An evaluation of Multiple Endmember Spectral Mixture Analysis applied to Landsat 8 OLI images for mapping land cover in southern Africa's Savanna.

remote sensing;southern Africa;savannas;MESMA

Fernando Ferreira Caixeta

University:University of Louisville

Department:Geography and Geosciences

关键词: remote sensing;    southern Africa;    savannas;    MESMA;   

Others  :  https://ir.library.louisville.edu/cgi/viewcontent.cgi?article=3463&context=etd

美国|英语

来源: The Universite of Louisville's Institutional Repository

【 摘 要 】

Mapping land cover in southern Africa’s savannas using traditional pixel based remote sensing techniques can be very challenging due to the heterogeneity of its vegetation structure and the spectral difficulty in separating similar land covers across various land uses. In order to overcome these complications, a Multiple Endmember Spectral Mixture Analysis (MESMA) provides a potential remote sensing approach to quantify spectral variation in the physical environment at a subpixel level. The MESMA approach was applied in the study area of the Mayuni Conservancy, in Namibia. Results show that 32.3% of the study area is covered by photosynthetic vegetation (PV), 32.0% by non-photosynthetic vegetation (NPV), 25.2% by bare soil (B) and 10.6% by shade. Post-classification validation shows that MESMA presented a moderate performance in estimating the proportions of land cover types in the study area. However the validation process is limited to the available resources and carries great subjectivity. It is concluded that future research on the matter should include a more consistent investigation on the endmember selection methodology and expand the study area inside of the same ecosystem.