A comparative study of UAV-based 3D point cloud analyses on landslide volume estimation for progressive rockslide

Rock slope instability, such as rockslides and rock falls, are common issues along mountain highways. These natural phenomena not only control the slope morphology but also pose substantial risk to the safety of road users. When highway authority responds to those disasters, the first task is to estimate the landslide volume and potential sliding volume for planning emergency measures. Recent advances in UAV-based 3D point cloud analyses have improved our ability to investigate landslides efficiency with unprecedented time resolution. However, different techniques involve several limitations that should be considered when approaching landslide volume estimation for progressive rockslides. This study demonstrates how the effects of multitemporal point cloud dataset alignment may hinder the analysis of landslide development in high steep highway slope. A specific progressive rockslide occurred in Northern Cross-Island Highway of Taiwan is discussed. The landslide initiated on 14 September 2022 after Typhoon Muifa leave Taiwan. The first disaster caused the road to be blocked at the mileage of 49.8K for two days and developed continually for the next one month. We obtained pre- and post-disaster UAV-based point cloud data for three major disasters during the period. The DEMs of Difference (DoD) and Iterative Closest Point (ICP) approaches were used to minimize the positioning error and estimate the landslide volume for each event. In addition, the feasibility of another common approach multiscale model-to-model cloud comparison (M3C2) was also discussed. The study provides authorities and practitioners with qualitative comparison regarding the application of UAV-based 3D point cloud analyses on landslide volume estimation for progressive rockslides. The results also benefit scientists in developing scenario modeling based on numerical simulation.