A rockfall is common geological disaster, and a natural phenomenon in which some dangerous rock masses on a slope roll down after being subjected to a force. It is usually characterized by unpredictability and high-speed motion, and can seriously threaten villages, communication lines, and traffic. Even if a small boulder rolls off a slope from a height of tens of meters, it can cause casualties(Chau et al. 1998) (Choi et al. 2009; Vanneschi et al. 2019; Youssef et al. 2015). The mechanisms of inducing rockfall disasters are usually diverse, and include freeze-thaw erosion, earthquakes, vegetation growth, and artificial excavation. These factors lead to differential weathering of rock masses, joint openings, pore pressure increases, and stress imbalances, which lead to rockfall disasters(Abebe et al. 2010; Chen et al. 1994; Dorren 2003; McCarroll et al. 1998; Valagussa et al. 2014; Wang et al. 2022; Wasowski and Del Gaudio 2000). Once a rockfall occurs, its trajectory and energy are controlled by multiple complex factors, including the size of the rock(Haas et al. 2012), the shape of the rock(Buzzi et al. 2012; Vijayakumar et al. 2011, 2012), and the surface properties of the slope (roughness, material properties, and covering)(Basharat et al. 2018; Wang and Lee 2010). The magnitudes and frequencies of different rockfall hazards vary in space and time, and the associated hazards often cannot be eliminated(Lan et al. 2010). Therefore, it is extremely important to study rockfall trajectories and energies through kinematic simulations to assess the risks of a rockfall disaster.
The methods and processes used for analyzing the spatiotemporal motions of rockfalls primarily include field investigations, theoretical analyses, and numerical simulations. First, through field investigations, the characteristics of dangerous rock areas, historical rockfall trajectories and slope terrain characteristics have been identified, and rock mass stability and failure modes have been analyzed. Then, the trajectories, kinetic energies, and threat ranges of rockfalls were obtained by two-or-three-dimensional (2D or 3D) numerical simulation methods(Bozzolo et al. 1988; Lan et al. 2007, 2010; Li and Lan 2015; Radtke et al. 2014; San et al. 2020; Sun et al. 2017; Wang et al. 2022; Wei et al. 2014); these results provide decision-making bases for subsequent disaster prevention and control (Choi et al. 2009; Fanos and Pradhan 2019; Vanneschi et al. 2019).However, describing rock mass characteristics using traditional field investigation methods becomes challenging when faced with difficult-to-approach slopes. In rockfall hazard assessments, the accuracy of the rockfall source location and the accuracy of the geographic data are two important factors that affect the reliability of the simulation results. The lack of geospatial slope data severely limits the application of numerical simulation methods(Lan et al. 2010).
Sturzenegger (Sturzenegger and Stead 2009) proposed a combination of field surveys and ground remote sensing technology to investigate and analyze rock mass characteristics in 2009, and there have been many applications of this method since then. However, for a rock mass on a high and steep slope, since the collection point of ground remote sensing data is at the bottom of the slope, when the location distribution of the rock mass is relatively high, the geometric information, spatial distribution characteristics, and location of the rockfall source cannot be accurately determined. High-resolution geographic data for slopes was also unavailable. The research and application of certain technologies, such as satellite remote sensing and interferometric synthetic aperture radar, have become relatively mature for use with geological disasters(Liu et al. 2022; Sun et al. 2016; Wang et al. 2021). However, these two methods are more suitable for early identification and monitoring of large-scale disasters, for small-area single geological disasters, these methods have the obvious limitations of redundant data collection, processing difficultly, and high cost (Lian et al. 2020).
With the rapid development of unmanned aerial vehicle (UAV) technology, near-Earth aerial remote sensing technology (with flying altitudes within 1,000m) overcomes the limitations of ground remote sensing. Oblique photogrammetry technology based on low-altitude remote UAV sensing is convenient and flexible, and is not affected by terrain during operation. This method can produce a 3D reality model that can reflect the real texture, color, and geometric shape of the modeling subject(Giordan et al. 2015; Ming et al. 2019; Vanneschi et al. 2019; Youssef et al. 2015). Additionally, high-resolution digital surface models (DSMs)(Mora et al. 2019) and digital orthophoto maps (DOMs)(Manfreda et al. 2018, 2019) can be acquired using this method. However, in Tibet and other remote alpine and valley areas, the GPS positioning accuracy of UAVs is low, and it is also affected by the terrain environment and image collection methods, making it difficult to guarantee the accuracy of image data collected in the field. When shooting images, due to the large vertical height difference of the slope, when collecting images by the conventional method of fixed flight height, the farther the area is from the camera position, the lower the image resolution. When building a 3D reality model, due to the different advantages and characteristics of different modeling methods, it is still difficult to obtain a 3D reality model that can satisfy both position accuracy and clear texture if only one method is used to process images. When investigating the characteristics of rock mass, the low resolution of the model makes it impossible to identify fine joints and fractures. Low positioning accuracy will lead to large errors in the interpretation of structural surface information, and even the statistics of the length and width of cracks may not match the actual ones. To complete the modeling work with high positioning accuracy and high resolution in complex environments requires a professional background in the field of surveying and mapping science, which is difficult for general geological researchers. Therefore, the comprehensive investigation and research of UAV technology in complex environment such as disaster characteristic investigation, formation mechanism analysis and risk assessment has not exerted its greatest advantage.
Chaya County is one of the important towns in the "Three Rivers Region" in southeastern Tibet. The expansion of the county will be an inevitable trend of future urban development with the implementation of the national rural revitalization strategy. At present, there are more than ten hidden dangers of large-scale geological disasters around the county seat, which seriously threaten residential houses and transportation facilities. However, the traditional survey method is difficult to complete the survey, because there are many unfavorable factors in this area, including the large number of disasters, the large distribution area, the high and steep terrain, the high distribution of provenance, and the existence of blind spots in sight. The analysis of disaster characteristics, mechanism, stability and disaster threat range can provide important reference value for the county's disaster prevention and future construction and development, and also can supply important basic data for the research on fault activity, landform formation and evolution in Changdu and its surrounding areas.
Therefore, taking a rockfall on the north side of Chaya County as the research object, this paper summarizes a set of rockfall disaster investigation procedures suitable for complex terrain combined with the UAV low-altitude remote sensing nap-of-the-object photogrammetry (Wang et al. 2022) technology and numerical simulation method. According to this process method, a 3D reality model, and a high-resolution DSM and DOM of the study area were obtained. The characteristics of rock masses in the study area and the formation mechanisms of rockfalls were analyzed. Multiple rockfall source points were accurately identified from the real 3D model, and the motion characteristics of potential rockfalls were simulated with the ArcGIS-based 3D numerical simulation software Rockfall Analyst(Lan et al. 2007, 2010; Zhang et al. 2019).