Detection of Tree Cover Dynamic on Belitung Island using Random Forest Regression

Belitung Island facesa series of interconnected environmental problems, particularly in forest conservation. Protected forest areas play a crucial role in supporting lifebut their sustainability is threatened by human activities such as mining exploitation and forest conversion for plantations. Therefore, protecting and restoring protected forest areas are priorities for maintaining the ecosystem's sustainability on Belitung Island. An evaluation was conducted to assess the ecological conditions of conservation areas on Belitung Island by visualizing changes in protected land cover to assist conservation efforts. In this study, the evaluation system for vegetation cover conditions on Belitung Island and Lalang MountainGrand Forest Park used random forest (RF) regression algorithms and remote sensing data. Satellite image data were used to determine the extent of vegetation cover on Belitung Island, utilizing combinations of bands from Landsat Satellites and MODIS Percent Tree Cover. Satellite images from 2013 to 2023 were used for comparison. This evaluation revealed several class changes in vegetation cover on Belitung Island based on percent tree cover classification over the years serving as an evaluation of land use in the areas under review. The R-squared value of 0.73indicated that the samples used to predict land cover demonstrated a relatively high level of accuracy. This study could serve as an effective means of predicting and estimating large-scale vegetation changes, as well as a monitoring tool for conservation areas on Belitung Island.