Repository logo

Doctoral Degrees (Agricultural Economics)

Permanent URI for this collectionhttps://hdl.handle.net/10413/6552

Browse

Browsing Doctoral Degrees (Agricultural Economics) by Subject "American bramble."

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    The integration of freely available medium resolution optical sensors with Synthetic Aperture Radar (SAR) imagery capabilities for American bramble (Rubus cuneifolius) invasion detection and mapping.
    (2018) Rajah, Perushan.; Mutanga, Onisimo.; Odindi, John Odhiambo.
    The emergence of American bramble (Rubus cuneifolius) across South Africa has caused severe ecological and economic damage. To date, most of the efforts to mitigate its effects have been largely unsuccessful due to its prolific growth and widespread distribution. Accurate and timeous detection and mapping of Bramble is therefore critical to the development of effective eradication management plans. Hence, this study sought to determine the potential of freely available, new generation medium spatial resolution satellite imagery for the detection and mapping of American Bramble infestations within the UNESCO world heritage site of the uKhahlamba Drakensberg Park (UDP). The first part of the thesis determined the potential of conventional freely available remote sensing imagery for the detection and mapping of Bramble. Utilizing the Support Vector Machine (SVM) learning algorithm, it was established that Bramble could be detected with limited users (45%) and reasonable producers (80%) accuracies. Much of the confusion occurred between the grassland land cover class and Bramble. The second part of the study focused on fusing the new age optical imagery and Synthetic Aperture Radar (SAR) imagery for Bramble detection and mapping. The synergistic potential of fused imagery was evaluated using multiclass SVM classification algorithm. Feature level image fusion of optical imagery and SAR resulted in an overall classification accuracy of 76%, with increased users and producers’ accuracies for Bramble. These positive results offered an opportunity to explore the polarization variables associated with SAR imagery for improved classification accuracies. The final section of the study dwelt on the use of Vegetation Indices (VIs) derived from new age satellite imagery, in concert with SAR to improve Bramble classification accuracies. Whereas improvement in classification accuracies were minimal, the potential of stand-alone VIs to detect and map Bramble (80%) was noteworthy. Lastly, dual-polarized SAR was fused with new age optical imagery to determine the synergistic potential of dual-polarized SAR to increase Bramble mapping accuracies. Results indicated a marked increase in overall Bramble classification accuracy (85%), suggesting improved potential of dual-polarized SAR and optical imagery in invasive species detection and mapping. Overall, this study provides sufficient evidence of the complimentary and synergistic potential of active and passive remote sensing imagery for invasive alien species detection and mapping. Results of this study are important for supporting contemporary decision making relating to invasive species management and eradication in order to safeguard ecological biodiversity and pristine status of nationally protected areas.