Exact characteristics of surface rupture zone are essential for exploring the mechanism of large earthquakes. Although the traditional field surface rupture investigation methods can obtain high-precision geomorphic data in a local area, it is difficult to rapidly get an extensive range of high-precision topographic and geomorphic data of the entire fault due to its limited measurement range and low efficiency. In addition, manual measurement is of tremendous workload, high cost, time-consuming and laborious, and the subjective differences in the judgment standards during the manual operation process may also cause the measurement results to be inconsistent with the actual terrain characteristics. In recent years, the development of photogrammetry technology has provided another more effective technical means for the rapid acquisition of high-precision topographic and geomorphic data, which has dramatically changed the way of geological investigation, improved the efficiency of fieldwork. At the same time, it also makes it a reality to reproduce the 3D tectonic features of field tectonic deformation indoors.
Structure from Motion(SfM)multi-view mobile photogrammetry technology is widely concerned for its convenience, fast and low-cost acquisition of high-resolution 3D topographic data in a working area of tens-kilometers scale. The emergence of this method has greatly improved the automation degree of photogrammetry. The technology obtains image sets by motion cameras, uses a feature matching algorithm to extract homonym features from multiple images(at least three images), determines the relative positional relationship of cameras during photography, and continuously optimizes by the nonlinear least square algorithm. Finally, the pose of cameras is automatically solved, and 3D scene structure is reconstructed. The technology can restore the original 3D appearance of the object in the computer by a set of digital images with a certain degree of overlap. In the applications of terrain mapping, this technology only needs to combine a small number of ground control points(GCPs)to quickly establish digital orthophoto maps(DOMs)and digital elevation models(DEMs)with high-precision. In this way, low altitude remote sensing platforms such as small and medium-sized UAVs have provided a foundation for SfM photogrammetry technology.
After the Madoi M_W7.4 earthquake occurred on May 22, 2021, our research team rushed to the site as soon as possible and conducted the rapid photogrammetry of the entire coseismic surface rupture zone in a short period with the use of the CW-15 VTOL fixed-wing UAV. We completed the collection of topographic data in an area with ~180km length and ~256km² area and collected 34302 aerial photographs. We used Agisoft PhotoScan TM software to process the images and generate DOMs quickly. The DOM resolution of the entire surface rupture was 2~7cm/pix, most of which were 3~5cm/pix. Then we used GIS software to vectorize the surface rupture. The centimeter-scale high-resolution DOMs could clearly display the coseismic surface rupture’s spatial distribution and the relative width. On this basis, the surface rupture could be accurately interpreted, and related parameters such as coseismic offsets could be extracted. In this study, the horizontal offsets measured by orthophoto images were basically consistent with the field measurement results, which proved the authenticity and reliability of the data obtained by the UAV photogrammetry method.
In order to obtain more detailed surface rupture vertical offset data, we used DJI Phantom 4 Pro V2.0 UAV to collect terrain information of several areas with the most significant rupture deformation. The DEM resolution obtained could reach centimeter-scale, and the accuracy was greatly improved. The high-resolution topographic and geomorphic data obtained by this method could accurately identify tiny fault features, clearly display sub-meter-level vertical offset features, significantly improve the accuracy of offset measurement, and achieve high-resolution 3D reconstruction of fault geomorphic.
In addition, we selected typical surface ruptures in the field, such as compressional stepovers, tensional cracks, and pressure ridges, and collected their 3D structural features using the iPhone 12 Pro LiDAR scanner. The 3D Scanner application was used to optimize the image, completely restore the “real object” in 3D to realize the indoor reconstruction of the 3D structure of surface ruptures and pressure ridges. The augmented reality(AR)imaging models could truly reflect the characteristics and details of surface ruptures, forming the same effect as field observations. This technology, which creates 3D models of close-range environments without any prior preparation, provides a novel, economical, and time-saving method to rapidly scan morphological features of small and medium-sized landforms(from centimeters to hundreds of meters)at high spatial resolution. This is the fastest and most convenient way to collect 3D models in field geological investigation without using external equipment, which provides a new idea for future geological teaching and scientific research.
Although photogrammetry technology still has some limitations, such as the short flight time of the flight platform, being easily affected by factors such as weather and altitude, and unsatisfactory aerial photography in densely vegetated areas, it is believed that these problems will be solved with the advancement of technology. Once solved, photogrammetry will become an essential technical means in quantitative and refined research on active tectonics.

Based on the broadband records of the digital seismic networks of Qinghai, the focal mechanisms of the Dingqing, Xizang earthquakes(M_S≥3.0) are of the obtained with Cut-and-Paste(CAP)inversion method and from USGS, seven of them are normal fault type with a little strike-slip component. The dominant direction of the fault strikes is near SN, the dominant distribution of dip angles is 58°~69°, and the dominant distribution of rake angle is -81°~-103°. The dominant direction of P axis is SWW, and that of T axis is SEE. The best double couple solution of the M_S5.5 earthquake in 2016 is 12°, 58° and -103° for strike, dip and rake angles, respectively, the second nodal plane solution is 216°, 34° and -70°, the centroid depth is 7.3km, and its moment magnitude is 5.3. For the M_S5.1 earthquake in 2020, the solution is 9°, 57°, -101° for strike, dip and rake angles, respectively, the second nodal plane solution is 209°, 35° and -74°, the centroid depth is 6.8km, and its moment magnitude is 4.9.
The double difference relative positioning method(HypoDD)is used to relocate the Dingqing earthquakes from February 1, 2015 to March 5, 2020. Broadband data of 9 seismic stations of Qinghai seismic network, Tibet seismic network and scientific array within about 400km around the epicenter are used, and the relocation of 217 earthquakes is obtained. After relocation, the Dingqing earthquake sequence is more clustered than before, with zonal distribution along NE-SW direction, which is in agreement with the fault strike of focal mechanism solutions, but not consistent with the major strike-slip faults in the region. The focal depths of the Dingqing earthquakes are close to the normal distribution, 75 percents of them range from 8 to 12km. The focal depths of earthquakes in 2015-2018 are confined in the range of 5~15km, and that in 2018—2020 are mainly from 7km to 12km, the range of focal depths is significantly reduced after 2018. After the occurrence of M_S5.5 earthquake in 2016, the earthquakes ruptured rapidly to the west and south, and most of the aftershocks were of magnitude 3 or below, and the sequence attenuation was fast, which may be because that the mainshock released most of the energy in the sequence. The aftershocks of the M_S5.1 earthquake in 2020 mostly ruptured along the horizontal direction or to the deep. The earthquakes occurring from 2019 to March 2020 are located in the middle of the sequence in spatial distribution, and there are two dominant directions of NE-SW and SSE in the spatial distribution of epicenters, showing an L-shape distribution. The reason may be that the earthquake encountered obstacles in the rupture along the NE-SW direction, the strain energy was not fully released, and then turned to the SSE faults after stress adjustment to induce subsequent aftershocks. In the NE direction of the “L-shape”, in addition to the M_S5.1 earthquake on January 25, 2020, there were also earthquakes with M_S5.5 on May 11, 2016 and M_S4.5 on October 12, 2017, while only a few earthquakes with M_S3.4 and below occurred in SSE direction, indicating that the NE-trending faults are the dominant area of Dingqing earthquakes activity in recent years.
Since the focal mechanism solutions of M_S5.5 earthquake in 2016 and M_S5.1 earthquake in 2020 are both of normal fault type, the dominant distribution direction of aftershocks is NE, according to the analysis of relocation, focal mechanism and geological structure background, it is inferred that the seismogenic structure of M_S5.5 earthquakes in 2016 and M_S5.1 in 2020 may be of a same normal fault type with NE direction. The fault plane may be nodal plane Ⅰ, i.e. the nodal plane with strike of 12°, dip angle of 58°, rate angle of -103° and strike of 9°, dip angle of 57° and rate angle of -101°. Because the Dingqing earthquakes occurred in the hinterland of the Qinghai-Tibet Plateau, the related research data on the distribution and attitude of small-scale faults is very scarce, so it is difficult to determine the seismogenic faults of the Dingqing earthquakes.