Tsoka, Jonathan2018-10-172018-10-172018Tsoka, Jonathan, (2018) An assessment of physical drivers for farm dam distribution in Midlands Kwazulu Natal, using GIS and remote sensing, University of the Witwatersrand, Johannesburg, https://hdl.handle.net/10539/25829.https://hdl.handle.net/10539/25829This research proposal is done in partial fulfillment of the requirements of the Master of Science degree in GIS and Remote Sensing at Witwatersrand University, Johannesburg, South Africa March 2018.The interest in farm dams emanates mainly from their use for livestock watering, irrigation and fisheries enhancement on a sustainable basis. While management information on large dams in South Africa is largely available, it is lacking for farm dams which cumulatively store large volumes of water. As a result they are barely considered as part of the water resources of a river basin. Data acquisition methods for obtaining information about farm dams are costly, time consuming and labour intensive. This study was an attempt to map farm dams and establish the factors driving their spatial distribution pattern in the Midlands, KwaZulu-Natal, South Africa, using cost effective, time saving and less laborious GIS and remote sensing techniques. A classified April 2017 Landsat 8 satellite image was used to identify all water bodies in the Umgeni River basin U2 quaternary catchment (U2) while Google Earth was subsequently employed for differentiating farm dams from other water bodies. There were approximately 2000 water bodies that were identified by the classification. These included large national dams, pools in golf courses, ponds and disused mine dumps. A total of 864 farm dams in the U2 region quaternary catchment was observed. Six physical factors, namely slope, aspect, elevation, land use, soil type and geology were assessed to establish to what extent they influenced the siting of farm dams. The results indicated the importance of soils and land use as farm dams were mainly found in clusters in areas where agricultural farm land is also found since water is required for crop and livestock production. The influence of other factors such as slope, geology and elevation were observed in the spatial distribution maps. They all gave significant p-values in their univariate analysis. Of the six variables only aspect gave non-significant results while the rest were significant. A binary multivariate logistic regression was created for forecasting future farm dam sites and to establish which sites are poorly sited. The other four factors were fitted to the model except aspect and geology type which had no significant p-values in the model. The model had an Akaike information criterion (AIC) score of 293.42 and had the best combination of variables relative to other models. It was validated using 500 farm dam sites and it predicted 86% correctly.Online resource (79 leaves)enGeographic information systemsSpatial analysis (Statistics)Thesis