Quantifying carbon stock and tree community composition in tropical forests through combining satellite and UAV analyses

Abstract

Abstract Monitoring tropical ecosystem services such as carbon stock and biodiversity with satellite remote sensing is essential for addressing climate change and biodiversity loss, but collecting ground truth data is costly. We investigated whether Unmanned Aerial Vehicles (UAVs) can reduce the costs. First, we developed a method to estimate Above-Ground Carbon (AGC) and biodiversity index (mixing ratio of pioneer and late-successional species) based on data derived from UAV-RGB images in four Forest Management Units (FMUs) in Malaysia. Second, we tested whether adding UAV-based ground truth (i.e., estimated carbon and biodiversity index) improves satellite-based models. We built machine learning models to estimate AGC and biodiversity index based on Landsat metrics and inventory data across Malaysia and Indonesia (395 plots). Accuracy was low without local inventory data (287 plots outside the four FMUs; R 2  = 0.43 and 0.46 for AGC and biodiversity, respectively). Adding UAV-based data ( n  = 934) significantly increased the accuracy ( R 2  = 0.51 and 0.48), which was comparable to the model with the full dataset including local inventory data ( R 2  = 0.53 and 0.60). These results underscore that integrating UAV and satellite analyses facilitates the monitoring of ecosystem services in tropical forests by reducing costs while maintaining accuracy.