Thesis (MSc) -- Stellenbosch University, 2012. / ENGLISH ABSTRACT: One of the main objectives of the South African Wine of Origin Scheme (SAWOS) is to guarantee the quality of wine products emerging from South Africa’s viticultural production areas by preventing the abuse of names of products originating from outstanding viticultural areas. The study of terroirs contributes to knowledge that assists in delimiting potential viticultural areas in South Africa. Terroirs are areas with homogeneous groups of natural factors that, with the aid of effective management, have the potential to produce a unique product over a recognizable period. Natural terroir units (NTU) reflect the integration of relatively homogenous environmental factors, which include topography, climate, soil and geology. This study investigates the use of geographic information systems (GIS) and remote sensing methods in the identification of NTU in the Robertson wine district. Existing topographical, soil, geological and climatic GIS data layers were collected at various scales. In addition to this spatial data, orthorectified and radiometrically corrected SPOT 5 and ASTER satellite imagery were used to classify the land use/cover using an object-based image analysis (OBIA) approach. Eight land use/cover classes were distinguished by a sequential ruleset and an overall accuracy of 81.2% was achieved. The land use/cover layer was combined with the slope aspect and soil landscape to provide a three-component NTU description. One hundred and seventy NTU were identified, of which fifty five units exist for agriculture. These NTU can be used for site selection of agricultural produce and effective planning and management of land use. Climate was not included in the delimitation of NTU because the coarse resolution of climatic data could not be used to distinguish between different NTU. Therefore, all NTU identified in this research has similar climatic conditions. The major drawback of GIS-assisted terroir studies is the difficulty of representing a number of NTU on one GIS map. Therefore, it is recommended to associate the NTU map with a table of the classes instead. Furthermore, the accuracy, scale and resolution of available GIS data in South Africa influence the delimitation of NTU. Although remote sensing was found to provide efficient methods for land cover mapping, the use of multiseasonal satellite images would classify vineyards more efficiently because such an approach accounts for the different growth cycles of grapevines. / AFRIKAANSE OPSOMMING: Een van die hoof doelwitte van die Suid-Afrikaanse Wyn van Oorsprongskema (SAWOS) is om die kwaliteit van wynbouprodukte afkomstig van Suid-Afrikaanse wingerdbougebiede te verseker. Die studie van terroirs is geïdentifiseer as ‘n metode om moontlike wingerdbougebiede in Suid-Afrika af te baken. Terroirs is gebiede met relatief homogene natuurlike faktore wat oor ‘n erkenbare tydperk en met behulp van effektiewe bestuur die vermoë het om ‘n unieke produk te lewer. Dié natuurlike faktore word, “natuurlike terroir eenhede” (NTE) genoem en sluit topografie, klimaat, grond en geologie in. Hierdie navorsing ondersoek die gebruik van geografiese inligtingstelsels (GIS) en afstandswaarnemingstegnieke om NTE in die Robertson-wyndistrik te identifiseer. Bestaande ruimtelike topografiese-, grond-, geologiese- en klimaatdata is op ‘n verskeidenheid skale versamel. Bykomend tot hierdie ruimtelike data, is ortogekorrigeerde en radiometries-gekalibreerde SPOT 5 en ASTER satellietbeelde gebruik om landgebruik/ -bedekking te klassifiseer. Objekgerigte beeldanalise (OGBA) is toegepas tydens hierdie klassifikasie en agt landgebruik/ -bedekkingsklasse is onderskei deur gebruik te maak van ‘n stapsgewyse reëlstel. OGBA het ‘n algehele akkuraatheid van 81.2% gelewer. Die landgebruik/ -bedekkingsdata is gekombineer met hellingaspek en die grondlandskap om ‘n drieledige NTE-beskrywing te lewer. Een honderd en sewentig NTE is geïdentifiseer waarvan daar vyf-en-vyftig eenhede vir landbou bestaan. Hierdie NTE kan aangewend word vir die selektering van geskikte terreine vir landbou-gewasse en effektiewe grondgebruikbeplanning en -bestuur. As gevolg van die bestaande klimaatdata se growwe resolusie, was dit ongeskik om te onderskei tussen verskillende NTE. Dus heers daar soortgelyke klimaatstoestande vir elke NTE wat in hierdie navorsing geïdentifiseer is. Die grootste stremmende faktor wat GIS-verwante terroir navorsing beïnvloed, is die uitdaging om ‘n groot getal NTE op een kaart voor te stel. Daarom is dit beter om die NTE kaart met ‘n tabel te assosieer. Verder beïnvloed die akkuraatheid, skaal en resolusie van beskikbare digitale geografiese data in Suid-Afrika die afbakening van NTE. Alhoewel bevind is dat afstandwaarneming ‘n effektiewe metode is om landbedekking te karteer, sal die gebruik van meerseisoenale satellietbeelde wingerde meer doelmatig karteer omdat dit verskeie siklusse in die groei van wingerde in ag neem. SLEUTELWOORDE EN FRASES terroir, natuurlike terroir eenhede (NTE), geografiese inligtingstelsels, afstandwaarneming, wingerdkunde