• Login
All challenges
      • SAF05 - Downscaling Oceanographic Data

          • Problem:
            The ocean circulation is a central player in regulating Earth’s climate and supporting marine life by transporting heat, carbon, oxygen, and nutrients throughout the world’s ocean. Small-scale perturbations in the ocean have been shown to have a tremendous consequences in the global circulation. Yet, our understanding of the processes governing these fluxes is still limited because of the relatively small spatial scales involved, which are difficult to observe with current remote sensing techniques.

            Situation & idea:
            One technique commonly used to enrich the wealth of information contained in satellite data is downscaling. It consists in reconstructing a high-resolution observation from a low-resolution one’s. This problem is similar to image super-resolution in the computer vision and pattern recognition community, which has shown tremendous progress in recent years thanks to deep learning. Given the availability of large datasets for ocean remote sensing, it appears very tempting to investigate the potential of deep learning models to downscale satellite-derived observations.
          • What the challenge owner would like to develop over 48h
          • The challenge is to develop an algorithm and train deep learning models to downscale oceanographic data from numerical simulations. The overall goal is to then be able to use this algorithm with real satellite data to better understand the small spatial scales involved in ocean circulation.
          • Which skills the challenge owner is looking for
          • Computer Scientist
            Data Scientist
Campus mondial de la mer
Technopôle Brest-Iroise
525, Avenue Alexis de Rochon
29280 Plouzané
  • Brest Métropole
  • Région Bretagne
  • https://www.tech-brest-iroise.fr/
Lettre d'informations : Désinscription | Mailings : Désinscription
Mentions-légales | Plan du site | Powered by diasite
Designed by diateam