061 Hybrid data assimilation and deep learning system for Earth monitoring

  • Full Time
  • Toulouse

Job title: 061 Hybrid data assimilation and deep learning system for Earth monitoring

Company: Centre National d’Etudes Spatiales

Job description: 24-061 Hybrid data assimilation and deep learning system for Earth monitoring


24-061 Hybrid data assimilation and deep learning system for Earth monitoring

  • Doctorat, 36 mois
  • Temps plein
  • Aucune expérience exigée
  • Maitrise, IEP, IUP, Bac+4
  • Digital technologies for remote sensing



This Phd is proposed within the framework of the RELEO chair, which is part of ANITI and supported by CNES.

Essential Variables (EVs) are key indicators to adequately describe and monitor the evolution of Earth’s changing climate. Accordingly, the regular and accurate mapping of variables such as Leaf Area Index, Land Surface Temperature, Evapotranspiration or Soil Moisture are becoming increasingly important for environmental applications.

Satellite imagery is an efficient tool to effectively predict EVs trends, define present conditions, and inform about their future evolution. Specifically, Sentinel missions offer incredible opportunities to provide accurate, timely and easily accessible information of the Earth surface. However, extracting useful information from raw satellite images acquired by multimodal sensors, which are nonstationary, not dense, multi-variate and have temporal gaps is challenging.

The retrieval of EV from noisy satellite observations is traditionally seen as an inverse modelling problem. In recent years, physics-constrained data-driven methodologies have been proposed to infer values of the physical model parameters from observations. Without relying on any simulation data, these methods have been proposed to train a deep neural network to map satellite observations into the physical model parameters in an unsupervised manner. To perform it, these methods consider that the parameters of the inverse problem are the latent variables of a semantic variational autoencoder in which the decoder part is a physical model incorporating knowledge about data generation. With the increasing availability of large scale datasets, this unsupervised framework is an attractive choice for modelling and forecasting the space-time EV dynamics from satellite observations.

This PhD aims to predict EVs’ future states by combining machine learning (ML) and data assimilation (DA) strategies from noisy and sparse satellite observations. On one side ML can benefit from DA, which accommodates real observations that are noisy, sparse, and only indirectly related to the physical state of interest like EV. DA follows the Bayesian approach to represent uncertainties of the observations, and retains existing physical knowledge. On the other side, DA could benefit from ML in handling more complex models and error distributions.

Given the unknown dynamical time propagation model for EVs, we will first focus on a fully data-driven physical model for forecasting EVs on hypothetical situations which are obtained by forcing specific assumptions about future weather scenarios. Once a dynamical model is developed, data assimilation strategies will be integrated to use real-time observations to progressively adjust the trajectories of EVs over time.

For the fully data-driven approaches, we will focus especially on learning dynamics in the reduced space known also as reduced order models (ROMs). These models may reduce the computational cost by approximating the large-scale systems by much smaller ones. Estimating dynamics in this reduced space can help prevent overfitting, especially in scenarios where the dynamics are uncertain and observations are noisy. Given the challenges of noisy observations and large-scale systems, we will consider ROMs in which the reduced space and the dynamics (within this reduced space) is learnt concurrently using ML techniques. We may consider using Koopman operators in the reduced (latent) space which enables to train such a dynamical model for long-term continuous reconstruction, even in difficult contexts where the data comes in irregularly-sampled time series. One can train these models directly on the physical values of the EVs. As an alternative one can also consider a loss function based on data log-likelihood in which the Gaussian probabilities can be evaluated by using the Ensemble Kalman filter. Within this project we will investigate different strategies.

Once a fully data-driven dynamical model is obtained, the trajectory of EVs over time can be progressively improved by considering DA strategies. DA uses the model trajectories, observations and a priori information to estimate the state of the system based on the uncertainties within a Bayesian framework. Traditional DA provides a weighted linear combination of the available information in which the weights are defined according to the uncertainties. Thanks to the latent space structure obtained from the previous step, we will consider applying data assimilation in the latent space. This provides a weighted nonlinear combination of the available information which may increase the accuracy.

From a methodological point of view, this Phd proposes to study the combination of machine learning and data assimilation strategies which is one of one of the hottest topics in Artificial Intelligence. The research conducted here will be integrated in ANITI 2.0 which CESBIO and CERFACS are actively involved.

For more Information about the topics and the co-financial partner (found by the lab !); contact Directeur de thèse –

Then, prepare a resumé, a recent transcript and a reference letter from your M2 supervisor/ engineering school director and you will be ready to apply online before March 15th, 2024 Midnight Paris time !


We are looking for enthusiastic people to join our interdisciplinary research group. The candidates must have a Master’s level in computer science (data science or similar). Applicants should preferably have a strong background in mathematics, signal and image processing and machine learning. A good knowledge of English and scientific programming is required.

Expected salary:

Location: Toulouse

Job date: Fri, 02 Feb 2024 01:32:04 GMT

Apply for the job now!

Submit your Resume!

Do you like the ai jobs 24 ?

Powered By Wischi | CW from Jobs in Germany.net