Monitoring of agricultural practices during the fallow period in Wallonia using Sentinel-2

Details

Supervisors

Defourny, Pierre ; Draye, Xavier

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

This thesis utilized Sentinel-2 remote sensing data to analyze the management of the fallow period within agricultural parcels in Wallonia. This fallow period, situated between one main crop's harvest and the germination of the subsequent crop, encompasses up to three phases: an initial bare soil phase, a cover crop phase, and a concluding bare soil phase. Features related to these phases, such as defining dates and the integral of GLAI for the cover crop phase, were extracted from processed time-series Vegetation Indices (VIs). Based on these features, parcels were categorized into three distinct agricultural practices through clustering. The developed model underwent evaluation and was compared with other models designed for monitoring agricultural practices and detecting bare soil (Sen4CaP L4C and L4E). In certain scenarios, the model aligned with these comparative models. The clustering analysis not only segmented parcels but also revealed a variable relationship between cover crop duration and biomass accumulation depending on the cluster. Furthermore, this analysis facilitated a comparison of different agricultural requirements, such as organic and conventional, and their influence on fallow management.