Effets de la couleur et de la brillance du sol sur les indices de végétation

Abstract

Variations in the spectral properties of soils related to their colour and brightness considerably influence the detection of sparse vegetation in heterogeneous environments using vegetation indices. During the last decade, a new generation of vegetation indices (NDVI, PVI, SAVI, MSAVI, TSAVI, TSARVI, ARVI, GEMI, and AVI) was developed in order to minimize these effects. To evaluate the sensitivity of these indices to soil colour and brightness and to test their potential for a more precise description of the vegetation cover for different cover rates, a number of simulations were carried out using a first order radiative transfer model. The model was adapted for studying directly the contribution of the optical properties of bare soils on the vegetation index. The results show that the first order radiative transfer model constitutes a valuable tool for analysing and understanding the interactions between the electromagnetic radiation, the vegetation cover and bare soil. It makes it possible to analyse the effect of colour and brightness on the reflectance factor and, consequently, on the vegetation index. The GEMI, AVI, NDVI, ARVI and PVI indices show lower performance for the management of sparse or moderately dense vegetation environments. They are marked by non-negligible errors related to the optical properties of bare soils. The AVI leads to results that do not reflect the theoretical behaviour of vegetation indices. As to the TSAVI, TSARVI, SAVI and MSAVI indices, they are more resistant to changes in the optical properties of soils and permit better discrimination between the vegetation from the bare soil background in an heterogeneous and relatively complex environment.