Centre de documentation
scientifique

Exploitation of hyperspectral data for assessing vegetation health under exposure to petroleum hydrocarbons / Guillaume Lassalle (2019)  

Public Titre : Exploitation of hyperspectral data for assessing vegetation health under exposure to petroleum hydrocarbons Type de document : Thèse/HDR Auteurs : Guillaume Lassalle, Auteur ; Arnaud Elger, Directeur de thèse ; Sophie Fabre, Directeur de thèse Editeur : Toulouse : Université Fédérale Toulouse Midi-Pyrénées Année de publication : 2019 Autre Editeur : Toulouse : Institut Supérieur de l’Aéronautique et de l’Espace Importance : 277 p. Format : 21 x 30 cm Note générale : bibliographie Thèse en vue de l'obtention du Doctorat de l'Université de Toulouse délivré par l'Institut Supérieur de l’Aéronautique et de l’Espace, spécialité : Surfaces et interfaces continentales, Hydrologie Agrosystèmes, écosystèmes et environnement Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications de télédétection [Termes IGN] canopée [Termes IGN] contamination [Termes IGN] feuille (végétation) [Termes IGN] hydrocarbure [Termes IGN] image aérienne [Termes IGN] image hyperspectrale [Termes IGN] indice de végétation [Termes IGN] modèle de transfert radiatif [Termes IGN] pollution des sols [Termes IGN] prospection pétrolière [Termes IGN] réflectance spectrale [Termes IGN] régression multiple [Termes IGN] signature spectrale [Termes IGN] surveillance de la végétation Index. décimale : THESE Thèses et HDR Résumé : (auteur) Oil exploration and contamination monitoring remain limited in regions covered by vegetation. Natural seepages and oil leakages due to facility failures are often masked by the foliage, making ineffective the current technologies used for detecting crude oil and petroleum products. However, the exposure of vegetation to oil affects its health and, consequently, its optical properties in the [400:2500] nm domain. This suggest being able to detect seepages and leakages indirectly, by analyzing vegetation health through its spectral reflectance. Based on this assumption, this thesis evaluates the potential of airborne hyperspectral imagery with high spatial resolution for detecting and quantifying oil contamination in vegetated regions. To achieve this, a three-step multiscale approach was adopted. The first step aimed at developing a method for detecting and characterizing the contamination under controlled conditions, by exploiting the optical properties of Rubus fruticosus L. The proposed method combines 14 vegetation indices in classification and allows detecting various oil contaminants accurately, from leaf to canopy scale. Its use under natural conditions was validated on a contaminated mud pit colonized by the same species. During the second step, a method for quantifying total petroleum hydrocarbons, based on inverting the PROSPECT model, was developed. The method exploits the pigment content of leaves, estimated from their spectral signature, for predicting the level of hydrocarbon contamination in soils accurately. The last step of the approach demonstrated the robustness of the two methods using airborne imagery. They proved performing for detecting and quantifying mud pit contamination. Another method of quantification, based on multiple regression, was proposed. At the end of this thesis, the three methods proposed were validated for use both on the field, at leaf and canopy scales, and on airborne hyperspectral images with high spatial resolution. Their performances depend however on the species, the season and the level of soil contamination. A similar approach was conducted under tropical conditions, allowing the development of a method for quantifying the contamination adapted to this context. In a perspective of operational use, an important effort is still required for extending the scope of the methods to other contexts and for anticipating their use on satellite- and drone-embedded hyperspectral sensors. Finally, the contribution of active remote sensing (radar and LiDAR) should be considered in further research, in order to overcome some of the limits specific to passive optical remote sensing. Note de contenu : General introduction 1- State-of-the-art of passive hyperspectral remote sensing for oil exploration and contamination monitoring in vegetated regions 2- Development of methods for detecting and quantifying oil contamination based on vegetation optical properties, under controlled conditions 3- Application and evaluation of the methods under natural conditions, from field scale to airborne hyperspectral imagery General conclusion Numéro de notice : 25946 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse française Note de thèse : Thèse de Doctorat : Spécialité : Surfaces et interfaces continentales, Hydrologie Agrosystèmes, écosystèmes et environnement : Toulouse : 2019 nature-HAL : Thèse DOI : sans En ligne : http://www.theses.fr/2019ESAE0030 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=96343