基于多期无人机影像的滑坡地表竖向变形测量模型试验与工程应用

滑坡地质灾害严重威胁人民生命财产安全，滑坡地表竖向变形测量属于滑坡监测与预警的重要组成部分。近年来，国内外学者开始尝试使用多期无人机影像开展滑坡地表变形监测，然而基于多期无人机影像的滑坡地表竖向变形测量精度研究却相对较少。首先通过大量室外模型试验，对CloudCompare、Global Mapper和PolyWorks三款软件地表竖向变形识别的结果进行了对比和研究；在此基础上，定量分析了三款软件地表竖向变形测量的精度。研究结果表明:当无人机影像分辨率优于3.0cm/像素，三款软件均能识别5.0cm及以上的地表竖向变形；在不同地表竖向变形工况下，Global Mapper地表竖向变形测量结果最为精确与稳定，竖向变形测量的中误差总体上分布于1.5~4.0 cm之间。三款软件地表竖向变形测量的中误差和测量误差的均值及标准差均与地表竖向变形值不呈现明显相关性，同时地表竖向变形量测误差均近似满足正态分布，因此可选取测量误差的95%置信区间对地表竖向变形测量结果进行修正。在此基础之上，运用Global Mapper软件开展了黑方台党川段滑坡地表竖向变形识别与测量应用，结果表明Global Mapper软件能较为准确地识别出滑坡变形区域并圈定其位置。 

Model tests of the vertical ground deformation measurement of landslide based on multiple UAV images and its application

The safety of people and property in mountainous regions can be affected by landslides, and measurement of vertical ground formation is a vital component of landslide monitoring and forecasting. Over the past few years, UAV images have gained increasing popularity in the landslide monitoring; however, studies on the measurement accuracy of UAV image-based landslide surface deformation monitoring are relatively limited. In this study, a series of outdoor model tests are conducted to investigate the measurement accuracy of the digital difference-based software of CloudCompare, Global Mapper, and PolyWorks in UAV image-based vertical ground deformation measurement. The results indicate that when the resolution of the input UAV image is better than 3.0 cm/pixel, all the selected software can recognize the ground deformation zones with vertical ground deformation larger than 5.0 cm; under different vertical ground deformation scenarios, the vertical ground deformation measurement obtained with Global Mapper tends to be more accurate, and the medium error of the measurement ranges from 1.5 cm to 4.0 cm. Neither the medium error of measurement nor the standard deviation of measurement error is greatly affected by the magnitude of vertical ground deformation. The statistical analysis of measurement error suggests that the measurement errors of the selected software all follow normal distributions; then, based on the statistics of the measurement errors, the 95% confidence intervals of the vertical ground deformations obtained from the selected software could be obtained. Furthermore, Global Mapper software is applied to identify and measure the vertical deformation of the Dangchuan landslides in Heifangtai, and the results indicate that this software can accurately identify the landslide deformation area and locate the related ranges.