Assessing the use of multispectra remote sensing in mapping the spatio-temporal variations of soil erosion in Sekhukhune District, South Africa

Thesis (M.Sc. (Geography)) --University of Limpopo, 2018

Soil erosion, which is a critical component of land degradation, is one of the serious global environmental problems often threatening food security, water resources, and biodiversity. A comprehensive assessment and analysis of remote sensing applications in the spatial soil erosion mapping and monitoring over time and space is therefore, important for providing effective management and rehabilitation approaches at local, national and regional scales. The overall aim of the study was to assess the use of multispectral remote sensing sensors in mapping and monitoring the spatio-temporal variations in levels of soil erosion in the former homelands of Sekhukhune district, South Africa. Firstly, the effectiveness of the new and freely available moderate-resolution multispectral remote sensing data (Landsat 8 Operation Land Imager: OLI and Sentinel-2 Multi-Spectral Instrument: MSI) derived spectral bands, vegetation indices, and a combination of spectral bands and vegetation indices in mapping the spatio-temporal variation of soil erosion in the former homelands of Sekhukhune District, South Africa is compared. The study further determines the most optimal individual sensor variables that can accurately map soil erosion. The results showed that the integration of spectral bands and spectral vegetation indices yielded high soil erosion overall classification accuracies for both sensors. Sentinel-2 data produced an OA of 83, 81% whereas Landsat 8 has an OA of 82.86%. The study further established that Sentinel-2 MSI bands located in the NIR (0.785-0.900 μm), red edge (0.698-0.785μm) and SWIR (1.565-2.280 μm) regions were the most optimal for discriminating degraded soils from other land cover types. For Landsat 8 OLI, only the SWIR (1.560-2.300 μm), NIR (0.845-0.885 μm) region were selected as the best regions. Of the eighteen spectral vegetation indices computed, Normalized Difference Vegetation Index (NDVI) and Soil Adjusted Vegetation Index (SAVI) and Global Environmental Monitoring Index (GEMI) were selected as the most suitable for detecting and mapping soil erosion. Secondly, the study assessed soil erosion in the former homelands of Sekhukhune, South Africa by applying a time-series analysis (2002 and 2017), to track changes of areas affected by varying degrees of erosion. Specifically, the study assessed and mapped changes of eroded areas (wet and dry season), using multi-date Landsat products 8 OLI and 7 Enhanced Thematic Mapper (ETM+)). Additionally, the study used extracted eroded areas and overlay analysis was performed together with geology, slope and the Topographic Wetness Index (TWI) of the area under study to assess whether and to what extent the observed erosional trends can be explained. ii Time series analysis indicated that the dry season of 2002, experienced 16.61 % (224733 ha) of erosion whereas in 2017 19.71% was observed. A similar trend was also observed in the wet season. This work also indicates that the dominant geology type Lebowa granite: and Rustenburg layered its lithology strata experienced more erosional disturbances than other geological types. Slopes between 2-5% (Nearly level) experienced more erosion and vice-versa. On the hand, the relationship between TWI and eroded areas showed that much erosion occurred between 3 and 6 TWI values in all the seasons for the two different years, however, the dry season of 2002 had a slightly higher relationship and vice-versa. We, therefore, recommend use and integration of freely and readily available new and free generation broadband sensors, such as Landsat data and environmental variables if soil erosion has to be well documented for purposes of effective soil rehabilitation and conservation. Keywords: Food security Global changes, Land degradation, Land-based ecosystems, Land management practices, Satellite data, Soil conservation, Sustainable Development; Topographic Wetness Index; Time series analysis.