Focus on… Maike Sonnewald and ocean models

on the January 14, 2019

Maike Sonnewald is a postdoctoral researcher working at MIT EAPS (Earth, Atmospheric and Planetary Sciences) and Harvard. She has been greeted by GIPSA-Lab and IGE with the support of Grenoble Alps Data Institute. For two weeks, she has shared her expertise at the interface of oceanography and complex system simulation.

The visit of Maike Sonnewald is part of a two-year project between the MEOM team of the Institute of Environmental Geosciences (IGE, UMR of Grenoble Alpes University, IRD, Grenoble INP and CNRS) and the GIPSA-Lab (UMR of Grenoble Alpes University, Grenoble INP and CNRS). The MEOM team produces numerical ocean simulations and the GIPSA-Lab conducts research on the statistical processing of data. Five scientists from Grenoble are involved: Pierre-Olivier Amblard (GIPSA-Lab), Sally Close (IGE), Jeanne Lefevre (GIPSA-Lab), Nicolas Le Bihan (GIPSA-Lab) and Thierry Penduff (IGE).

Maike Sonnewald is driving the intersection of modern statistics/machine learning and traditional modeling/theory work, and sees the opportunity to help the GIPSA and IGE labs meet as a wonderful opportunity. Having expertise working with oceanography and data science, Dr Sonnewald maintains that oceanographers need work together with data scientists to learn more about the ocean. Together, they are working at the exciting interface between disciplines to find a common language and leverage their expertise and learn more about the ocean.

In the framework of the OCCIPUT project, the MEOM team has produced a pioneering ensemble of fifty gloal ocean simulations to study the chaotic features of the ocean circulation and their sentivity to the atmospheric variability. The same atmospheric variability is prescribed on the fifty members, whose initial conditions were slightly perturbed, and the variability of the fifty oceans are compared. The hypothesis is that there are localized sources of chaotic variability that propagate in the rest of the ocean.


Legend : Contribution of the oceanic chaos to the interannual-to-decadal variability of climate-relevant ocean indices in the North Atlantic (America on the left, Europe on the right): Northward heat transport by the ocean acrouss 41°N and 59°N, oceanic heat content contained between these two latitudes, integrated across the basin from the surface to the abyss.

The collaboration between M. Sonnewald, MEOM and GIPSA-Lab scientists aims to develop tools and methods to process the massive OCCIPUT dataset and add value to this experiment. Machine learning techniques will help reveal links between chaotic events taking place in different parts of the Atlantic Ocean. These connections will then have to be interpreted by oceanographers.

Methodological and scientific outcomes are expected from this collaborations. Beyond the knowledge gained about ocean dynamics and processing techniques, the tools that will be developed may be useful for other climate sciences. Applications are also expected for ensemble ocean forceasting such as those performed by Mercator-Océan.
For more information

Data Science for climate science: et
The thermal imprint of the oceanic chaos:
OCCIPUT web site:
The OCCIPUT simulation:

Published on January 14, 2019