Estimating stand attributes of complex forest types in subtropical mountain environments by combining the shadow fraction method with analyses of high-dynamic-range photographs

Shadow fractions can be overestimated because of topographic shadows, which can occupy a significant area on aerial photographs of mountainous terrain. In this study, we first used high-dynamic-range (HDR) image analysis techniques to extract the original canopy shadow from the topographic shadows on aerial photographs. Subsequently, we applied the shadow fraction method to estimate selected forest attributes (stand height, basal area, and stem volume). In this paper, we discuss the effects of tree shadow fraction normalization, auxiliary spectral information, and forest type on forest attribute estimation. HDR image analysis successfully extracted canopy shadow information from topographic shadows. The tree shadow fraction normalization method had no obvious effect. The shadow fraction enhanced spectral information to estimate stand attributes. Using shadow fractions resulted in better estimates of stand height for mixed-hardwood forest ( = 0.45), basal area for mixed-hardwood forest ( = 0.50), and stem volume for conifer–hardwood forest ( = 0.43). This difference in estimated results is related to the shade patterns produced by stand structures in the different forest types.