Assessing forest structure with LiDAR: A method benchmark in the Rohrach Natural Forest Reserve

Abstract

The Rohrach Natural Forest Reserve, located in the northern foothills of the Alps, serves as a long-term reference area for preserving and studying natural forest development. In this study, three methods for assessing forest structure—classical field-based inventory, terrestrial laser scanning (TLS), and drone-based LiDAR (UAV-LS)—were systematically compared. Building on an initial survey conducted in 1996, 44 sample plots were revisited and supplemented with high-resolution 3D measurements. The results show that TLS provides highly accurate volume estimates that closely match those obtained through the classical inventory. The UAV-based approach enabled a comprehensive, area-wide survey of the 48-ha study site and yielded average growing stock values of 547 m³/ha—almost identical to those from the classical inventory (549 m³/ha). However, automated detection of lying deadwood using UAV data underestimated the volume by up to 50% compared to the line intersect method. Whether this discrepancy is due to the line intersect method’s assumption of randomly distributed logs being unsuitable for this site, or whether UAV-LS underestimates deadwood due to canopy shadowing or algorithmic omission of fine structures, requires further investigation. The study also highlights persistent challenges in tree segmentation within steep or densely vegetated areas: overlapping crowns often result in misclassifications or undetected stems. While TLS continues to offer the highest geometric accuracy, UAV-LS provides the advantage of rapid, large-scale data acquisition with minimal disturbance to sensitive environments. These findings underscore the importance of integrated methodological approaches for effective long-term monitoring in natural forest ecosystems.