Remote sensing applications in vegetation mapping with special reference to the Langebaan area, South Africa
Master Thesis
1981
Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Department
Faculty
License
Series
Abstract
Interpretation of remote sensing products as a procedure used during the process of vegetation mapping has developed from a purely visual process of image identification to one which can utilize computerised methods to aid consistent identification of vast quantities of digitally stored/recorded spectral information. A description of the Landsat satellite system which is currently providing imagery to potential South African users in the form of digitally stored data and photographic products is given and sources of digital spectral data at other than satellite scales of resolution are described. A brief description of some image processing systems already operational in South Africa and being utilized for land cover mapping is also included. An introduction to the concept of computer analysis of numerical spectral data is given. The difference in approach between workers interpreting geological or other surface features as opposed to those wishing to simplify an image into categories is emphasized. This explains the local effort being expended on development of computer 'classification' routines as opposed to other methods of computer based image processing in vegetation mapping. The first paper presents a review of current use of remote sensing products in vegetation mapping in South Africa and the potential of more recently available products and processes in this field. The relative merits of different film types is discussed, as is the problem of scale of survey, scale of remote sensing product and scale of final mapping. The position that computer analysis of spectral data occupies in a scheme designed to show the relationships between different scales of survey is described. The second paper describes an example of the application of computer classification techniques to Landsat data in mapping vegetation in the Langebaan area, South Africa, at a semi-detailed scale of operation. The results of this exercise are illustrated together with a map produced using visual air photo interpretation techniques, backed by field checking. More detail of the specific relationships between plant community structure, canopy cover, scale of survey and reflectance values in the map classes produced is then given in the third paper. Computer classification matches well with major structural divisions; finer structural sub-division descriptions of sample plots correlate well with floristic divisions. A combination of digital analysis of remote sensing products ·and field checking based on structural schemes is recommended as a rapid mapping process. A report compiled on the full range of investigations carried out up to the end of 1980 into the usefulness of various remote sensing products for studying and mapping the Fynbos Biome is included. The overall mapping objective of this investigation was to determine the extent of the Fynbos Biome and of the major land use types within it. It was decided that the 'reconnaissance' level of operation at 1:250 000 final mapping scale is best suited to meet the overall mapping objective. General conclusions are drawn as to the current status of remote sensing applications in vegetation/land cover mapping in South Africa and elsewhere. Developments and refinements in techniques subsequent to the carrying out of the investigations reported in this account are briefly discussed.
Description
Includes bibliography.
Keywords
Reference:
Jarman, M. 1981. Remote sensing applications in vegetation mapping with special reference to the Langebaan area, South Africa. University of Cape Town.