Temporal Analysis of Forest Cover Change owing to Environmental Impacts using Machine Learning

Abstract

Analyzing changes in the forest canopy is essential for sustainable forest man agement. Forests get affected by deforestation such as global warming, human activities, and a variety of natural calamities such as forest fires. Forest fires are growing increasingly dangerous as they threaten the ecosystem, economy, and hu man safety. The risk of forest fires rises during the dry seasons. So, it is crucial to examine the change in forest cover. Remote sensing technology, together with the environmental factors affecting forested areas, allows for a more thorough and precise examination of forests over time. This research aimed to determine the areas that are frequently affected by forest fires (hotspots) and also conducted a temporal analysis of those areas to examine forest cover change. Moreover, en vironmental factors affecting the forest cover change were investigated. In this study, the change analysis was conducted for a period of 6 years from 2017 to 2022 using Sentinel-2 imagery. An object-based approach, SNIC was adopted to cluster the whole forest region and then per-cluster change detection analysis was carried out. As a case study, Brunei Darussalam’s Forest cover area was analysed. Our critical analysis shows that in the clusters where fire events occurred, the area of those particular clusters was reduced due to the forest fire and lower level of precipitation. The change in the area of the clusters where forest fires occurred for the selected years was noted as: 66.11%, 69.46%, 68.32%, 73.88%, 77.27%, and 78.70%. This study will not only be beneficial for foresters to assess the causes of forest cover loss and develop strategies for forest conservation and afforestation as well as to determine the species in hotspot areas.