Review on the Application of Airborne LiDAR in Active Tectonics of China: Dushanzi Reverse Fault in the Northern Tian Shan

High-resolution topographic data are fundamental for active tectonics studies. Within the last 2 decades, the airborne light detection and ranging (LiDAR) system provided a solution for the accurate and ef ﬁ cient acquisition of detailed geomorphic features. The use of LiDAR data for the identi ﬁ cation of microstructural and geomorphic features, fault zone activity analysis, and earthquake disaster assessment remains challenging and has been the focus of active tectonics studies. Based on the LiDAR data of Dushanzi anticline – a reverse fault zone in Xinjiang, China, our group carried out a signi ﬁ cant number of active tectonic research studies. By reviewing the speci ﬁ c content of these works, we summarized the main application of LiDAR for a speci ﬁ c structure, the Dushanzi Reverse Fault in the northern Tianshan. In addition, other applications of LiDAR in active tectonics are summarized in this paper. These studies show that high-resolution LiDAR data facilitate detailed studies of the fault activity and paleoseismicity. We hope more researchers can realize the advantages of LiDAR technology in active tectonics research and apply LiDAR technology into their own practical work, so as to promote the development of active tectonics research.


INTRODUCTION
A moderate to large magnitude (M > 5.5) earthquake may break the ground surface and form various types of earthquake-related landforms, such as fault scarps, offset terraces, gullies, and alluvial fans, and other geomorphic markers (Hilley and Arrowsmith, 2008;Zhou et al., 2020a).These earthquakerelated landforms provide valuable information about earthquakes that have occurred over hundreds to tens of thousands of years.Therefore, studies of earthquake-related landforms play a crucial role in gathering this type of information including the earthquake magnitude, recurrence time, and coseismic slip distribution (Deng and Wen, 2008;Arrowsmith and Zielke, 2009;Zhang et al., 2013;Bi et al., 2016), especially considering the increasing availability of high-resolution topographic and imagery data for tectono-geomorphic interpretation (Shi et al., 2017;Brandenberg et al., 2020).
Among high-resolution topographic data, light detection and ranging (LiDAR)-derived digital elevation models (DEMs) represent one type of high-resolution basemaps that enable the detailed mapping and surveying of the fault geometry and displaced geomorphic markers.Because LiDAR can penetrate the surface vegetation, LiDAR-derived DEMs reveal the high-resolution "bare-earth" topography, which is one of the most desirable benefits of LiDAR data with respect to fault mapping, particularly in densely vegetated areas.Several fault scarps generated by surface rupture earthquakes, which were not identified in previous geological surveys and aerial imagery, could be identified in forest-covered areas by using airborne LiDAR data (Harding and Berghoff, 2000;Kondo et al., 2008).
Within the last 2 decades, the LiDAR data have been increasingly applied in active tectonics including fault zone mapping (Arrowsmith and Zielke, 2009), the reconstruction of surface deformation (Gold and Cowgill, 2011;Zielke et al., 2012), investigation of geomorphic deformation (Hurst et al., 2013), and quantification of near-field co-seismic displacements (Hudnut et al., 2002;Oskin et al., 2012).The results of these studies generally challenge prior research findings and greatly promote the progress of active tectonics research (Kondo et al., 2008;Zielke et al., 2010;;Zhou et al., 2016).In particular, LiDAR is being used on a large scale in countries with active structures and relevant seismic hazards, such as the United States (Oskin et al., 2012;Zielke et al., 2012) and New Zealand (Clarke and Burbank, 2010;Langridge et al., 2014).
China is one of the most active seismic regions in the world and in recent decades, large-magnitude earthquakes have occurred frequently and abruptly, resulting in serious damage and numerous casualties, such as the 1976 Mw 7.8 Tangshan earthquake (Huang and Yeh, 1997), 2008 Mw 7.9 Wenchuan earthquake) (Xu et al., 2009), and 2021 Mw 7.4 Maduo earthquake (Yuan et al., 2022).Therefore, in China, LiDAR technology and surveys applied to investigate fault activity and earthquake surface deformation are crucial and will improve the country earthquake disaster assessment.
In recent years, Chinese scholars have carried out active fault mapping and fault activity studies in the Haiyuan and Altyn Tagh fault zones using LiDAR data (Zhang et al., 2013;Chen et al., 2014;Ren et al., 2015;Chen et al., 2018;Kang et al., 2019).Our research group carried out surveys in the Dushanzi anticline zone of the northern Tian Shan, including the morphological dating of terraces risers (Wei et al., 2015), paleoseismicity investigation using the fault scarp (Wei et al., 2019), and the automated mapping of landforms (Wei et al., 2017).In this paper, we summarized the application of LiDAR in active tectonics of China by reviewing the specific content of these works.In addition, some studies of high-resolution data in other aspects of active tectonics are generalized in this paper.Furthermore, we discussed the strategies to promote the application of the LiDAR technique in active tectonics studies of China as well as potential breakthroughs.

Introduction to the LiDAR Technique
The LiDAR technique involves the calculation of the distance between a laser emission point and an observed object based on the time between the emission and reception of the laser beam according to Eq 1.It operates on the premises that the time is accurately recorded and the laser propagation speed is the speed of light (Axelsson, 1999;Wehr and Lohr, 1999) where R is the distance between the laser emitter and the observed object surface (m), c is light speed of light (3 x 10 8 m/s), and t is the difference between the laser emission and reception time (s).
The observation system consists of a laser scanner (with position and orientation system (POS)), global positioning system (GPS), and inertial navigation system (IMU).Depending on the application of the observation system, four measurement methods can be utilized, including satellite-borne, airborne, vehicle-mounted, and ground-based methods.The raw data obtained by LiDAR measurements contain elevation and laser echo intensity information in a point cloud format.And the raw data accuracy depends on the distance between the measurement system and measured landform, GPS measurement accuracy of the ground control points, and weather conditions.Current LiDAR measurement equipment is manufactured by large international companies and preprocessed with the manufacturer's self-contained data processing algorithm and software.Preprocessing includes the calculation of the absolute coordinates of each measurement point, scanning angle, and GPS time, in reference to POS and GPS data of the aircraft as well as a ground reference point (Ackermann, 1999;Carter et al., 2007;Meigs, 2013).Generally, the most common data provided to the user are preprocessed and solved standard point cloud data (generally in Log ASCII Standard File format, LAS format).These point cloud data are surface data, including buildings and vegetation.The user can filter and classify the point cloud data based on the particular needs of a given study.And then, a Digital Surface Model (DSM) can be extracted from the data.If it is necessary to extract vegetation information from forest surveys, non-ground point (e.g., vegetation) information will be retained.To investigate active tectonics and earthquake records in the landscape, vegetation and "man-made" structures points are removed, so that the final product corresponds exclusively to the real topography (Cunningham et al., 2006).Finally, the classified point cloud data can be quantitatively analyzed and highresolution DEM data can be extracted.

Advantages of Airborne LiDAR Data
At present, airborne LiDAR and Structure from Motion (SfM) technology are the two most common techniques for obtaining high-resolution DEM data in China.
Compared with the digital photogrammetry technology, the advantages of airborne LiDAR are as follows (Baltsavias, 1999).1) LiDAR is a direct measurement of three-dimensional coordinates of the earth surface; while the traditional photogrammetry or Radar Interferometry (InSAR) obtain three-dimensional data of the surface through indirect methods.2) LiDAR can partially transmit vegetation, and effectively remove the effect of the vegetation (Figure 1).As for the photogrammetry technology, the influence of vegetation can only be removed by estimating vegetation height, which greatly affects the accuracy of topography.3) As an active sensor, the airborne LiDAR system has strong anti-interference ability, and is not affected by weather conditions such as light and clouds.
Structure from Motion (SfM), another technology of aerial photogrammetry, has developed rapidly due to its advantages of low cost, high flexibility and the ability of quickly acquiring highprecision 3D topographical data (Klinger et al., 2011).Especially in the arid areas or areas after crop harvesting or wilting, the point cloud data generated by the UAV-SfM technology can also accurately reflect the surface features (bare earth) (Wu et al., 2019;Ren et al., 2021).Therefore, the UAV-SfM technology has been an important complementary to LiDAR.

LIDAR SURVEYS IN DUSHANZI REVERSE FAULT, XINJIANG, CHINA Regional Background and the Data Acquisition
The east-west striking Tian Shan Mountains are one of the most active intracontinental mountain belts in the world (Figure 1A) (Molnar and Tapponnier, 1975;Burchfiel et al., 1999).The northern Tian Shan deformed rapidly because of the ~S-N trending continental collision of India and Asia.The deformation caused by the ongoing collision is accommodated by a series of east-west trending, subparallel zones of reverse faulting and folding (I, II, and III in Figure 2B), deforming a significant part of the northern piedmont of the Tian Shan (Tapponnier and Molnar, 1979;Allen et al., 1991;Sobel et al., 2006).Because of the ongoing deformation, seismic activity in this region is frequent (Molnar and Ghose, 2000).Approximately 80 earthquakes with magnitude greater than M 5 (three of which exceeded M 7) have occurred in the region since 1716 (Wang et al., 2004).
The Dushanzi Reverse Fault (DRF), which is located at the west end of the Dushanzi-Anjihai fold-thrust belt (Belt III), is the youngest reverse fault-fold belt.Imagery interpretation and field surveys showed that the DRF is ~15 km long and 6 km wide.At the surface, the DRF can be divided into two segments (Deng et al., 1996a;Deng et al., 2000): 1) the main fault, which is a reverse fault, and 2) a secondary fault, which is a shallower dipping thrust fault with a right lateral oblique slip component.These fault segments cross through the Kuitun River and piedmont slope of the mountains, forming prominent northfacing, south-side-up fault scarps, which are 2-10 m high, on terraces and alluvial fans (Avouac and Peltzer, 1993;Deng et al., 2000).
Many rivers originate from active glaciers in the axial parts of the Tian Shan and flow north and across the anticline belt (Figure 2B) (Molnar et al., 1994).These rivers flow along very deep valleys, where several terraces are present and provide information about past changes of the climate and regional tectonics.The DRF area is characterized by continental arid climate, with an annual precipitation of <200 mm/yr and the structural and geomorphic features of the DRF are well preserved under such arid climatic conditions (Deng et al., 2000).
The data acquisition area was divided into two parts, representing a T shape.This area comprises the DRF as well as the fluvial terraces at the hanging wall of the thrust zone.The E-W area is 22 km long, 3 km wide, and includes the surface rupture zone of the northern piedmont.And the S-N area is about 25 km long, 4 km wide, and includes the Kuitun fluvial terraces.
The average shot density and mean point spacing of the acquired LiDAR data are 6.5 point per m 2 and 0.39 m, respectively (Zhanyu et al., 2014).Firstly, 12 static GPS control points with mm-scale measurement accuracy was used to determine the relative vertical accuracy and mean square deviation, which are 0.12 and 0.078 m, respectively (Zhanyu et al., 2014).Subsequently, the optimal DEM resolution of the ground point cloud data was determined to be 0.5 m by using the inverse distance weighting algorithm (Zhanyu et al., 2014).At this resolution, the DEM data can be used to describe and characterize the geomorphological features of the DRF in detail.

Morphological Dating Using Fluvial Terrace Risers
Dating of the landforms displaced is essential to estimate the period associated to the measured offsets, allowing to quantify a fault slip rate and infer the occurrence patterns of large earthquakes (Zhang et al., 2013;Chen et al., 2014).However, in arid and semiarid regions, dating with 14C is challenging because of the lack of organic material or the material exceeds its dating range (older than 50ka).Despite the luminescence dating or cosmogenic nuclide dating are alternatives (Lal, 1991;Gosse and Phillips, 2001;Schaller et al., 2001;Cockburn and Summerfield, 2004), they are costly or time-consuming.Morphological dating, based on profile change, may provide estimates of the age of scarps developed in unconsolidated alluvial material (Culling, 1960;Nash, 1980;Hanks et al., 2000;Riebe et al., 2000;Crosby et al., 2004;Wei et al., 2015;Xu et al., 2021).
As shown in Figure 3A, the terrace riser separates adjacent terrace treads.When the low tread of a riser is abandoned, the riser is generally represented by nearly vertical slopes.Due to the gravitational collapse, the angle of the riser's free face would be reduced to angle of repose (generally about 28-37 °, depending on the deposition characteristics) (Pierce and Colman, 1986;Hsu and Pelletier, 2004;Harkins and Kirby, 2008).Following this short process, the riser degradation is dominated by less vigorous diffusive hillslope processes (Avouac, 1993).At this stage, the morphological age (kt) of the terrace riser is the sole measure of the profile morphology and a product of the time (t) because diffusive processes have become the active mechanisms for morphological change (Ren et al., 2015).Assuming the region's transport coefficient (k) value is uniform, when we obtained the morphologic age (kt) of these risers and the absolute age of one of them, we can efficiently determine the other terrace riser's relative age (Wallace, 1977;Harkins and Kirby, 2008).
Many geologists have extensively studied the general evolution of terrace risers (Pierce and Colman, 1986;Hsu and Pelletier, 2004;Clarke and Burbank, 2010).However, only a fraction of topographic profiles have been used for each location because the data acquisition largely depends on labor-intensive field surveys (Nash, 1980;Hanks et al., 2000).DEMs generated from airborne LiDAR allow us to collect and systematically analyze the scarp morphology without labor-consuming field surveys because of the detailed topographical resolution (resolution ≤0.5 m) and digital format (DuRoss et al., 2019;Crosby et al., 20202020;Bello et al., 2021).Based on the high-resolution LiDAR-derived DEMs, Wei et al. (2015) extracted 159 topographic profile swaths of terrace risers and analyzed to determine a range of best-fitting morphological age (Carter et al., 2007).
In every scarp site, Wei et al. (2015) collected a DEM swath with a width of 10 m that represents the terrace riser profile (Wei et al., 2015).Firstly, all grid points in the swath are projected onto the center line (black line in Figure 3B).Next, the above projected points are grouped at 1-m intervals, and constructed the center point of the binned data as a synthetic elevation profile, which represents the terrace riser's morphology (Figure 3C).From this profile, the gradient profile is calculated and then compared with these gradient profiles to determine the terrace riser's "best-fit" kt value (Figure 3D) (Wei et al., 2015).
Combined the morphological age (kt) of the T1/T2 riser and its independently absolute age (Molnar et al., 1994;Deng et al., 1996a;Deng et al., 2000), a locally applicable sediment transport coefficient (k) was determined (5.5 ± 1.6 m 2 /ky).Subsequently, the coefficient was used to estimate the age ranges of the terraces T2/T3, T3/T4, T4/T5, and T5/T6, which are 11.6 ± 3.4, 6.5 ± 1.4, 5.3 ± 1.1, and 4.2 ± 1.2 ka, respectively [Figure 4,22].It needs to be emphasized that the ages represent the lower terrace's abandonment ages.Poisson et al. (Poisson and Avouac, 2004) mapped the fluvial terraces along the Kuitun River and obtained the abandonment ages of some terrace by optically stimulated luminescence (OSL) samples or radiocarbon samples, in which T3, T4 and T6 are 10-11 ka B.P, 6.8-7.5 ka B.P. and 3.3-3.5 ka B.P, respectively.Through comparison, the ages obtained by these two approaches are consistent (Wei et al., 2015).It demonstrates that the DEM-based morphological dating technique is reliable and it can be used to quickly estimate the abandonment age of landform terraces.This method can be applied to a variety of other areas for which high-resolution DEMs are available.

Investigating Paleoseismicity Using Fault Scarp Morphology
Tectonic landforms, such as fault scarps and displaced channels, are the result of large magnitude earthquake surface ruptures overprinting surface processes.Fault scarp, as a typical seismic landform, is frequently used to indicate the fault location, estimate fault slip rate and infer past events characteristics (Wells and Coppersmith, 1994;Hilley and Arrowsmith, 2008).
Previous studies have employed the offset clustering distribution to identify the paleo-earthquake events and co-seismic displacements (McGill and Sieh, 1991;Grant and Sieh, 1994).Limited to the data accuracy, the identification and extraction of information from the tectonic landforms have led to persistent challenges (Chen et al., 2006).Within the last 2 decades, the DEM derived from airborne LiDAR enabled researchers to quickly obtain high-resolution three-dimensional (3D) topographic data.Therefore, the comprehensive information, including the scarp height and slope change, can be used to study the paleoseismic history (Wei et al., 2020).
To study the morphological characteristics of a fault scarp in unconsolidated material, Wei et al. (2018) selected a wellpreserved scarp (~2 km) of the DRF.Firstly, profiles of scarps normal to the fault were extracted from the DEM with lengths depending on the width of the surface deformation, generally spanning 40-160 m (Figure 5A).To measure the scarp heights, they fitted separate straight lines with a simple least squares regression to the profile segments on each side of the scarp slope as the original geomorphic surfaces and the vertical separation between the lines was measured as the vertical displacement at the midpoint of the scarp slope [Figure 5B,23,86].
As for the slope change of unconsolidated-sediment fault scarp, the generally evolution as follows.After one event occurred, the scarp morphology would be reset and lost its previous morphology consequently due to the gravitational collapse (Figure 6A).According to the degradation model, there are multiple slope breaks in the fault scarp morphology, and each slope break represents a surface rupture event for a compound fault scarp formed by multiple fault rupture 80,81].Different morphologies of the cumulative reverse fault scarps were obtained through the numerical models, and can be summed up in three overall cases (Carretier et al., 2002;Wei et al., 2019).To determine the slope breaks in profile, Wei et al. (2018) compute the gradient, the first derivative, along each elevation profiles and document the gradient changes.To avoid the overinterpretation of gradient changes, they identified these slope breaks by visually checking the gradient changes in the calculated profiles (Wei et al., 2019).
Based on the 237 vertical displacements in various landform surfaces (Figure 7A), Wei et al. (2018) acquired the normal distribution of the along-strike scarp heights, expressed as a COPD (cumulative offset probability distribution) (Figure 7B).Due to the repeated surfacerupturing events exist on a fault, the offsets clusters with similar displacements are interpreted as the cumulative slip associated with successive paleoearthquakes (McGill and Sieh, 1991;Yeats et al., 1997;Zhou et al., 2020b).In addition, based on the gradient analysis of 94 scarp profiles, Wei et al. (2018) determined several slope breaks that account for several surface-faulting events in difference terraces [Figure 7B,23].Statistical analysis shows that the selected fault segment has most likely experienced four individual rupture events since the formation of terrace T2 and three events since the formation of T1 (Wei et al., 2019).
Combining the surface rupture length (14-18 km), mean slip of single event (1.8 ±0.5 m) and the cascade rupture length (55-60 km), the magnitude of the DRF was estimated between Mw 6.5-6.7 (Wells and Coppersmith, 1994;Wei et al., 2019).In addition, based on the previous studies about the time frame for terraces along the Kuitun River (Deng et al., 2000;Poisson and Avouac, 2004;Yang et al., 2011) and the surface-rupturing event numbers, they further estimated the average recurrence interval of large earthquakes in DRF.
This study in Wei et al. (2019) demonstrated that fault scarps can be served as a geomorphic marker to quantify the number of surface-rupturing events and estimate cumulative   (Wei et al., 2019).In case A, (A) and (B) show the scarp morphology, the associated slope profiles and the profile of fault scarp (inset box) for faults cutting the surface roughly at the same place.In case B, (C) and (D) show the scarp morphology, the associated degraded slope profiles and the profile of fault scarp (inset box) for the successive faults are distant enough to preserve the morphology inherited from previous erosion period.In case C, (E) and (F) show the scarp morphology, the associated degraded slope profiles and the profile of fault scarp (inset box) for the distance between faults is moderate, and a portion of the previous scarp can be preserved from gravitational collapse.(G) and (H) show the fault scarps and corresponding slope profiles after multi-event earthquakes, as in Cases B and C (Wei et al., 2019).and single-event vertical displacements (Wei et al., 2019).The results also highlight the importance of high-resolution LiDAR-derived DEM in paleoearthquake surveys.

Automated Geomorphological Mapping
Geomorphological maps, which are often used to describe topographic features, can provide information about the origin and evolution of landforms.In addition, in some tectonically active areas, geomorphological maps can also provide information about the risks associated with earthquakes and landslides (Vedder and Wallance, 1970;Arrowsmith and Zielke, 2009).In the past, mapping in geomorphic analysis generally depended on the stereoscopic interpretation of aerial photographs and time-consuming, laborious field surveys (Masoud and Koike, 2006).The results of geomorphologic analysis may be biased due to subjective experiences, especially in areas which are steep and difficult terrain (Williams et al., 2012).
Therefore, the demand for automatic geomorphological mapping based on DEM is increasing (MacMillan et al., 2000;Lohani and Mason, 2001;Ostaficzuk, 2005;Prima et al., 2006;Evans, 2012).The previous researches are usually used to obtain the large-scale topography, such as basins and ranges, because its resolution is generally about 20-80 m horizontally and 10 m vertically (Smith et al., 2006;Suwandana et al., 2012).
Within the last 2 decades, the explosion in sub-meter resolution airborne LiDAR data raises the reliability and practicality of geomorphological mapping (Nissen et al., 2012;Lin et al., 2013;Dong, 2015).Compared with largescale DEM data mentioned above, DEMs derived from LiDAR are of great advantage with respect to the detailed description of microscopic topographic elements related to tectonic activities and surface processes.Therefore, Wei et al. (2017) utilized high-resolution LiDAR data from the Dushanzi anticline for automated landform mapping research (Wei et al., 2017).
After the data analysis, Wei et al. (Wei et al., 2017) utilized four types of morphometric attributes, including the relief variance, fractal dimension, area rate, and crossover length, to define the topographic features of specific terrains.The above-mentioned four morphometric attributes were used as morphometric variables in classification algorithms.The automated mapping process consists of four steps (Figure 8): 1) using the moving window routine to generate thematic raster maps of the terrain properties; 2) obtaining the feature vectors for different landform types by delineating the training areas; 3) computing the Mahalanobis distance between the predefined classes and each unclassified pixel; and 4) classifying each pixel of the thematic raster maps with the minimum Mahalanobis distance classifier and generating a map (Wei et al., 2017).
Compared with the field mapping (Deng et al., 1995) and image interpretation, the automated landform classification method can be used to reconstruct a detailed geomorphological map with a classification accuracy of 72.9% and a kappa coefficient (κ) of 0.66 (Figure 9) (Wei et al., 2017).The results of the interpretation of the automatic classification reveal alluvial landforms with a notable east-west direction.This earthquake rupture zone is closely related to the thrust fault, which should be avoided in major engineering projects (Wei et al., 2017).With the increasing popularity of high-resolution digital terrain data, automatic landform classification is expected to be widely applied to geomorphological characterization and mapping analyses due to its high efficiency and accuracy.

LIDAR APPLICATION IN OTHER ASPECTS OF ACTIVE TECTONICS
As mentioned, we reviewed our works which were based on the LiDAR data, including dating the terrace risers through its morphology, investigating paleoseismicity using fault scarp, and automated geomorphological mapping.In addition, in the field of active tectonics, LiDAR data also play a great role in coseismic deformation, offset measurement, and the monitoring of inter-earthquake deformation.

Cumulative Displacement Measurements
The first application of LiDAR in active tectonics was a survey of surface rupture associated with the Hector Mine earthquake in southern California in 1999 (Hudnut et al., 2002).The results showed that the resolution and accuracy of data yielded by LiDAR exceeded what could be obtained by traditional methods (Meigs, 2013).Subsequently, based on the sub-meter topographic data using airborne LiDAR and UAV photogrammetry (UAV-SfM), the researchers successfully reconstructed the offset accumulation pattern of multiple faults from geomorphic evidence (Scharer et al., 2010;Zielke et al., 2010;Klinger et al., 2011;Zielke et al., 2015).
In China, Ren et al. (2015) used the Lidar data to measure the cumulative offsets along Haiyuan fault and reevaluate the singleevent offsets of the 1920 Haiyuan Ms 8.5 earthquake (Ren et al., 2015).They found the previously reported maximum displacement of the 1920 earthquake might be the result of at least two earthquakes, which challenged our understanding of the Haiyuan Fault's characteristic (Ren et al., 2015).And the approach has been applied on other large faults in China, such as the Altyn Tagh Fault (Ren et al., 2015), the Fuyun fault (Liang et al., 2021), the Tianjingshan fault (Li et al., 2017), and the Xianshuihe fault (Ma et al., 2022).
However, it has also been pointed out that the applicability of cumulative offset probability distribution (COPD) is an important topic.Lin et al. (2020) combined the statistical modeling with global historical coseismic surface rupture data to evaluate the conditions required for COPD (Zhou et al., 2020a;Zhou et al., 2020b).The results show that robust COPD interpretation on large fault needs low offset variability, sufficient offset measurements and other constraints like paleoseismic event sequence data (Ren et al., 2015;Zhou et al., 2020a).In addition, the results from LaDiCaoz (Zielke and Arrowsmith, 2012), a commonly used offset measurement software, often produces systematic errors in the offset measurement process because it requires the field survey or subjective determination of the strike of faults and displaced gullies.Thus, developing a fully-automatic repeatable and verifiable offset-measuring algorithm can be considered in the future.

Co-Seismic Deformation Detection Using Pre and Post-Earthquake LiDAR DEM
After the 2010 EL Mayor-Cucapah earthquake, Oskin et al. (2012) obtained the co-seismic deformation field by subtraction of pre-earthquake and post-earthquake Lidar data, providing a new idea for future co-seismic deformation research (Nissen et al., 2012;Oskin et al., 2012).
Chinese scholars have also noticed the advantages of LiDAR technology in coseismic deformation observation and applied it to practical work.After the 2010 Yushu Earthquake, Yuan et al. (2012) used ground-based terrestrial laser scanning surveys (TLS) to scan the Changusi segment, and obtained fine 3D data of the surface rupture zone at this point, and also obtained its vertical and horizontal amount of co-earthquake dislocations (Yuan et al., 2012).
After the 2021 Maduo Earthquake, Yuan et al [2022] used UAV-SfM to take photographs of the surface rupture of the Maduo Earthquake systematically, and quickly obtained the detailed surface rupture distribution and horizontal and vertical displacement distribution maps of the Maduo Earthquake (Yuan et al., 2022).This work has demonstrated great efficiency of UAV in processing post-earthquake emergency data, and also improves the precision and accuracy of coearthquake dislocation measurements.
Based on the large-scale acquisition of high-resolution DEM data, it is possible to obtain more pre-earthquake data.Therefore, the study of co-seismic deformation based on Lidar data will be increasingly applied to active tectonics, and provide a basis for the study of earthquake occurrence mechanism and deformation mechanism.

SUMMARY
Within the last 2 decades, the explosive development of LiDAR acquisition technology has pushed active tectonics research to a new level (Karabacak et al., 2011;Langridge et al., 2014;Ren et al., 2014).Compared with the conventional topographical data, LiDAR data has unique technical advantages.
Firstly, the main advantage of LiDAR is that it can obtain the actual surface topographic information by eliminating vegetation data even in densely vegetated areas (Meigs, 2013;Glennie et al., 2014).This property enables the LiDAR becomes applicably for tectono-geomorphic features analysis and the extraction of the detailed geometry and kinematics of the fault zone (Gold and Cowgill, 2011;Papanikolaou et al., 2015).Secondly, after the debut of LiDAR as a means for survey the co-seismic surface rupture, its stunning accuracy and resolution highlights the potential in active tectonic research (Hudnut et al., 2002).And since then, more and more studies employed LiDAR data and advanced our understanding of active tectonics, such as the coseismic deformation field analysis using before and after earthquake differential LiDAR data (Nissen et al., 2014;Ren et al., 20182018), identifying both the on-and off-fault deformation along strike slip faults (Oskin et al., 2012;Ren et al., 2015;Zielke et al., 2015) or monitoring the after-slip (Wilkinson et al., 2010).In addition, compared with optical remote sensing satellite measurements, the airborne LiDAR is less affected by weather than, thus it can be applied for emergency works in the case of natural disasters, such as earthquakes and volcanoes.Thus, with the development of LiDAR surveying, numerous active fault systems in United States, New Zealand, and Japan have been extensively surveyed via airborne LiDAR, and a rapidly growing LiDAR datasets is now available (e.g., www.opentopography.org)(Zielke et al., 2015).
The ~20 years of development of LiDAR survey in active tectonics have allowed an unprecedented improvement in this research field, promoting the active tectonics research from a qualitative to a more and more quantitative one (Zielke et al., 2015).And in China, based on the LiDAR data, the researchers have carried out the active faults mapping, fault activity studies, and the identification of earthquake recurrence behavior in Haiyuan, Dushanzi, and Altyn Tagh fault (Chen et al., 2010;Zhang et al., 2013;Chen et al., 2014;Ren et al., 2015;Wei et al., 2015;Chen et al., 2018;Kang et al., 2019).Nevertheless, the main foci of these studies were the interpretation of tectonic landforms and the extraction of the offsets, which have significant limitations with respect to innovative data analysis and data mining methods.In addition, the current problems of LiDAR include the high cost of LiDAR scanning, and the difficulty of implementing point cloud filtering in areas with steep slopes such as mountains and canyons.
Therefore, the efficient application of LiDAR technologies and data for the analysis of fault activity and earthquake disaster assessment should be the focus of future active tectonics studies.This is also critical for promoting studies of the evolution of the tectonic geomorphology, seismic fault rupture modes, and other related issues in China.We hope that more researchers will recognize the advantages of LiDAR technology in active tectonics studies and apply this technology to promote the development of active tectonics research.

FIGURE 1 |
FIGURE 1 | Comparison of different resolution data at the same location (Bevis et al., 2005).(A) 10 m resolution DEM.(B) 0.5 m resolution DEM; the elevation information obtained from the last pulse of LiDAR, represents the bare-earth DEM; the white arrows indicate fault trace.(C) 0.5 m resolution DEM; the elevation information obtained from the first pulse of LiDAR, represents the data obtained by UAV-SfM.
FIGURE 2 | (A) Simplified tectonic map showing the location of the Tian Shan.Intracontinental regions within the Eurasian continent were deformed due to the collision between India and Eurasia; (B) Three approximately east-west stretching fold and fault zones (labeled Belts Ⅰ, Ⅱ, and Ⅲ) along the northern piedmont of the Tian Shan.The white rectangle represents the study area(Wei et al., 2019).

FIGURE 3 |
FIGURE 3 | Usage of the DEM to conduct swath profile-based morphological dating (Wei et al., 2015).(A) Geometry and diffusive degradation of the terrace riser profile.(B) Hillshade plots were used to identify the terrace riser trace and determine the riser profile sites.(C) Full scarp modeling under a range of kt values was used for the gradient profile.(D) Calculated RMS 2 misfits (errors) between the synthetic riser slopes and modeled slopes over a range of kt values.

FIGURE 4 |
FIGURE 4 | Diffusion ages of the terrace risers estimated based on the analysis of the riser diffusivity and its location of different terraces (Wei et al., 2015).The histograms on the left show the probability distribution of the diffusion ages and frequency distribution of ages, respectively.The map on the right indicate the location of different terrace risers.(A) Distribution of river terraces along the Kuitun River; (B) Enlarged view of the black box in (A); (C) Hillshade from LiDAR-derived DEM, the area is the same as (A).

FIGURE 5 |
FIGURE 5 | Vertical displacement measurement and slope breaks detection for a fault scarp (Wei et al., 2019).(A) Hillshade plots were used to identify the fault scarp and determine the profile site; (B) Measurement of vertical displacement and detection of slope breaks from scarp and gradient profiles, respectively.

FIGURE 6 |
FIGURE 6 |The thrust fault scarp morphologies in multiple seismic activities(Wei et al., 2019).In case A, (A) and (B) show the scarp morphology, the associated slope profiles and the profile of fault scarp (inset box) for faults cutting the surface roughly at the same place.In case B, (C) and (D) show the scarp morphology, the associated degraded slope profiles and the profile of fault scarp (inset box) for the successive faults are distant enough to preserve the morphology inherited from previous erosion period.In case C, (E) and (F) show the scarp morphology, the associated degraded slope profiles and the profile of fault scarp (inset box) for the distance between faults is moderate, and a portion of the previous scarp can be preserved from gravitational collapse.(G) and (H) show the fault scarps and corresponding slope profiles after multi-event earthquakes, as in Cases B and C(Wei et al., 2019).

FIGURE 7 |
FIGURE 7 | Distribution of the displacement along the fault segment (A) and its characteristics of vertical displacement clustering (B) (Wei et al., 2019).

FIGURE 9 |
FIGURE 9 | Comparison of the map obtained from supervised classification of landforms (automated classification) (C) with the field-based geomorphological map (A) and interpretation-based geomorphological map (B) (Wei et al., 2017).