Evaluating experimental ITD algorithms for accurate hybrid volume estimation

Abstract

Accurate estimation of forest stand volume remains a major challenge in remote sensing–based forest inventory. This study assesses the performance and stability of several individual tree detection (ITD) algorithms applied to airborne LiDAR data in conjunction with forest inventory information from the Forest Management Plan (FMP). Algorithmic performance was compared across LiDAR point clouds of varying densities, and results were evaluated against both management plan projections and measured timber harvest records. Among the tested methods, the adaptive AutoLMF algorithm achieved the highest correspondence with reference data in terms of tree detection and stand-volume estimation. Simpler algorithms such as LMF and LMFx provided reliable estimates of average canopy height, particularly under low-density LiDAR conditions. A hybrid estimation framework combining LiDAR-derived structural metrics with management data improved the consistency and accuracy of stand-level volume estimates. The results demonstrate a transferable approach for enhancing inventory precision in structurally homogeneous forest stands and highlight the importance of algorithm calibration when applying ITD techniques to legacy or low-resolution LiDAR datasets.