GraphDeco

Deep learning speeds up ice fow modelling by several orders of magnitude
Presentation | Team members | Collaborations | Publications | Job offers | Contact

 

Deep learning speeds up ice fow modelling by several orders of magnitude

Guillaume Jouvet, Guillaume Cordonnier, Byungsoo Kim, Martin Lüthi, Andreas Vieli, Andy Aschwanden
Journal of Glaciology, page 1-14 - 2021
Download the publication : JOG-21-0059.pdf [7.5Mo]  
This paper introduces the Instructed Glacier Model (IGM) – a model that simulates ice dynamics, mass balance and its coupling to predict the evolution of glaciers, icefields or ice sheets. The novelty of IGM is that it models the ice flow by a Convolutional Neural Network, which is trained from data generated with hybrid SIA + SSA or Stokes ice flow models. By doing so, the most computationally demanding model component is substituted by a cheap emulator. Once trained with representative data, we demonstrate that IGM permits to model mountain glaciers up to 1000 × faster than Stokes ones on Central Processing Units (CPU) with fidelity levels above 90% in terms of ice flow solutions leading to nearly identical transient thickness evolution. Switching to the GPU often permits additional significant speed-ups, especially when emulating Stokes dynamics or/and modelling at high spatial resolution. IGM is an open-source Python code which deals with two-dimensional (2-D) gridded input and output data. Together with a companion library of trained ice flow emulators, IGM permits user-friendly, highly efficient and mechanically state-of-the-art glacier and icefields simulations.

Images and movies

 

See also

Full source code and datasets are available at Source Code .

Acknowledgements and Funding

We acknowledge Ed Bueler, Fabien Maussion and an anonymous referee for their valuable comments on the original manuscript. We are thankful to Constantine Khroulev and Olivier Gagliardini for support with PISM and Elmer/Ice, respectively. The Python code PyPDD by J. Seguinot has greatly helped the integration of the PDD model in IGM. The emulator was originally motivated to speed-up an ice fow model within the framework of project 200021-162444 supported by the Swiss National Science Foundation (SNSF). Michael Imhof is acknowledged for providing the script that served to produce Fig. 5.

BibTex references

@Article{JCKLVA21,
  author       = "Jouvet, Guillaume and Cordonnier, Guillaume and Kim, Byungsoo and L{\"u}thi, Martin and Vieli, Andreas and Aschwanden, Andy",
  title        = "Deep learning speeds up ice fow modelling by several orders of magnitude",
  journal      = "Journal of Glaciology",
  pages        = "1-14",
  year         = "2021",
  url          = "http://www-sop.inria.fr/reves/Basilic/2021/JCKLVA21"
}

Other publications in the database

» Guillaume Jouvet
» Guillaume Cordonnier
» Byungsoo Kim