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

Accueil du site → Doctorat → Allemagne → 2018 → Landsat derived land surface phenology metrics for the characterization of natural vegetation in the Brazilian savanna

Humboldt-Universität zu Berlin (2018)

Landsat derived land surface phenology metrics for the characterization of natural vegetation in the Brazilian savanna

Schwieder, Marcel

Titre : Landsat derived land surface phenology metrics for the characterization of natural vegetation in the Brazilian savanna

Auteur : Schwieder, Marcel

Université de soutenance : Humboldt-Universität zu Berlin

Grade : Doctor rerum naturalium (Dr. rer. nat.) 2018

Résumé
The Brazilian savanna, known as the Cerrado, covers around 24% of Brazil. It is characterized by a unique biodiversity and a strong gradient in vegetation structure. Land-use changes have led to almost half of the Cerrado being converted into cultivated land. The mapping of ecological processes is, therefore, an important prerequisite for supporting nature conservation policies based on spatially explicit information and for deepening our understanding of ecosystem dynamics. New sensors, freely available data, and advances in data processing allow the analysis of large data sets and thus for the first time to capture seasonal vegetation dynamics over large extents with a high spatial detail. This thesis aimed to analyze the benefits of Landsat based land surface phenological (LSP) metrics, for the characterization of Cerrado vegetation, regarding its structural and phenological diversity, and to assess their relation to above ground carbon. The results revealed that LSP metrics enable to capture the seasonal dynamics of photosynthetically active vegetation and are beneficial for the mapping of vegetation physiognomies. However, the results also revealed limitations of hard classification approaches for mapping vegetation gradients in complex ecosystems. Based on similarities in LSP metrics, which were for the first time derived for the whole extent of the Cerrado, LSP archetypes were proposed, which revealed the spatial patterns of LSP diversity at a 30 m spatial resolution and offer potential to enhance current mapping concepts. Further, LSP metrics facilitated the spatially explicit quantification of AGC in three study areas in the central Cerrado and should thus be considered as a valuable variable for future carbon estimations. Overall, the insights highlight that Landsat based LSP metrics are beneficial for ecosystem monitoring approaches, which are crucial to design sustainable land management strategies that maintain key ecosystem functions and services

Présentation

Version intégrale (4,8 Mb)

Page publiée le 4 janvier 2019