Practical information

November 21, 2025, at 2 p.m. in the Pal B1.01 amphitheater on the Agro-Saclay campus, 22 place de l’agronomie in Palaiseau.

Supervision

Emmanuelle VAUDOUR, DR INRAE, UMR EcoSys Palaiseau
Dominique ARROUAYS, IREx INRAE, UR Info&Sols Orléans
Anne RICHER-de-FORGES, IR INRAE, UR Info&Sols Orléans

Abstract

Soil organic carbon (SOC) plays an essential role in agricultural sustainability and climate change mitigation. The growing availability of satellite Earth observation data offers new opportunities to quantify and monitor SOC on a large scale. However, the production of accurate SOC maps from remote sensing is highly dependent on the quality of bare soil mosaics. This thesis explores how bare soil mosaics derived from Sentinel-2 time series can be optimized to improve SOC prediction at the regional scale, in particular through the use of spectral index thresholds.

Section 1 presents the conceptual framework and the main scientific questions related to the use of remote sensing bare soil mosaics for COS mapping. Section 2 describes the study areas and the datasets used. The appropriate use of spectral indices requires a systematic understanding of their effectiveness and limitations. Section 3 therefore evaluates a wide range of spectral indices focused on soil monitoring, highlights the problems posed by their application, and proposes a reference catalog for future studies. Section 4 examines the predictive potential of bare soil mosaics for estimating SOC in Chernozem regions and evaluates the gains obtained by integrating them with older soil maps in a Digital Soil Mapping framework. Section 5 focuses on the use of a seven-year Sentinel-2 time series to generate bare soil mosaics, evaluating different index thresholding strategies. It also analyzes the contribution of soil moisture indices and stratified models by soil type to improve the accuracy of SOC predictions. Finally, Section 6 discusses the main results, methodological contributions, and implications for COS monitoring at the regional level, with prospects for future research at the national level.