Mapping Eucalyptus species using high-resolution multispectral imagery

dc.contributor.authorNyirahagenimana, Solange
dc.date.accessioned2020-09-10T11:49:08Z
dc.date.available2020-09-10T11:49:08Z
dc.date.issued2019
dc.descriptionThis research report is submitted in partial fulfilment of the requirements for the degree of Master of Science to the School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Johannesburg, 2019en_ZA
dc.description.abstractAlthough Eucalyptus species have several negative impacts on the environment, the management of these highly alien invasive species has been often complicated due to the socioeconomic benefits that they provide. Therefore, determining the extent and distribution of these notorious alien plant species can help to prioritize resource allocation for management intervention and to mitigate the impact of their invasion on native plant diversity and water resources. Mapping Eucalyptus species using traditional field survey is costly, timeconsuming and inconvenient for large-scale plantations. Remote sensing using low spatial and spectral resolution imagery is also complicated due to the spectral confusion between tree species. This study investigated the utility of multispectral SPOT 6 satellite imageries using Random Forest (RF) and Support Vector Machine (SVM) classification algorithms to discriminate and map Eucalyptus species in ‘Tom Jenkins Plantation’ at Rietondale, Pretoria East. The separability analysis performed to discriminate between Eucalyptus species using the Transformed divergence (TD) and Jeffries Matusita (JM) tests produced results ranging from 0 to 1, much lower than the benchmark values of 2 and 1.41for TD and JM, respectively. Nevertheless, at least one (the Random Forest) of the two classifiers produced a high level of overall accuracy (88.46% with a kappa coefficient of 0.87). The overall accuracy for the SVM classifier was only 55.26% (with a kappa coefficient of 0.50). The RF algorithm attained the user and producer’s accuracies ranging from 67% to 95% while SVM obtained 0% user’s accuracy for Eucalyptus classes. The study concludes that the SPOT 6 data combined with Random Forest are important in mapping invasive species and can help environmental managers to plan effective management of Eucalyptus species.en_ZA
dc.description.librarianTL (2020)en_ZA
dc.facultyFaculty of Scienceen_ZA
dc.format.extentOnline resource (47 leaves)
dc.identifier.citationNyirahagenimana, Solange (2019) Mapping Eucalyptus species using high-resolution multispectral imagery, University of the Witwatersrand, Johannesburg, <http://hdl.handle.net/10539/29578>
dc.identifier.urihttps://hdl.handle.net/10539/29578
dc.language.isoenen_ZA
dc.schoolSchool of Animal, Plant and Environmental Sciencesen_ZA
dc.subject.lcshTrees--Breeding
dc.subject.lcshEucalyptus grandis--South Africa
dc.subject.lcshForest conservation
dc.titleMapping Eucalyptus species using high-resolution multispectral imageryen_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
Mapping Eucalyptus species using high-resolution multispectral imagery_Full Report.pdf
Size:
1.45 MB
Format:
Adobe Portable Document Format
Description:
Main Work
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Collections