PRIORITISED DEFORESTATION ALERTS USING A MULTI-SENSOR REMOTE SENSING APPROACH FOR THE MONITORING OF TROPICAL FORESTS IN SABAH REGION, MALAYSIA

摘要

The main driver of land use change and forest cover loss in Sabah over the past few decades has been the demand for palm oil and the increasing profitability of its cultivation. Although total forest cover in Sabah remains high, and rates of deforestation low by regional standards, significant conservation issues remain. The regional government of Sabah launched a jurisdictional approach in 2015. with the aim to prevent further loss of natural capital and achieve deforestation-free supply chains by 2025. We propose a methodology that can assist with achieving this goal: a near real-time deforestation monitoring system based on a combination of freely available Sentinel-1 radar, and Sentinel-2 and Landsat-5.7,8 optical imagery. The prioritised deforestation alerts for early warning coming from this system can be used to plan timely interventions on the ground. Current studies show that the automated system can detect deforestation with high accuracy over large areas, and the inclusion of Sentinel-1 radar provides insights weeks earlier than when relying on traditional Landsat observations only. The implementation of this new system can play an instrumental role in helping to demonstrate Sabah's palm oil can be 100% traceable, transparent and sustainable by 2025.