Evaluation of UAV-based 3D modeling methods for landslides under different route plans

Abstract

Currently, aerial photographs are acquired by unmanned aerial vehicles (UAVs) along preplanned flight routes, and a 3D model is constructed to investigate and measure the features of geological disasters. In this work, the effectiveness and quality of three kinds of UAV routes used for landslide investigations are explored on the basis of a survey of a landslide in Luquan County, Kunming city, Yunnan Province. The advantages and disadvantages of 3D models based on the "horizontal round‒trip route (HRr)", "terrain follow route (TFr)" and "inclined route (Ir)" methods are studied and compared through quality evaluations of the final products. The results show that the 3D models based on HRr and TFr have different degrees of distortion and blurring, whereas the 3D model based on Ir has a finer surface resolution. Ir offers the best performance among the three routes and can more accurately capture the true surface of the survey features. Correspondingly, the checkpoint accuracies of the HRr, TFr, and Ir methods are 0.155, 0.064 and 0.033 m, respectively, indicating that the 3D model based on Ir is the most accurate. This research indicates that inclined routes provide better applicability in the investigation and monitoring of landslides.