Inter-Sensor Error Analysis between TLS and MLS for Urban Street Trees toward Digital Twin Development

Purpose: This study compares tree metric estimates derived from terrestrial laser scanning (TLS) and mobile laser scanning (MLS) for urban street-tree inventory, focusing on inter-sensor consistency under an identical processing workflow. Method: Point clouds collected along the same street segment near Bomun Station, Seoul, were acquired using Leica RTC360 (TLS) and Leica Pegasus Two: Ultimate (P2U, MLS). After exporting both datasets to LAS, we applied the same LiDAR360 workflow (ground normalization, individual tree segmentation, and attribute extraction) and matched 41 trees (N=41). We compared TreeHeight, DBH, CrownDiameter, CrownArea, and CrownVolume using mean difference (Δ=P2U−RTC360), MAE/RMSE, and paired nonparametric tests. Result: P2U tended to underestimate TreeHeight and crown-related metrics compared with RTC360 (Δ TreeHeight=−1.02m; ΔCrownVolume=−103.38m³ ), while DBH showed no significant difference (Wilcoxon p=0.6078). CrownVolume exhibited the largest discrepancy and variability (MAE=124.29m³ ; RMSE=183.40m³ ). Sensitivity analysis excluding outlier candidates confirmed that the overall pattern remained unchanged. Conclusion: Under the same commercial workflow, TLS and MLS results are not directly interchangeable for crown metrics without additional QC or calibration, whereas DBH appears relatively stable in this case study.