Description du sujet de thèse

The tropical Atlantic Ocean experiences significant variability across multiple temporal and spatial scales, with impacts on climatic conditions and extreme events in surrounding countries (e.g., rainfall, cyclones). The PIRATA (Prediction and Research Moored Array in the Tropical Atlantic) observation network, established in 1997 by France, Brazil, and the United States, plays a key role in monitoring the region's climate variability and understanding the underlying mechanisms. The network includes 18 metocean moorings that require annual oceanographic campaigns. This project aims to optimize the tropical Atlantic observing system in the context of reducing greenhouse gas emissions. This includes formulating recommendations by proposing various configurations under different reduction scenarios and timeframes. The assessment will rely on process-oriented indicators and a methodology accessible to non-specialists, in order to facilitate the transfer of results toward a sustainable, integrated tropical Atlantic observing system.

The project methodology is based on three main components:
(1) the implementation of a set of climate-relevant metrics commonly used to validate climate models, including major climate modes (such as the AMM and Atlantic Niño), the seasonal cold tongue, Kelvin and Rossby wave dynamics, and thermocline depth variations;
(2) the development of a regional monthly gridded product using advanced methods such as machine learning, based on all available in-situ data (PIRATA, ARGO) over the 2000–2024 period. Strategies ranging from optimal interpolation to advanced neural networks based on Implicit Neural Representations will be explored to generate this gridded dataset and quantify its uncertainties;
(3) the execution of Observing System Simulation Experiments (OSSEs) using outputs from regional NEMO simulations, testing various scenarios (e.g., number and type of observations, reduction and/or relocation of PIRATA moorings), with the primary goal of assessing the required sampling levels and the complementarities among different observational datasets.

Contexte de travail

This PhD is funded by the MITI (Mission for Transverse and Interdisciplinary Initiatives) of the CNRS and thus has a strong transdisciplinary dimension, bridging Physical Oceanography and Numerical Methods. The doctoral candidate will be encouraged to integrate into the PIRATA oceanographic scientific community at LEGOS, as well as participate in the numerical methods research conducted at IRIT, in order to build connections between these two disciplines and foster fruitful collaboration. A distribution of time between LEGOS and IRIT will be discussed at the beginning of the project to ensure the student benefits from both complementary research environments. Workshops involving other teams, such as INRIA in Grenoble, will be organized throughout the project.

Contraintes et risques

We are looking for a highly motivated candidate with an excellent academic background, holding a MSc degree (or equivalent) in physical oceanography or numerical sciences, and with a strong interest in the interdisciplinary connection between these two fields. Good writing and communication skills in English are required.