Offre en lien avec l’Action/le Réseau : MACLEAN/– — –

Laboratoire/Entreprise : CESBIO
Durée : 12 months
Contact : mathieu.fauvel@inrae.fr
Date limite de publication : 2022-06-20

Contexte :
Satellite remote sensing is an active field of research, with application in environmental, agricultural and climate change science. Several satellite missions have been launched in the last decades, such as the European Copernicus program, and provide massive Earth observation open access data.
Among the various products obtained from these missions, large scale land cover mapping is surely the most operational. Nowadays, such mapping is used for a large range of environmental applications and of primarily importance in the context of climate change. Several open and pri- vate achievements were announced recently (e.g., https://www.theia-land.fr/en/ceslist/land- cover-sec/ or https://viewer.esa-worldcover.org/worldcover) but this topic is still an impor- tant active field of research and engineering. One of the main limiting problems is the ability to efficiently process the very large amount of data that researchers and engineers are faced with.
In this context, the open-source software iota2 (source: https://framagit.org/iota2-project/ iota2, documentation: https://docs.iota2.net/develop/) has been initiated by the CESBIO-Lab as a generic processing chain to fully process recent satellite time series, such as SENTINEL-1 and SENTINEL-2 or Landsat-8. It allowed to produce the first map of the land cover over the metropolitan French territory (e.g., https://theia.cnes.fr/atdistrib/rocket/#/collections/ OSO/21b3e29b-d6de-5d3b-9a45-6068b9cfe77a).
To extend the development of the software, outside of the CESBIO lab, the PARCELLE project was set up to foster the applicability of iota2 to other large scale mapping problems. Three main topics are considered in the project.
1. A quantitative and qualitative assessment of the performances of iota2 for different types of landscapes (e.g., South-Africa or South-America) and/or different land cover types.
2. The methodological integration of state-of-the-art algorithms from the project partners.
3. Promote the usage of iota2 trough training and scientific meeting.
Ultimately, the improvements of the chain will be used to enrich several Centre d’Expertise Scientifique (CES) of the national data center Theia

Sujet :
The first mission of the recruit is to work on the development of new features for iota2, such as deep learning algorithms applied at large scale (super-resolution, classification, inversion . . . ). Appli- cant could check the project repository for more details (https://framagit.org/iota2-project/ iota2/-/issues).
The second mission is related to give training for others members of the project and institutional users (e.g. https://docs.iota2.net/formation/ and its repository https://gitlab.cesbio.omp. eu/fauvelm/formation-iota2). Also, some times will be devoted to answer users questions (mainly trough the issues interface of the gitlab repository).
The third mission of the recruit will be to coordinate the different developments carried out by the partners. As such, other issues may emerge during the project.

Profil du candidat :
The applicant must have a solid background in python (numpy, pandas, scikit learn, pytorch), sci- entific computing, linux, and distributed version control system (git). Experience in software doc- umentation (docstrings, sphinx) will be appreciated, as well as some knowledge in remote sensing image processing, geomatics, geographic information systems and data bases.
The applicant should send a detailed CV, motivation letter, reference letters and, if possible, links to developed software to the contacts.

Formation et compétences requises :
The applicant must have a solid background in python (numpy, pandas, scikit learn, pytorch), sci- entific computing, linux, and distributed version control system (git). Experience in software doc- umentation (docstrings, sphinx) will be appreciated, as well as some knowledge in remote sensing image processing, geomatics, geographic information systems and data bases.
The applicant should send a detailed CV, motivation letter, reference letters and, if possible, links to developed software to the contacts.

Adresse d’emploi :
CESBIO,
Centre d’Etudes Spatiales de la Biosphère, 31400 Toulouse

Document attaché : 202206081219_cdd_parcelle.pdf