Quantifier l'impact biophysique du climat sur les changements dans le secteur agricole européen

Agricultural management practices influence climate through biophysical processes such as surface albedo and energy balance, yet their impacts remain understudied compared to biogeochemical effects. Three critical uncertainties in current assessments of agricultural biophysical impacts arise from: (1) The limited quantification of albedo-mediated radiative forcing induced by management practice; (2) The scarcity of high-resolution spatiotemporal data on bare soil albedo dynamics across heterogeneous croplands; and (3) The inadequate representation of crop development and management practices in land surface models. To address these gaps, this thesis integrates machine learning, remote sensing, and process-based modeling to evaluate the albedo-mediated biophysical climate impacts of agricultural practices in European croplands.First, I quantified the albedo-mediated biophysical impacts of seven common management practices on climate in European winter wheat systems at site scale using interpretable machine learning. By analyzing albedo data from eight flux-tower sites, I found that nitrogen fertilizer application has the highest positive albedo impact because of increasing crop cover. Herbicide use reduced albedo by suppressing fallow-period weeds. These findings highlight the biophysical climate impact of chemical inputs which are often overlooked.Second, I developed the first high-resolution, spatiotemporally continuous bare soil albedo dataset for European cropland regions (5-day intervals, 300 m resolution) using Sentinel-2 reflectance observations (2018-2020). Two random forest models disentangled spatial and temporal drivers, achieving R² values of 0.35-0.46 (RMSE=0.02). This dataset provides a critical tool for assessing albedo-mediated climate impacts of soil management, addressing the lack of granular albedo data for European fragmented croplands.Third, I advanced the ORCHIDEE-CROP land surface model to represent the effect of residue cover and foliar yellowing of winter wheat on albedo, surface roughness, and soil evaporation. Simulations across 12 European sites showed that foliar yellowing during senescence and post-harvest residue retention cools the surface by raising the surface albedo. Additionally, wheat residues can improve soil water storage, demonstrating their dual role in climate adaptation and mitigation.This thesis demonstrates that agricultural practices such as residue management significantly influence climate through spatiotemporal variation of albedo and energy balance. The improved process-based modeling provides effective insights and evaluation tools for region-specific agricultural mitigation strategies. Linking biophysical and biogeochemical effects highlights the imperative for multidimensional evaluations in climate policy and agricultural management. Future research could prioritize developing hybrid frameworks that combine high-resolution satellite, machine learning, and multiscale process-based models, enabling the design of climate-resilient agricultural systems in a warming world.