Forest Inventory with Terrestrial LiDAR: A Comparison of Static and Hand-Held Mobile Laser Scanning

TLDR

Static terrestrial laser scanning is increasingly effective for forest inventories, yet occlusion and tree irregularities still limit attribute extraction accuracy. The study aimed to determine whether a hand‑held mobile laser scanner could reduce occlusion and improve forest parameter estimation compared to single‑scan and multi‑scan TLS approaches. Researchers compared a hand‑held mobile laser scanner with single‑scan and multi‑scan TLS across diverse forest types, measuring ground, canopy, and DBH metrics. The mobile scanner achieved the most accurate DBH estimates (bias −0.08 cm, RMSE 1.11 cm) and 91 % tree coverage, while single‑scan TLS was competitive for ground extraction and multi‑scan TLS best for canopy, demonstrating that MLS enables efficient large‑scale 3D forest structure acquisition.

Abstract

The application of static terrestrial laser scanning (TLS) in forest inventories is becoming more effective. Nevertheless, the occlusion effect is still limiting the processing efficiency to extract forest attributes. The use of a mobile laser scanner (MLS) would reduce this occlusion. In this study, we assessed and compared a hand-held mobile laser scanner (HMLS) with two TLS approaches (single scan: SS, and multi scan: MS) for the estimation of several forest parameters in a wide range of forest types and structures. We found that SS is competitive to extract the ground surface of forest plots, while MS gives the best result to describe the upper part of the canopy. The whole cross-section at 1.3 m height is scanned for 91% of the trees (DBH > 10 cm) with the HMLS leading to the best results for DBH estimates (bias of −0.08 cm and RMSE of 1.11 cm), compared to no fully-scanned trees for SS and 42% fully-scanned trees for MS. Irregularities, such as bark roughness and non-circular cross-section may explain the negative bias encountered for all of the scanning approaches. The success of using MLS in forests will allow for 3D structure acquisition on a larger scale and in a time-efficient manner.