presented CERISE results at the American Meteorological Society (AMS) conference in Baltimore, MD, USA.
The 5 day conference was themed around "Living in a Changing Environment" and brought together scientists and researchers from around the globe.
Ewan's presentation on “Towards Ensemble Land Data Assimilation at ECMWF” was part of the “Land Data Assimilation for Improved Model Output” session on Thursday 1st Feb 2024.
This talk was an opportunity to showcase Ewan's work from CERISE; incorporating more ensemble information into the ECMWF land surface data assimilation system which has enabled an improvement in forecasts of surface temperature. Ewan reports that there was a good response from other attendees and collaborations were strengthened with colleagues at NOAA and NASA who are experimenting with similar techniques and who shared valuable insights into our results and how we might better condition our ensemble forecasts.
Ewan reports that Artificial Intelligence & Machine Learning were big topics at this year’s AMS with many talks highlighting the benefit and physical validity of new data-driven forecasting systems, including ECMWF’s own AIFS, which created a great deal of buzz. Ewan comments "On CERISE we have developed a machine learning model of the land surface for creating large ensembles to characterise uncertainty in surface conditions". This topic was also included in Ewan's talk.
There were several other talks leveraging machine learning to emulate the response of the land surface to the atmosphere which sparked new ideas for future directions with this work.
