Tree counting of tropical tree plantations using the maximum probability spectral features of high-resolution satellite images and drones

Information on tree plantation structures, such as tree type, density, and tree height, is essential for developing smart agriculture and plantation management strategies to support production estimation and investment, including biomass for carbon sequestration estimation. In this study, multisource remote sensing data from radar (Sentinel-1 C band), optical (Pléiades), and drones (multispectral drone) were used to support effective and cost-efficient sustainable tree plantation management in Siak Regency, Riau Province, Indonesia. Tree plantation maps were created using the difference backscatter VH and VV from Sentinel-1. Tree counting was then performed using Pléiades red, green, and blue visible bands and multispectral drone bands using a maximum a posteriori pixel-based classifier integrated with a filter function and statistical estimation. The validation of the tree map using manual measurements yielded accuracies ranging from approximately 79% to 97%. Tree heights were calculated from the difference between the Digital Surface Model (DSM) derived from drone data and the Digital Terrain Model (DTM) obtained from DEM Nasional (DEMNAS) data. Further improvements in the current map accuracy can be achieved using a combination of remote sensing and field measurements of tree structure inventories. • Tree plantations map, tree structure and tree height inventory produced using integrated SAR, optical imaging and UAV • Tree counting developed using The Maximum A Posteriori (MAP) classifier of high-resolution satellite image and drone • Tree height developed using drone DSM and DTM data