SRGAN을 활용한 Sentinel-2 영상의 초해상화와 GEDI LiDAR를 통한 캐노피 추정

The increasing importance of understanding the changes in forest ecosystems due to climate change has led to consistent research in estimating AGB (Above-Ground Biomass), which typically involves surveying or the use of aerial and satellite imagery. These methods, however, face challenges related to cost, labor, and determining the size of research areas. For the reason, it is unavoidably relying on freely available satellite imagery data. Thus, this study aims to use SRGAN (Super-Resolution Generative Adversarial Network) for enhancing the resolution of Sentinel-2 images and to estimate forest canopy height using GEDI (Global Ecosystem Dynamics Investigation) LiDAR. Specifically, the super-resolution process employed SRGAN to enhance the 10m spatial resolution of Sentinel-2 images to 2.5m, improving the ability to detect changes in forests using Sentinel-2 images. Furthermore, canopy height values from GEDI data were interpolated for unmeasured areas using OK (Ordinary Kriging) and IDW (Inverse Distance Weighting), allowing for the estimation of forest canopy height over a large area. Finally, this study analyzed the average values and distribution of the forest canopy, and utilized both GEDI and Sentinel-2 data for a more precise understanding of the forest ecosystem. Therefore, this research proposes a cost-effective method for extensive forest ecosystem monitoring, contributing to sustainable forest management and conservation.