Traitement automatique de données océanographiques pour l'interpolation de la ∫CO₂ de surface dans l'océan Atlantique tropical, en utilisant les données satellitaires

Abstract : This thesis work consists of using satellite data of SST (sea surface temperature), SSS (sea surface salinity), and Chl-a (chlorophyll-a), in order to interpolate the CO2 fugacity (fCO2) in the surface of the tropical Atlantic ocean, for seasons of the period 2002-2013. Three data types were used: in situ (SOCAT V.3 DB (database)); satellite (MODIS-A, Sea-WIFS, and SMOS sensors); and assimilated (SODA V.2.2.4 DB). The first step was the data classification based on SST. The second step was the fCO2 interpolation (for each class of each season), using feedforward NNs (artificial neural networks) with a backpropagation learning method. Obtained results (RMSEs (root mean square error) between 8,8 and 15,7 µatm) confirm the importance of: process each season separately, pass through data classification step, and choose the best NN on the basis of generalization step results. This allowed the development of 138 monthly fCO2 CSV (Comma-separated values) file, with 4 km x 4 km spatial resolution, for the period from July 2002 to December 2013.
Document type :
Theses
Liste complète des métadonnées

https://tel.archives-ouvertes.fr/tel-01417518
Contributor : Abes Star <>
Submitted on : Thursday, December 15, 2016 - 4:38:11 PM
Last modification on : Thursday, October 4, 2018 - 10:02:26 AM
Document(s) archivé(s) le : Thursday, March 16, 2017 - 4:33:53 PM

File

These_MOUSSA_HADJER_2016.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01417518, version 1

Collections

Citation

Hadjer Moussa. Traitement automatique de données océanographiques pour l'interpolation de la ∫CO₂ de surface dans l'océan Atlantique tropical, en utilisant les données satellitaires. Océan, Atmosphère. Université de Perpignan, 2016. Français. ⟨NNT : 2016PERP0025⟩. ⟨tel-01417518⟩

Share

Metrics

Record views

267

Files downloads

559