Postdoc position
Title: Synergy between ocean observing systems over boundary currents
Description
We are looking for a “data scientist” to come and work at LOPS in the Ocean and Climate team. The objective of this postdoc is to develop a new ocean product that will statistically combine complementary information from satellites and in situ ocean measurements to produce a 3-dimensional time series with sufficient resolution to track eddies in the Gulf Stream region.
Location: Laboratory for Ocean Physics and Satellite remote sensing, Ifremer – Brest, France.
Contract: 18 months
Starting date: As soon as possible
How to apply
Applications (detailed CV including a list of publications, cover letter explaining how your interests and experiences would make you a suitable candidate, name/email of at least 2 referees) and/or informal inquiries should be sent to Guillaume Maze (gmaze@ifremer.fr). Deadline: November 15th 2020
Who can apply:
A PhD in mathematics or information and communication science and technology with experience in environmental science, but also a PhD in physical oceanography, meteorology or climate science with strong experience in statistics.
Details on the project
The ocean is a turbulent and chaotic dynamical system, therefore complex to understand and predict. This is particularly true in the western regions of the mid-latitude oceans, where very large currents narrow to intense fronts, meanders and numerous eddies about 100km wide. These regions are called “western boundary current extensions”. Because they are turbulent, these regions are one of the main sources of uncertainty in the assessment of the ocean’s role in climate and their behaviour under the influence of global climate change is still limited. These uncertainties can be reduced if we improve our ability to diagnose the temporal evolution of the three-dimensional structure of the extensions of the western edge currents.
None of the existing ocean observing systems is able to provide an accurate time series of three-dimensional eddy scale thermal fields for these western edge current extensions. Satellite measurements have a high local frequency and precise horizontal resolution, but they capture only the surface signature of the ocean’s interior structures. On the other hand, in situ measurements of the interior of the ocean collected by autonomous probes (such as Argo floats) provide a very accurate vertical thermal structure of the ocean, but they are rare and lack horizontal resolution.
Activities
The product will be developed using a methodology based on neural network regression methods that have already been successfully tested in this context (Tokunaga and Maze, 2019; Agudelo, 2020). Neural networks are able to model complex and highly non-linear relationships between explanatory variables and therefore have the potential to perfectly capture the relationships between different ocean, satellite and in-situ observations.
This postdoc proposes to work on real data for the turbulent Gulf Stream region of the North Atlantic Ocean, while keeping in mind possible applications to other ocean currents. We will apply the regression method to predict a time series of the three-dimensional structure of the ocean temperature. The model training will be done on satellite sea surface height data co-located with deep, in-situ temperature data. Several neural network architectures will be tested and compared. The statistical properties of model predictions (gradient intensities, covariance matrices) will be analyzed and compared to those of more classical data interpolation techniques.
The procedure for generating the new product and its documentation will be made available to the partners of the EARISE project.
Context
The work of this postdoc is part of a long collaboration between LOPS and IMT-A (Pierre Tandeo) and takes part in the H2020 EARISE project and its WP2.3, which aims at improving the extraction of information from the Argo network in boundary currents. This project involves many European partners, and within this topic more specifically BSH (German Institute of Oceanography) and SOCIB (the institute in charge of the Observing and Forecasting System of the Balearic Islands Coast). The postdoc will present his work during the project meetings but also in conferences and publications.