Estimation of forest canopy height using orthoimage-refined digital elevation models

Tree height is an important factor when making estimates for the management of disasters, the atmospheric environment, and forest biomass. However, there are limits to the accurate estimation of tree heights due to the inherent deviations in Shuttle Radar Topography Mission (SRTM) digital terrain elevation data (DTED) and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) global digital elevation model (GDEM) data, which are used to capture surface elevations in wide areas. There are also other problems, such as errors introduced during automatic terrain extraction from satellite images and when matching treetop areas with tree heights. In this study, we propose a method for producing refined DEMs from stereo orthoimages, which are subtracted from existing digital terrain models to calculate tree heights over a wide area. Experimental results showed that our proposed method yields tree heights that are closer to those measured manually than tree heights estimated via SRTM DTED or ASTER GDEM are. Our method also facilitates the generation of polygons by updating the average tree height in existing vector interim terrain data, thereby improving their accuracy.