Abstract: The research region covered by the secondary forest and it has quickly extended. Estimating either beneficial economic or environmental impacts needs accurate maps. Sentinel-2 Multispectral Instrument (MSI) with its distinctive synoptic coverage capability provides accurate and instantly valuable data. Three classification techniques (Artificial Neural Network (ANN), Support Vector Machine (SVM) and Spectral Angle Mapper (SAM) were investigated in this paper utilizing Sentinel-2 MSI image with training samples of various sizes to map secondary forest cover. SVM had the perfect output with overall accuracy varying from 86% to 92% and a coefficient of Kappa from 0.76 to 0.85, depends entirely on the sample size of the training data (varying from 20 to 500 pixels per class). SVM's benefit was more apparent once the sample size of the training was lower. ANN needed the involvement of the user, the level of his / her knowledge and experience affected the accuracy. The SAM algorithm surpassed alike SVM and ANN in aspects of speed and efficiency for large-scale secondary forest mapping. Furthermore, the maximum threshold values of the SAM spectral angle for a large training sample size are in agreement with the outcomes of previous studies, that implies the potential subjectivity of the SAM threshold. If verifiable, the SAM algorithm will be a simple and robust methodology commonly for large-scale mapping of secondary forest.
Keywords: Artificial Neural Network; Vector Machine Support; Spectral Angle Mapper; Sentinel-2 MSI.
Title: Secondary Forest Mapping by Utilizing Sentinel-2 MSI Imagery
Author: Jwan Al-Doski, Shattri B. Mansor, Zailani Khuzaimah
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Research Publish Journals
