Application of Landsat 8 Image in An Assessment of Hydrothermal Alteration Mapping in Dense Vegetation: A Case Study from Kokap Area, Kulon Progo

Abstract

DOI:10.17014/ijog.9.1.45-60

Several researchers through geochemical analysis have proven the presence of gold mineralization in Kokap, Kulon Progo, as a result of hydrothermal alteration. Alteration mapping with optical remote sensing images in tropical areas is very difficult due to atmospheric conditions, dense vegetation cover, and rapid weathering. This study aims to assess the ability of Landsat 8 images in the mapping of hydrothermal alteration in Kokap, Kulon Progo, with the Principles Component Analysis (PCA) method. Three conventional machine learning methods, including artificial neural network (ANN), maximum likelihood classification (MLC), and support vector machine (SVM) were compared to find an optimal classifier for hydrothermal alteration mapping. The experiment revealed that the MLC method offered the highest overall accuracy. Two alteration zones were mapped, i.e. argillic zone and propylitic zone. The comparison results showed that the MLC classification of band ratio images of 5:2 and 6:7 yielded a classification accuracy of 56.4% and kappa coefficient of 0.36, which was higher than those of other machine learning methods and band combinations. The combination of Landsat 8 with DEM succeeded in increasing accuracy to 59.5% with kappa coefficient of 0.4.