Nova D. Doyog¹、Chinsu Lin^{2, *}

¹ Department of Forestry and Natural Resources。300 University Rd., Chiayi 60004, Taiwan。

² Department of Forestry and Natural Resources, National Chiayi University。300 University Rd., Chiayi 60004, Taiwan

^* Corresponding author: chinsu@mail.ncyu.edu.tw

Tropical forests are complex ecosystems providing many different ecosystem services and covers 45% (1.8 billion hectares) of the world’s forest cover (4.06 billion hectares). The world’s forest cover had faced a declining trend from 4.13 billion ha in 1990 to 4.06 billion ha in 2000 to 4.03 billion ha in 2005 to 4.02 billion ha in 2010, and eventually to 4.00 billion ha in 2015. This study was conducted to determine the drivers of forest degradation of tropical forest using optical sensing based approach. The forest degradation was determined through monitoring the AGB productivity of Pinus kesiya forest in the Philippines from 2014 to 2019 using Landsat OLI images and a machine learning based kNN algorithm with the aid of allometric model and inventory data. The surface reflectance of the Landsat images was restored using FLAASH model. Pansharpening technique was also applied to improve the spatial resolution while retaining the spectral resolution of the reflectance image. The result showed that pansharpened surface reflectance image was able to derive AGB maps with a satisfactory accuracy of root mean square percentage error, RMSPE = 15%. Human induced activities such as agriculture expansion, forest fires, and landslides were observed within the forest and played as the main drivers of forest degradation from 2014 to 2019 indicated by the AGB loss. Significant AGB losses were mainly observed in the edges of pine forest particularly of those adjacent to agriculture and built-up areas. Strengthening community-based forest management could mitigate the degradation of tropical forest and improving the biomass stocks for better REDD achievements.