The past centuries have been marked with massive land conversions from one land use category, usually natural vegetation, to another. The forces that drive these land use changes are complex and poorly understood. However, the study of land has been revolutionised by the introduction of spatial tools such as remote sensing and GIS that automate these complex issues and assist in the solutions of these geographic problems. Land use identification and classification techniques were used in conjunction with GIS to consistently and accurately extract and incorporate land use data from a series of remotely sensed images of Richards Bay Mineral's Zulti South Mineral lease. Eight land use types from Zulti South were identified and mapped from six different remotely sensed images taken at different time periods between the 21 st of September 1990 and the 1st of June 2001. This mapping technique was shown to have an accuracy of 87.6%. The data collated from this study enabled the monitoring and representation of the temporal and spatial differences in land use within a GIS. From the analysis carried in the GIS the land use dynamics within the lease could be quantified and modelled. The time series of the land use datasets indicated how much of the landscape is changing, what changes have occurred and where these changes are taking place. Accurate and timely mapping of land use provides vital information on the state of the mineral lease area and its environment, and facilitates the development of spatial trends from which predictions of land use and land use change can be made. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 2001.

The aim of this study was to investigate the role of spatial and categorical resolution of satellite images in landcover classification. Three images namely, SPOT 5, Landsat TM, and MODIS were used, each of varying spatial resolution. Landcover classes were chosen for each of the classifications, were placed into groups of 11, and then merged to 8. This was to evaluate the effect that the categorical resolution plays on the final classification algorithm. Three traditional classifiers were used to create landcover maps. It was found that the higher resolution imagery produced higher accuracies at the 11 class level and these accuracies were improved by reducing the number of classes to 8. The coarser resolution imagery was able to classify larger features more accurately than the smaller features. This allowed the conclusion to be drawn that, before classifications are to be done, the size of the features to be detected should be considered when deciding which imagery to use. To improve upon the accuracy of the maximum likelihood classifier, an Artificial Neural Network was trained using ancillary data and the SPOT 5 image. Results showed an increase of over 30% in the classification accuracy of the ANN. Specific classes were easily identified, showing the ability of the ANN to classify imagery from a complex savanna environment. Experiments with various parameters of the neural network confirmed that there are no general guidelines that can be applied to a neural network to obtain high classification accuracy. / Thesis (M.Sc.) - University of KwaZulu-Natal, Pietermaritzburg, 2007.

The Ferncliffe Catchment Conservancy has been identified, within the context of Pietermaritzburg, KwaZulu-Natal, for its important geographical and ecological features. The mapping and communication of these features to the broader community resident within the Conservancy have been envisaged as an important undertaking. A most effective way of achieving this goal was to use a Geographic Information System in the mapping exercise and in creating an inventory of the resources in the Conservancy and a monitoring database. Such spatial information would then provide stakeholders with a spatial context within which to appreciate the natural resources available and the problems associated with them. In undertaking this task, spatial data were acquired in digital form as well as from aerial photographs and 1:50000 topo-cadastral maps. These data were imported into ArcView GIS Version 3.1 where the mapping of the various resources was done. An inventory of the resources was created and a spatial database linking attributes that describe the physical environment, the natural vegetation, agricultural activities and the built environment, was set up. It became evident that using a Geographic Information System for natural resource management provides for integration of spatial information which would otherwise be contained in several separate databases and maps. Further, these data can be readily accessed, queried, upgraded and manipulated. For conservancies in urban and rural KwaZulu-Natal, and indeed, the rest of South Africa, to achieve their aims in natural resource management and monitoring, such an approach would be most efficient and effective. / Thesis (M.Env.Dev.)-University of Natal, Pietermaritzburg, 2000.

Wetland vegetation is of fundamental ecological importance and is used as one of the vital bio-indicators for early signs of physical or chemical degradation in wetland systems. Wetland vegetation is being threatened by expansion of extensive lowland areas of agriculture, natural resource exploitation, etc. These threats are increasing the demand for detailed information on vegetation status, up-to-date maps as well as accurate information for mitigation and adaptive management to preserve wetland vegetation. All these requirements are difficult to produce at species or community level, due to the fact that some parts of the wetlands are inaccessible. Remote sensing offers nondestructive and real time information for sustainable and effective management of wetland vegetation. The application of remote sensing in wetland mapping has been done extensively, but unfortunately the uses of narrowband hyperspectral data remain unexplored at an advanced level. The aim of this study is to explore the potential of hyperspectral remote sensing for wetland vegetation discrimination at species level. In particular, the study concentrates on enhancing or improving class separability among wetland vegetation species. Therefore, the study relies on the following two factors; a) the use of narrowband hyperspectral remote sensing, and b) the integration of vegetation properties and vegetation indices to improve accuracy. The potential of vegetation indices and red edge position were evaluated for vegetation species discrimination. Oneway ANOVA and Canonical variate analysis were used to statistically test if the species were significantly different and to discriminate among them. The canonical structure matrix revealed that hyperspectral data transforms can discriminate vegetation species with an overall accuracy around 87%. The addition of biomass and water content variables improved the accuracy to 95.5%. Overall, the study demonstrated that hyperspectral data and vegetation properties improve wetland vegetation separability at species level. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2010.