Mapping the dominant forest tree distribution using a combined image classification approach in a complex Eastern Mediterranean basin

Abstract

A land use/cover (LUC) classification strategy based on an unsupervised k-means, object-based and expert knowledge classification technique was cross-checked using Landsat satellite datasets along with ancillary data for mapping dominant forest tree species of the Goksu River Basin in the Eastern Mediterranean Region. Eight dominant forest tree species were classified as juniper (Juniperus excelsa Bieb), Taurus fir (Abies cilicica Ant. & Kotschy), Turkish pine (Pinus brutia Tenore), black pine (Pinus nigra Arnold), cedar (Cedrus libani Rich), oak (Quercus pubescens Schwarz and Quercus cerris Pall), stone pine (Pinus pinea L.) and plane (Platanus orientalis L.). The results of the combined classification approach (CCA) were compared with a traditional maximum likelihood classifier (MLC) for a better understanding of the benefits of the CCA. Kappa values of the CCA and the MLC were derived as 0.77 and 0.44 respectively.