This month we publish a video from work in progress in SASIP-WP4, provided by C. Durand, T. Finn, A. Farchi, M. Bocquet, E. Òlason @ CEREA, École des Ponts.
Fig.: Movie produced by C. Durand, T. Finn, A. Farchi, M. Bocquet, E. Òlason @ CEREA, École des Ponts This is a work in progress in WP4. By using simulation outputs from neXtSIM and some ERA5 reanalysis atmospheric forcings, we trained a neural network to predict the sea-ice thickness up to 12 hours. Then, by cycling the learned neural network, we built a surrogate model of neXtSIM sea-ice thickness. This surrogate model is shown in the right panel of the video, while neXtSIM simulation output is shown on the left panel. The surrogate model can exhibit large scale dynamics of the sea-ice thickness over the full year, with good advection properties, while fine-scale dynamics are diffused. This work has led to the submission of a paper in The Cryosphere (https://doi.org/10.5194/egusphere-2023-1384) and the video is identified under the following DOI https://doi.org/10.5446/62131.news
