Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Canada → La télédétection et les indices de végétation pour la détection de la végétation éparse et moyennement dense cas de l’environnement urbain

Université de Sherbrooke (1996)

La télédétection et les indices de végétation pour la détection de la végétation éparse et moyennement dense cas de l’environnement urbain

Bannari, Abderrazak

Titre : La télédétection et les indices de végétation pour la détection de la végétation éparse et moyennement dense cas de l’environnement urbain

Auteur : Bannari, Abderrazak

Université de soutenance : Université de Sherbrooke

Grade : Philosophe Doctor (Ph. D.) en Télédétection 1996

Résumé
The specific objectives of this thesis are twofold. First, our goal is to develop a vegetation index which characterizes sparse and moderately dense vegetation covers, independently from exterior physical disturbances namely : the effect of soil optical properties, i.e. color and brightness, related to the heterogeneity and specificities of this environment, the disturbances introduced by the atmosphere which are variable through time and space and, the effect of spatial and spectral resolutions specific to each sensor. These factors control the interaction processes between the electromagnetic radiation, the atmosphere, the vegetation cover and the underlying soil and, consequently introduce quite severe limitations for the detection of vegetation covers using vegetation indices. Secondly, we evaluate the contribution of the vegetation index to classification precision for thematic mapping applications. For this purpose, we carried out our analyses based on ground-based spectroradiometric data, narrow spatial (7 m) and spectral (30 nm) airborne dam (MEIS-II) and other wide spatial and spectral resolution satellite (TM) data. The study of the sensitivity of vegetation indices to atmospheric disturbances was carried out using the H5S radiative transfer model. As to the analysis of the contribution of the vegetation index to classification precision, we used the maximum likelihood algorithm, and verified the precision by means of the Kappa coefficient. In order to study the spectral properties of bare soils on vegetation covers, we propose a radiative transfer model which permits to decompose the resulting reflectance measured at ground level over a"soil-vegetation cover" mixture into two principal components : the first is intrinsect to the vegetation cover and the second, characteristic of the underlying bare soil, is transmitted through the vegetation cover. The results of the ground simulations for different rates of vegetation cover and different soil colors and brightnesses demonstrate the performance of the proposed model for enhancing the effect of soil optical properties on individual spectral reflectances and consequently, on vegetation indices. The analysis of the results based on the ground measurements, the airborne or satellite data and the simulations of the H5S atmospheric model show that the vegetation indices converge towards the same conclusions and demonstrate that none of the indices remains stable and independent in relation to overall exterior effects. However, the TSAVI and ARVI indices are distinct from the others by their complementary characteristics. Based on the individual performances of these two indices, we propose a new vegetation index : the TSARVI (Transformed Soil Atmospherically Resistant Vegetation Index). This new index has the advantage of adequately describing sparse ar moderately sparse vegetation independently from soil effects, the atmosphere and sensor characteristics

Présentation

Version intégrale (13,3 Mb)

Page publiée le 15 avril 2009, mise à jour le 8 février 2018