Figurit Homepage
news

SASIP December Newsletter

Follow Us

AGU SASIP Highlights

We’re thrilled to share that exciting research from our project will be showcased at AGU24 in Washington D.C. from December 9–13!

Cryosphere ❄️

Monday 9th

  • 16:42 (C14A-05): Simon Driscoll creates the first observational emulator of melt ponds on Arctic sea ice using machine learning (Abstract).

  • 13:40 - 17:30 (Poster Hall)

    • Christopher Horvat share exciting insights on Global gridded products of sea ice concentration, wave climate, and sea ice floe size distribution with ICESat-2: examining marginal ice zone variability and passive microwave uncertainties (Abstract).


    • Aikaterini Tavri investigates Wave-Induced Ocean Surface Mixing in the Arctic Using Coupled Wave - Sea Ice Model Simulations (Abstract).


    • Serena Vu showcase her research on Retrieving Ocean Eddy Signatures in the Marginal Ice Zone from Sea Ice Drift Fields Derived from Sequential Synthetic Aperture Radar Observations (Abstract).

Non Linear Geophysics 🔬

Thursday 12th

  • 08:30 (NG41A): Simon Driscoll convenes a presentation on Data-Driven Science: Developments in Machine Learning Subgrid-Scale Parameterizations and in Reanalyses Across Earth System Modeling (Abstract).


  • 13:40 - 17:30 (Poster Hall): Simon Driscoll introduces Data-driven emulation of melt ponds on Arctic sea ice (Abstract).

Atmospheric Sciences 🌫️

Friday 13th

  • 08:30 - 12:20 (Poster Hall): Bingjie Zhao design An Adaptive Threshold Selection Method for Supporting Robust Intercomparisons of Extreme Climate Events (Abstract)

  • 10:42 (A52B-03): Melanie Lauer gives a talk on Assessing the interaction between precipitation and sea ice in Antarctica (Abstract).

News

Join the SASIP December Webinar on Thursday, Dec 19 at 2:30 PM CET!

“Overcoming the limitations of data assimilation in complex physical settings with machine learning” with Ivo Pasmans from the University of Reading.  


Popular data assimilation (DA) methods, designed to improve estimates of the truth by incorporating information from observations, typically assume that errors are Gaussian and model are linear. However, these assumptions fail in the context of complex sea ice models. In this presentation, Ivo Pasmans will outline the fundamentals of DA and demonstrate how machine learning techniques can help address its limitations. Specifically, he will show that applying DA in the latent space of a variational autoencoder enables the generation of post-DA states that adhere to the physics embedded in the model.

Missed our latest webinar? No worries, you can watch it now on Youtube!

“From alpha to beta ocean: Exploring the role of surface buoyancy fluxes and seawater thermal expansion in setting the upper ocean stratification”, with Romain Caneill from CNRS, France.

.
.
  • This project is supported by Schmidt Sciences, LLC.