Multi Temporal Remotely Sensed Image Modelling For Deforestation Monitoring

Abstract

Tropical rainforest in Indonesia faces critical issue related to deforestation. Human activities which convert forest cover into non-forest cover has been a major issue. In order to sustain the forest resources, monitoring on deforestation and forest cover prediction is necessary to be done. Remotely sensed data, Landsat images, with acquisition in 1996, 2000, and 2005 are used in this study. In this study area, forest cover decreased around 6 % in the period of 1996 - 2005. For the purpose of forest cover modelling, three model (i.e. Stochastic Markov Model, Cellullar Automata Markov (CA_Markov) Model, dan GEOMOD) were tested. Based upon the Kappa index, GEOMOD performed better with the highest Kappa index. Therefore, GEOMOD is recommended to forecast forest cover.