Comparaison de plusieurs méthodes de télédétection multibande et hyperspectrale pour la détermination du taux de recouvrement des résidus de cultures agricoles

Abstract

The vegetation cover fraction of agricultural soils is often used as input into erosion or hydrological models at the watershed scale. Many methods exist for quantifying green vegetation and most of them are based on vegetation indices that take advantage of the spectral contrast between the red and near infrared, that is typically present in chlorophyll producing plants. However, the detection and quantification of non-living vegetation, such as crop residues that are efficient in reducing erosion and runoff on agricultural soils, is often problematic because of its optical properties that make this type of vegetation difficult to discriminate from bare soil. A few methods were however developed but there is some ambiguity as to the validity of their results. The objective of this research is mainly oriented towards evaluating existing techniques for crop residue detection in order to find the most suitable one for accurate mapping of the residue cover fraction. The techniques evaluated include the Soil Adjusted Corn Residue Index (SACRI), the Corn Residue Index Multiband (CRIM), the Normalized Differential Index (NDI) and the spectral unmixing technique."--Résumé abrégé par UMI