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

Accueil du site → Doctorat → Allemagne → Satellite-based monitoring of pasture degradation on the Tibetan Plateau : A multi-scale approach

Philipps-Universität Marburg (2015)

Satellite-based monitoring of pasture degradation on the Tibetan Plateau : A multi-scale approach

Lehnert, Lukas

Titre : Satellite-based monitoring of pasture degradation on the Tibetan Plateau : A multi-scale approach

Auteur : Lehnert, Lukas

Université de soutenance : Philipps-Universität Marburg

Grade : Doktorgrades der Naturwissenschaften (Dr. rer. nat.) 2015

Résumé partiel
The Tibetan Plateau has been entitled Third-Pole-Environment’’ because of its outstanding importance for the global climate and the hydrological system of East and Southeast Asia. Its climatological and hydrological influences are strongly affected by the local vegetation which is supposed to be subject to ongoing degradation. The degradation of the Tibetan pastures was investigated on the local scale by numerous studies. However, because methods and scales substantially differed among the previous studies, the overall pattern of degradation on the Tibetan Plateau is hitherto unknown. Consequently, the aims of this thesis are to monitor recent changes in the grassland degradation on the Tibetan Plateau and to detect the underlying driving forces of the observed changes. Therefore, a comprehensive remote sensing based approach is developed. The new approach consists of three parts and incorporates different spatial and temporal scales : (i) the development and testing of an indicator system for pasture degradation on the local scale, (ii) the development of a MODIS-based product usable for degradation monitoring from the local to the plateau scale, and (iii) the application of the new product to delineate recent changes in the degradation status of the pastures on the Tibetan Plateau. The first part of the new approach comprised the test of the suitability of a new two-indicator system and its transferability to spaceborne data. The indicators were land-cover fractions (e.g., green vegetation, bare soil) derived from linear spectral unmixing and chlorophyll content. The latter was incorporated as a proxy for nutrient and water availability. It was estimated combining hyperspectral vegetation indices as predictors in partial least squares regression. The indicator system was established and tested on the local scale using a transect design and textitin situ measured data. The promising results revealed clear spatial patterns attributed to degradation, indicating that the combination of vegetation cover and chlorophyll content is a suitable indicator system for the detection of pasture degradation on local scales on the Tibetan Plateau. To delineate patterns of degradation changes on the plateau scale, the green plant coverage of the Tibetan pastures was derived in the second part. Therefore, an upscaling approach was developed. It is based on satellite data from high spatial resolution sensors on the local scale (WorldView-type) via medium resolution data (Landsat) to low resolution data on the plateau scale (MODIS). The different spatial resolutions involved in the methodology were incorporated to enable the cross-validation of the estimations in the new product against field observations (over 600 plots across the entire Tibetan Plateau). Four methods (linear spectral unmixing, spectral angle mapper, partial least squares regression, and support vector machine regression) were tested on their predictive performance for the estimation of plant cover and the method with the highest accuracy (support vector machine regression) was applied to 14 years of MODIS data to generate a new vegetation coverage product. In the third part, the changes in vegetation cover between the years 2000 and 2013 and their driving forces were investigated by comparing the trends in the new vegetation coverage product against climate variables (precipitation from tropical rainfall measuring mission and 2 m air temperature from ERA-Interim reanalysis data) on the entire Tibetan Plateau


Version intégrale

Page publiée le 27 septembre 2017