The application of a landscape diversity index using remote sensing and geographical information systems to identify degradation patterns in the Great Fish River Valley, Eastern Cape Province, South Africa

• Main Author: Tanser, Frank Courteney
• Format: Others
• Language: English
• Published: Rhodes University 1997
• Subjects: Geographic information systems; Earth sciences > Remote sensing; Environmental degradation > South Africa > Eastern Cape > Great Fish River Valley; Forest degradation > South Africa > Eastern Cape > Great Rish River Valley
• Online Access: http://hdl.handle.net/10962/d1005488

Using a range of satellite-derived indices I describe. monitor and predict vegetation conditions that exist in the Great Fish River Valley, Eastern Cape. The heterogeneous nature of the area necessitates that the mapping of vegetation classes be accomplished using a combination of a supervised approach, an unsupervised approach and the use of a Moving Standard Deviation Index (MSDI). Nine vegetation classes are identified and mapped at an accuracy of 84%. The vegetation classes are strongly related to land-use and the communal areas demonstrate a reduction in palatable species and a shift towards dominance by a single species. Nature reserves and commercial rangeland are by contrast dominated by good condition vegetation types. The Modified Soil Adjusted Vegetation Index (MSA VI) is used to map the vegetation production in the study area. The influence of soil reflectance is reduced using this index. The MSA VI proves to be a good predictor of vegetation condition in the higher rainfall areas but not in the more semi-arid regions. The MSA VI has a significant relationship to rainfall but no absolute relationship to biomass. However, a stratification approach (on the basis of vegetation type) reveals that the MSA VI exhibits relationships to biomass in vegetation types occurring in the higher rainfall areas and consisting of a large cover of shrubs. A technique based on an index which describes landscape spatial variability is presented to assist in the interpretation of landscape condition. The research outlines a method for degradation assessment which overcomes many of the problems associated with cost and repeatability. Indices that attempt to provide a correlation with net primary productivity, e.g. NDVI, do not consider changes in the quality of net primary productivity. Landscape variability represents a measure of ecosystem change in the landscape that underlies the degradation process. The hypothesis is that healthy/undisturbed/stable landscapes tend to be less variable and homogenous than their degraded heterogenous counterparts. The Moving Standard Deviation Index (MSDI) is calculated by performing a 3 x 3 moving standard deviation window across Landsat Thematic Mapper (TM) band 3. The result is a sensitive indicator of landscape condition which is not affected by moisture availability and vegetation type. The MSDI shows a significant negative relationship to NDVI confirming its relationship to condition. The cross-classification of MSDI with NDVI allows the identification of invasive woody weeds which exhibit strong photosynthetic signals and would therefore be categorised as good condition using NDVI. Other ecosystems are investigated to determine the relationship between NDVI and MSDI. Where increase in NDVI is disturbance-induced (such as the Kalahari Desert) the relationship is positive. Where high NDVI values are indicative of good condition rangeland (such as the Fish River Valley) the relationship is negative. The MSDI therefore always exhibits a significant positive relationship to degradation irrespective of the relationship of NDVI to condition in the ecosystem.