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

Accueil du site → Doctorat → Allemagne → 2008 → Time series generation and classification of MODIS data for land cover mapping

Julius-Maxilimiams-Universität Wûrzburg (2008)

Time series generation and classification of MODIS data for land cover mapping

Colditz, René Roland

Titre : Time series generation and classification of MODIS data for land cover mapping

Auteur : Colditz, Rene Roland

Université de soutenance : Julius-Maxilimiams-Universität Würzburg

Grade : Doktorgrade 2008

Each main chapter of this thesis contains a first section with an in-depth discussion, necessary to understand the subsequent datasets, methods, and results. The above-described goals of this thesis are approached in chapter 3 (Time Series Compilation) and chapter 4 (Automated Time Series Classification). The second chapter comprises a discussion of the sensors suitable for time-series generation and describes the MODIS sensor and the MODIS land product generation process. Besides important background information for readers unfamiliar with the sensors, in particular with the MODIS instrument, this chapter introduces the products and study sites used in this thesis. Chapter 3 (Time Series Compilation) highlights important considerations for time-series production and describes the process of MODIS time-series generation using the developed software tool TiSeG. The analysis section illustrates the selection of different data-quality levels and their effect on time series in Germany using visual and statistical measures. The effect of changes in the quality indicators is shown for the recently released MODIS data collection 5. A last section analyzes the temporal shifts of phenological patterns in South Africa. The fourth chapter, Automated Time Series Classification, discusses several issues to be considered for classification processes, outlines the classification procedure used in this study, and analyzes the results. The analysis section contains a detailed classification assessment of South Africa and Germany. A sensitivity analysis evaluates important classification parameters for both main study sites. A third part focuses on the capabilities for land-cover updating. A final part indicates the flexibility of the classification procedure for mapping vegetation types in Namibia. The thesis concludes with a final discussion and an outlook for critical requirements and scientific developments in the near future.


Version intégrale

Page publiée le 17 mars 2008, mise à jour le 13 mai 2019