ESTIMATION OF STAND HEIGHT AND FOREST VOLUME USING HIGH RESOLUTION STEREO PHOTOGRAPHY AND FOREST TYPE MAP

Abstract

Abstract. Traditional field methods for measuring tree heights are often too costly and time consuming. An alternative remote sensing approach is to measure tree heights from digital stereo photographs which is more practical for forest managers and less expensive than LiDAR or synthetic aperture radar. This work proposes an estimation of stand height and forest volume(m3/ha) using normalized digital surface model (nDSM) from high resolution stereo photography (25cm resolution) and forest type map. The study area was located in Mt. Maehwa model forest in Hong Chun-Gun, South Korea. The forest type map has four attributes such as major species, age class, DBH class and crown density class by stand. Overlapping aerial photos were taken in September 2013 and digital surface model (DSM) was created by photogrammetric methods(aerial triangulation, digital image matching). Then, digital terrain model (DTM) was created by filtering DSM and subtracted DTM from DSM pixel by pixel, resulting in nDSM which represents object heights (buildings, trees, etc.). Two independent variables from nDSM were used to estimate forest stand volume: crown density (%) and stand height (m). First, crown density was calculated using canopy segmentation method considering live crown ratio. Next, stand height was produced by averaging individual tree heights in a stand using Esri’s ArcGIS and the USDA Forest Service’s FUSION software. Finally, stand volume was estimated and mapped using aerial photo stand volume equations by species which have two independent variables, crown density and stand height. South Korea has a historical imagery archive which can show forest change in 40 years of successful forest rehabilitation. For a future study, forest volume change map (1970s–present) will be produced using this stand volume estimation method and a historical imagery archive.