Optical and radar remote sensing data for forest cover mapping in peninsular Malaysia

Abstract

This study aims to map forest cover in Peninsular Malaysia using satellite images as deforestation is of concern in the recent decades, and is an important environmental issue for the future too. The Carnegie Landsat Analysis System-Lite (CLASlite) program was used in this study to detect forest cover in Peninsular Malaysia using Landsat satellite data. The results of the study show that CLASlite algorithm misclassified some oil palm, rubber and urban areas as forest vegetation. A reliable forest cover map was produced by first combining Landsat and ALOS PALSAR images to identify oil palm, rubber and urban areas, and then subsequently removing them. The HH and HV polarization data of ALOS PALSAR (threshold method) could detect oil palm plantations with 85.26 per cent of overall accuracy. For urban area detection, Enhance Build up Index (EBBI) using spectral bands from Landsat provided higher overall accuracy of 94 per cent. These methods produced a forest cover reading of 5 914 421 ha with an overall classification accuracy of 94.5 per cent. The forest cover (including rubber areas) detected in this study is 0.38 per cent higher than the percentage of 2010 forest cover detected by the Forestry Department of Peninsular Malaysia. The technique described in this paper presents an alternative and viable approach for updating forest cover maps in Malaysia.