Hierarchical Integration of Multi-Source Lidar for Forest Volume Estimation across Individual Trees and Plot Scales

Accurate stem volume estimation remains contingent on an analytical tree (i.e., tree felled and sectionally measured to obtain reference stem volume) as the calibration benchmark for remote sensing and allometric equation, thereby limiting scalability and operational efficiency in large-scale forest carbon assessment. To overcome this bottleneck, we propose a hierarchical integration framework for multi-source lidar-based stem volume estimation, termed “hierarchical integration of volume estimation” (HIVE), which establishes a lidar-driven estimation chain linking individual trees to plot-level assessments. Specifically, we first establish a volumetric benchmark from terrestrial laser scanning (TLS) by reconstructing stems with the quantitative structure model (QSM), which serves as a surrogate for analytical trees. We then transfer this benchmark to airborne laser scanning (ALS) by mapping the lidar biomass index (LBI) and tree height to TLS-derived reference volume, effectively coupling nondestructive TLS-based volumetric calibration with broad spatial coverage. Cross-scale calibration and validation then enforce tree-plot consistency, enabling accurate upscaling without reliance on traditional field calibration based on the manual diameter at breast height and tree height measurements. For 41 Japanese larch (Larix kaempferi ) trees, TLS-based TreeQSM yielded stem volume estimates consistent with analytical-tree measurements (R 2 = 0.97, relative root mean square error [rRMSE] = 12.90%) and outperformed traditional allometric equations (R 2 = 0.96, rRMSE = 14.34%). Furthermore, using k-fold cross-validation on 81 TLS-ALS paired trees, the tree-level LBI model showed consistent predictive performance (R 2 = 0.81, rRMSE = 27.62%). When transferred to the plot level, stem volume estimates across 67 plots agreed well with field survey references (R 2 = 0.72, rRMSE = 23.53%). This approach provides an innovative solution for efficient largescale forest volume estimation by substantially reducing reliance on traditional field surveys for model calibration and opens new possibilities for carbon stock assessment and sustainable forest management.