Sentinel-2 images are expected to improve global crop monitoring even in challenging tropical small agricultural systems that are characterized by high intra- and inter-field spatial variability and where satellite observations are disturbed by the presence of clouds. To overcome these constraints, we analyzed and optimized the performance of a combined Random Forest (RF) classifier/object-based approach and applied it to multisource satellite data to produce land use maps of a smallholder agricultural zone in Madagascar at five different nomenclature levels. The RF classifier was first optimized by reducing the number of input variables. Experiments were then carried out to (i) test cropland masking prior to the classification of more detailed nomenclature levels, (ii) analyze the importance of each data source (a high spatial resolution (HSR) time series, a very high spatial resolution (VHSR) coverage and a digital elevation model (DEM)) and data type (spectral, textural or other), and (iii) quantify their contributions to classification accuracy levels. The results show that RF classifier optimization allowed for a reduction in the number of variables by 1.5- to 6-fold (depending on the classification level) and thus a reduction in the data processing time. Classification results were improved via the hierarchical approach at all classification levels, achieving an overall accuracy of 91.7% and 64.4% for the cropland and crop subclass levels, respectively. Spectral variables derived from an HSR time series were shown to be the most discriminating, with a better score for spectral indices over the reflectances. VHSR data were only found to be essential when implementing the segmentation of the area into objects and not for the spectral or textural features they can provide during classification.

The estimation of crop yield plays a major role in decision making and management of food supply. This paper aims to estimate corn dry masses and grain yield at field scale using an agrometeorological model. The SAFY-WB model (simple algorithm for yield model combined with a water balance) is controlled by green area index (GAI) derived from optical satellite images (GAIopt), and the GAI derived from synthetic aperture radar (SAR) satellite images (GAIsar) acquired over two crop seasons (2015 and 2016) in the south-west of France. Landsat-8 mission provides the optical data. SAR information (σ° _VV , σ° _{V H} , and σ° _VH,VV ) is provided by Sentinel-1A mission through two angular normalized orbits (30 and 132) allowing a repetitiveness from 12 to 6 days. σ° _VH,VV is successfully used to derive GAIsar (R ² = 0.72, relative root mean square error (rRMSE) = 10.4%) over the leaf development stages of the crop cycle from a nonlinear function. Others SAR signal (σ° _VV and σ° _VH ) are too much related to soil moisture changes. At the opposite of GAIopt, GAIsar cannot be used alone in the model to accurately estimate vegetation parameters. Finally, the robustness of the results comes from the combination of GAI derived from SAR and optical data. In this condition, the model is able, thanks to the inclusion of a new “production module,” to simulate dry masses and yield (R ² > 0.75 and rRMSE <; 12.75%) with good performances in the diagnostic approach. In the context of forecast, results offer lower performances but stay acceptable, with relative errors inferior to 13.95% (R ² > 0.69).

This work aims to provide daily estimates of the evolution of popcorn dry masses at the field scale using an agro-meteorological model, named the simple algorithm for yield model combined with a water balance model (SAFY-WB), controlled by the Green Area Index (GAI), derived from satellite images acquired in the microwave and optical domains. Synthetic aperture radar (SAR) satellite information (σ°_VH/VV) was provided by the Sentinel-1A (S1-A) mission through two orbits (30 and 132), with a repetitiveness of six days. The optical data were obtained from the Landsat-8 mission. SAR and optical data were acquired over one complete agricultural season, in 2016, over a test site located in the southwest of France. The results show that the total dry masses of corn can be estimated accurately (R² = 0.92) at daily time steps due to a combination of satellite and model data. The SAR data are more suitable for characterizing the first period of crop development (until the end of flowering), whereas the optical data can be used throughout the crop cycle. Moreover, the model offers good performances in plant (R² = 0.90) and ear (R² = 0.93) mass retrieval, irrespective of the phenological stage. The results also reveal that four phenological stages (four to five leaves, flowering, ripening, and harvest) can be accurately predicted by the proposed approach (R² = 0.98; root-mean-square error (RMSE) = seven days). Nevertheless, some important points must be taken into account before assimilation, namely the SAR signal must be corrected with respect to thermal noise before being assimilated, and the relationship estimated between the GAI and SAR signal must be performed over fields cultivated without intercrops. These results are unique in the literature and provide a new way to better monitor corn production over time.

The present study aims to investigate the potential of multi-configuration Sentinel-1 (S-1) synthetic aperture radar (SAR) images for characterizing four wheat parameters: total fresh mass (TFM), total dry mass (TDM), plant heights (He), and water content (WC). Because they are almost independent on the weather conditions, we have chosen to use only SAR. Samples of wheat parameters were collected over seven fields (three irrigated and four rainfed fields) in Southwestern France. We first analyzed the temporal behaviors of wheat parameters (He, TDM, TFM and WC) between February and June 2016. Then, the temporal profiles of the S-1 backscattering coefficients (VV, VH), the difference (VH − VV), the sum of the polarizations (VH + VV) and their cumulative values are analyzed for two orbits (30 and 132) during the wheat-growing season (from January to July 2016). After that, S-1 signals were statistically compared with all crop parameters considering the impact of pass orbit, irrigation and two vegetative periods in order to identify the best S-1 configuration for estimating crop parameters. Interesting S-1 backscattering behaviors were observed with the various wheat parameters after separating irrigation impacts and vegetative periods. For the orbit 30 (mean incidence angle of 33.6°); results show that the best S-1 configurations (with coefficient of determination (R²) > 0.7) were obtained using the VV and VH + VV as a function of the He, TDM and WC, over irrigated fields and during the second vegetative period. For the orbit 132 (mean incidence angle of 43.4°), the highest dynamic sensitivities (R² > 0.8) were observed for the VV and VH + VV configurations with He, TDM and TFM over irrigated fields during the first vegetative period. Overall, the sensitivity of S-1 data to wheat variables depended on the radar configuration (orbits and polarizations), the vegetative periods and was often better over irrigated fields in comparison with rainfed ones. Significant improvements of the determination coefficients were obtained when the cumulative (VH + VV) index was considered for He (R² > 0.9), TDM (R² > 0.9) and TFM (R² > 0.75) for irrigated fields, all along the crop cycle. The estimate of WC was more limited (R² > 0.6) and remained limited to the second period of the vegetation cycle (from flowering onwards). Whatever parameters were considered, the relative errors never exceeded 23%. This study has shown the importance of considering the agricultural practices (irrigation) and vegetative periods to effectively monitor some wheat parameters with S-1 data.

La thèse répond aux objectifs d'évaluer le potentiel d'estimation de la production de parcelles de maïs, ainsi que de leurs besoins hydriques. Les travaux, validés dans le Sud-Ouest de la France, reposent sur le couplage du modèle agro-météorologique à de l'imagerie satellitaire pour étendre le suivi à une échelle régionale. Le but est de s'affranchir des contraintes nuageuses imposées sur les acquisitions optiques, grâce à l'utilisation du radar Sentinel-1. L'estimation de la production comprend la biomasse sèche (épi, plante) ainsi que le rendement à l'échelle parcellaire. Au-delà des améliorations du modèle, l'assimilation combinée de l'optique et du radar donne de bons résultats et améliore les simulations basées sur l'optique en comblant l'absence de données en début de saison (forte nébulosité). Les résultats de l'estimation du contenu en eau sont plus contrastés. Le détail des erreurs révèle notamment des défauts liés aux propriétés du sol ou aux déclarations des agriculteurs.