A fine-scale classification of land cover in the North-west Sandveld
| dc.contributor.advisor | Van der Merwe, J. H. | en_ZA |
| dc.contributor.author | Lotz, Tamarin | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Arts and Social Sciences. Dept. of Geography and Environmental Studies. | en_ZA |
| dc.date.accessioned | 2012-02-03T09:20:47Z | en_ZA |
| dc.date.accessioned | 2012-03-30T10:23:23Z | |
| dc.date.available | 2012-02-03T09:20:47Z | en_ZA |
| dc.date.available | 2012-03-30T10:23:23Z | |
| dc.date.issued | 2012-03 | en_ZA |
| dc.description | Thesis (MSc)--Stellenbosch University, 2012. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: A land cover classification showing the landscape structure of a given area is necessary to make appropriate measures for environmental planning. The environmental impacts from insensitive human activities have led to a severe loss of biodiversity in the Cape Floristic Region over time. The natural biodiversity of the North-West Sandveld in particular, has suffered severe loss and a high level of fragmentation. The rapid growth of certain agricultural practices in the Sandveld has led not only to loss of biodiversity, but the secondary effects of excessive water extraction, invasive alien fauna and flora species and harmful run-off from toxic herbicides and pesticides. To plan effectively, an accurate map of a suitable resolution needs to be created to effectively display spatial information. The primary aim was to demonstrate that a semi-automated fine-scale, land cover classification using object-oriented image analysis is possible for a large local area to examine the environmental issues pertaining to the Sandveld. Towards this aim, a model to classify land cover of the study area was developed and its effectiveness analysed and interpreted. To meet these requirements, pre-processed SPOT 5 satellite imagery was used to digitize certain classes and to generate frame, border and Normalized Differentiation Vegetation Index (NDVI) layers for the object-oriented classification in eCognition. The accuracy of the results was determined using the Kappa coefficient which gave an accuracy level of 70%. The environmental impacts were determined after area calculations were done on each class. The results showed that the natural areas still made up the greatest percentage of the Sandveld but that it is highly fragmented, especially along the coast and many areas, although left in a natural state, were being overgrazed by livestock. The temporary irrigated, temporary non-irrigated strip agriculture and permanent agriculture classes made up the most of the remainder of the areas and had the largest impact on the Sandveld‟s biodiversity. For the biodiversity of the Sandveld to remain intact, a balance between enforcing the law and encouraging inhabitants of the Sandveld to encourage more environmentally balanced practices needs to be created. Stewardship programmes and education will greatly enhance the effectiveness of any conservation efforts. | en_ZA |
| dc.description.abstract | AFRIKAANSE OPSOMMING:'n Landbedekkingsklassifikasie wat die landskapstruktuur van 'n gegewe gebied aandui, is noodsaaklik om gepaste omgewingsbeplanning toe te pas. Die effek van intensiewe menslike aktiwiteite op die omgewing oor 'n lang tydperk, het gelei tot die verlies van biodiversiteit in die Kaapse Blommestreek. Die natuurlike biodiversiteit van veral die Noordwes Sandveld is deur ernstige verliese, asook 'n vlak van fragmentasie beïnvloed. Behalwe dat die vinnige toename van sekere landboupraktyke in die Sandveld gelei het tot die verlies van biodiversiteit. Sekondêre faktore soos die oormatige water-ontrekking, indringerspesies van fauna en flora en die afloop van skadelike onkruiddoder en pes-weerende middels in rivierstelsels is ook verantwoordelik vir verlies van biodiversiteit. Om effektiewe beplanning moontlik te maak, is die skepping van 'n akkurate kaart met 'n geskikte resolusie wat die ruimtelike inligting effektief voorstel, nodig. Die hoof doel van hierdie studie was om 'n semi-geoutomatiseerde fynskaal landbedekkingsklassifikasie in 'n groot plaaslike gebied wat die omgewingskwessies rakende die Sandveld te bestudeer en demonstreer, deur gebruik te maak van objek-gerigte beeldanalise. Die effektiwiteit van 'n model wat ontwikkel is om die studiegebied te klassifiseer, was ontleed en geïnterpreteer. Om hierdie vereistes te bevredig, is voorverwerkte SPOT 5 satellietbeelde gebruik om sekere klasse te versyfer, asook om raam-, grens- en genormaliseerde plantegroei differensiasie indekslae vir beeldgerigte klassifikasie in eCognition te skep. Die akkuraatheid van die resultate was bepaal deur die Kappa-koëffisiënt wat 'n akkuraatheidsvlak van 70% gelewer het. Die omgewingsimpakte was bepaal deur opperlakteberekeninge vir elke klas te maak. Die resultate dui aan dat natuurlike gebiede steeds die grootste persentasie van die Sandveld beslaan, maar dat dit hoogs gefragmenteerd is, veral langs die kuslyn en dat baie gebiede, wat wel steeds in 'n natuurlike toestand is, oorbewei word deur vee. Die tydelike spilpuntbesproeide landbougebiede, tydelike nie-besproeide strooklandbou en permanente landbouklasse, beslaan die meeste van die oorblywende landbedekking van die Sandveld. Hierdie klasse het ook die grootste impak op die studiegebied se biodiversiteit. Om die biodiversiteit van die Sandveld te behou, moet 'n balans tussen die afdwing van wetgewing en die aanmoedig van meer omgewingsgebalanseerde praktyke in die Sandveld gevind word. Rentmeesterskap programme en opvoeding sal die effektiwiteit van enige bewaringspogings aansienlik verbeter. | af_ZA |
| dc.format.extent | 138 p. : ill., maps | |
| dc.identifier.uri | http://hdl.handle.net/10019.1/19907 | |
| dc.language.iso | en_ZA | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | |
| dc.subject | Object-oriented classification | en_ZA |
| dc.subject | Environmental planning -- South Africa -- Sandveld | en_ZA |
| dc.subject | Land cover classification -- South Africa -- Sandveld | en_ZA |
| dc.subject | Dissertations -- Geography and environmental studies | en_ZA |
| dc.subject | Theses -- Geography and environmental studies | en_ZA |
| dc.subject.other | Geography and Environmental Studies | en_ZA |
| dc.title | A fine-scale classification of land cover in the North-west Sandveld | en_ZA |
| dc.type | Thesis |