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

Accueil du site → Master → Afrique du Sud → A Remote sensing change detection study in the arid Richtersveld region of South Africa

University of the Western Cape (2007)

A Remote sensing change detection study in the arid Richtersveld region of South Africa

Main, Russell Stuart

Titre : A Remote sensing change detection study in the arid Richtersveld region of South Africa

Auteur : Main, Russell Stuart

Université de soutenance : University of the Western Cape

Grade : Magister Scientiae - MSc (Biodiversity and Conservation Biology) 2007

Résumé
The Richtersveld falls within the Succulent Karoo and Desert biomes. It forms part of a biodiversity hotspot with a high diversity of succulent plant species, many of which are endemic to the area. The area is also characterised by varied geological, socio-economic and climatic conditions, which are further complicated and accentuated by impacts such as open-cast mining, livestock grazing and the illegal harvesting of sensitive species. Existing research has pointed towards to the loss of keystone species, changes in vegetation cover and losses in overall biodiversity as a result of a combination of the above conditions and impacts. Remote sensing is increasingly being used to detect or monitor change, because of its ability to capture information on a large scale in a repeatable and digital manner. Remote sensing of vegetation cover changes in arid regions is, however, particularly challenging due to low vegetation cover, bright soil background, as well as morphological and physiological desert plant adaptations. This study made use of remote sensing technologies in order to investigate possible vegetation cover changes that have taken place over time, and which may have manifested through a combination of threats to the region. The aims of the study were addressed using three key questions that sought to a) gain an understanding of the relationship between vegetation response and moisture, in order to interpret b) temporal and c) spatial vegetation cover changes. A spatially and temporally representative remotely sensed dataset was used together with techniques that are repeatable and able to quantify change with limited human bias. The dataset consisted of periodic (1991, 1997 and 2004) 30 meter Landsat images as well as continuous (2000 to 2005) 16-day 250 meter MODIS NDVI imagery. The data were analysed using a combination of techniques that included pixel- and objectbased classifications, vegetation index differencing, principle component analysis, and others. The results give an objective reflection of a region that, from a rather course remote sensing perspective, appears to have a temporally predictable yet spatially complex vegetation relationship with available moisture. However, the region appears to be without significant temporal or spatial vegetation cover changes. Despite the stable results, it is argued that remote sensing research in the Richtersveld, using updated technologies and techniques, should continue into the future as it still holds great potential as a decision support tool in sensitive environments with high propensities for change.

Présentation

Version intégrale (11,64 Mb)

Page publiée le 28 novembre 2014, mise à jour le 8 janvier 2019