Individual-Tree Stem Volume Modeling Using Handheld LiDAR and UAV-SfM Photogrammetry in a Mediterranean Mixed Forest

Abstract

Integrating ground-based and aerial remote sensing for individual tree-level stem volume modeling remains underexplored in Mediterranean mixed forests, despite the growing need for cost-effective, automated forest inventory approaches. This study evaluated the combined use of Handheld Laser Scanning (HLS) and Unmanned Aerial Vehicle (UAV)-based Structure from Motion (SfM) photogrammetry for individual-tree stem volume modeling in a mixed stand in Castilla y León, Spain, dominated by Pinus halepensis, Pinus pinea, Quercus faginea, and Cupressus sempervirens. Two open-source HLS processing tools; the Forest Structural Complexity Tool (FSCT) and 3D Forest Inventory (3DFin), were compared for individual tree attribute extraction, with FSCT outperforming 3DFin across all species. Reference stem volumes were derived by applying species-specific Spanish National Forest Inventory (SNFI) allometric equations to FSCT-extracted diameter and height values. Random Forest models were then built using UAV-SfM crown metrics as predictors, testing two image overlap configurations: 80 × 80 F (80% front and side overlap) and 80 × 60 CF (80% front, 60% side, cross-flight). The 80 × 80 F configuration produced the best-performing model (R2 = 0.730), with 80 × 60 CF achieving comparable accuracy (R2 = 0.688), results confirmed by spatially independent leave-one-plot-out cross-validation (LOPO-CV R2 = 0.627 and 0.613, respectively). These results show that combining HLS and UAV-SfM through a predominantly open-source workflow offers a viable, reproducible approach to stem volume modeling in structurally complex Mediterranean mixed forests.

Plant phenotyping relevance

HLSとUAV-SfMを用いて個体樹の直径・樹高から幹材積を推定する再現可能な計測・解析ワークフローを構築し、複数ツールと飛行条件を比較検証しているため、植物表現型取得が中心である。

Code and data availability

The supplied article blocks describe HLS point clouds, UAV-SfM imagery, crown metrics, and Random Forest models, but contain no data or code availability statement, no public repository deposit, and no authors' URL for datasets, images, scripts, or trained models. Only third-party open-source tools (FSCT, 3DFin, WebODM

No evidence-backed public reproduction asset is currently recorded.