open access

https://hdl.handle.net/2115/100457

Rapid Debris Volume Estimation Using UAV-Based 3D Point Clouds : Validation Experiments and Application to the 2024 Noto Peninsula Earthquake

英語

2024 Noto Peninsula Earthquake

2026-07-01

Proceedings of the 10th International Symposium on Reliability Engineering and Risk Management (ISRERM 2026)

Rapid and accurate estimation of debris volume is crucial for effective disaster response and recovery planning. In recent years, unmanned aerial vehicles (UAVs) combined with photogrammetry-based 3D point cloud reconstruction have emerged as promising tools for large-scale damage assessment. However, the reliability of debris volume estimation is affected by multiple factors, including UAV flight conditions, point cloud generation methods, and parameter settings in data processing, which remain insufficiently validated. This study proposes a practical framework for rapid debris volume estimation by integrating controlled validation experiments and real-world application. First, a series of experiments using an artificial debris model composed of stacked cardboard boxes were conducted to quantitatively evaluate the influence of UAV flight altitude, flight path, modeling approaches, and ground estimation parameters on volume estimation accuracy. Second, the proposed methodology was applied to UAV data acquired in the Ukai district of Suzu City following the 2024 Noto Peninsula Earthquake to assess its applicability under complex field conditions. The results demonstrate that circular flight paths and tiled point cloud models with smoothing significantly improve estimation accuracy and stability. In real disaster scenarios, the presence of uneven terrain, dispersed debris, and boundary ambiguity introduces additional challenges that are not fully captured in controlled experiments. Based on these findings, practical guidelines for reliable debris volume estimation in post-disaster environments are discussed.

会議発表論文

Version of Record