Microfracture density in fault damage zones can reflect spatial variability that decays in intensity as a function of distance from the fault, which is crucial in understanding the mechanical, seismological, and fluid-flow properties of the fault system. However, few studies explored the characteristics of fracture density between the two sides of active dip-slip faults due to rare field observations. Here, we measured and modeled microfractures across an active thrust fault associated with the 2008 Mw 7.9 Wenchuan earthquake in the Longmen Shan, eastern Tibetan Plateau. The results showed that the microfracture density at the Qingping site developed more intensely in the hanging wall than in the footwall for an exposed thrust fault, indicating an asymmetrical pattern. The hidden thrust fault at the Jushui site showed that microfractures developed more intensely in vertical planes in the hanging wall than in the footwall, whereas microfractures developed similarly in horizontal planes within the two sides, indicating a quasiasymmetrical pattern. Comparing the data at the two sites with computational modeling, we suggest that fault geometry might exert a first-order control of the asymmetrical microfracture density pattern, which is helpful for revealing different deformational behaviors of rock masses in the fault damage zones and better understanding the hanging-wall effect for evaluating seismic hazards on active thrust faults.

A fault damage zone, expressed as a zone with numerous fractures surrounding a narrow fault core, has been considered to be related to coseismic loading and, therefore, has the potential to reveal the rock deformational mechanics and past earthquake rupture conditions [1-7]. Moreover, such a damage zone is expected to act as conduits, barriers, or combined conduit-barrier systems that play a fundamental role in crustal fluid flow [8-10]. Therefore, quantitative determination of characteristics of fractures in the fault damage zone is critical to understand the mechanical and seismological properties of the fault system.

Geometrically, fracture density is one of the key parameters in evaluating the spatial variability that decays in intensity as a function of distance from the fault [11, 12]. Many studies have measured micro/mesofracture density on fault-perpendicular transects to show that fracture density decreases gradually away from the fault core, which can be simplified to fit either an exponential decay model [13] or a power law decay model [14, 15] in the fault damage zone. Moreover, previous studies have suggested that the characteristics of fracture density might be influenced by the amount of slip across the fault, the size of the fault, lithology, rupture processes, and movement history [8, 13, 16]. For example, Caine et al. [8] suggested that a wide damage zone may indicate the effect of more repeated seismic events with greater accumulative deformation than that of a narrow damage zone. Ostermeijer et al. [12] investigated bedrock exposure at the Borrego Fault in detail and suggested that the distribution patterns of fracture density might be more complex due to subsidiary faults. However, few studies compared the characteristics of fracture density between the two sides of active dip-slip faults (e.g., thrust faults or normal faults). Whether the fracture density within the two sides is symmetrical? Are there any dominant differences in spatial patterns for fractures within the two sides? If so, what are the primary factors contributing to the differences? Answering these questions can help us better to understand the deformational mechanisms of fractures in fault damage zones, which is also helpful for us in understanding spatial–temporal evolution of the fault zone architectures [8].

To address the aforementioned issue, theoretical analyses of seismic ruptures for different fault geometries might have some implications. For example, active dip-slip faults characterized by a nonvertical fault plane show variations in the normal stress that results in asymmetrical deformation within the two sides of the fault [17]. Influenced by multiple earthquakes and the overprinting of successive stages of fault deformation [8], the fault damage zone of active dip-slip faults might possibly show a pattern of an asymmetrical fracture density within the two sides. To date, this speculation has been scarcely verified by field-based evidence due to difficulties in revealing a natural bedrock transect that can be well exposed across the two sides of active dip-slip faults. Because the downthrown side of active dip-slip faults always subsides due to fault movement and is then covered by thick deposits.

Here, we first analyzed microfracture (cm-scale) density in the fault damage zone within the two sides of the seismogenic thrust fault related to the 2008 Mw 7.9 Wenchuan earthquake in the Longmen Shan, eastern Tibetan Plateau (Figure 1). We suggest that the microfracture density in the hanging wall is more intensely developed than that in the footwall for both exposed and hidden active thrust faults. Our results indicate that fault geometry might exert a first-order control of the asymmetrical microfracture density pattern across active thrust faults, which further increases our understanding of the mechanics of the hanging-wall effect for evaluating seismic hazards.

The Longmen Shan thrust fault is located at the eastern margin of the Tibetan Plateau and primarily consists of three northeast-trending branch faults, including the Maoxian–Wenchuan fault (MW), Yingxiu–Beichuan fault (YBF), and Jiangyou–Guanxian fault (JGF; Figure 1). The thrust fault began to develop in the Late Triassic and experienced strong contractional deformation during the Indo-China movement and Himalayan movement, resulting in a height deviation of up to 4000 m between the Tibetan Plateau and Sichuan Basin within a range of several tens of kilometers [18, 19]. Stratigraphically, Neoproterozoic basement rocks surrounded by Paleozoic sedimentary strata crop out in the core part of the Longmen Shan [20]. To the west of the Longmen Shan, a thick (>8–10 km) sequence of strongly folded Triassic flysch deposits was primarily developed [21-23]. To the east, the strata are Late Triassic–Quaternary deposits [24, 25].

The 2008 Mw 7.9 Wenchuan earthquake occurred along the YBF and JGF, which were Holocene active and produced approximately 240 and 90 km lengths of coseismic surface ruptures, respectively. Moreover, ∼7 km surface ruptures were produced by a northwest-trending Xiaoyudong fault [26-28]. Comparing the two parallel-trending seismogenic faults, the average coseismic vertical offset was ∼3 to 5 m on the YBF, whereas the JGF shows an average coseismic vertical offset at ∼1 to 2 m [26-28], indicating that coseismic thrust deformation on the YBF is much stronger than that on the JGF. Moreover, multiple paleoseismic trenches on the two faults revealed that previous seismic events showed greater coseismic vertical offsets on the YBF than on the JGF [29]. Although the fault collectively produced more than 300 km of coseismic surface ruptures, the downthrown side of the fault was mainly covered by slope/fluvial-related deposits [28]. Based on the interpretation of high-resolution images and detailed field investigations, we find that only two sites show well-exposed bedrock transects across the two parallel-trending faults, which are located at the Qingping site on the YBF and the Jushui site on the JGF (Figure 1).

3.1. Sampling for Thin Sections

Determining the spatial distribution of fracture density within a fault zone has been primarily focused on mesoscale and microscale damage [12]. The bedrock exposures at the two sites are partly covered by deposits and vegetation; therefore, we collected oriented rock samples to make thin sections, in which we counted cm-scale microfractures [12]. Considering that the Longmen Shan thrust fault shows maximum compressive stress toward the southeast [30], we chose a horizontal plane (horizontal thin section [HTS]) and two vertical planes (vertical thin section [VTS1] and VTS2; Figure 2) for each sample to observe the three-dimensional structures of the microfractures. Moreover, we prepared three thin sections for each plane for comparison. Each thin section was produced with a length of 3 cm and a width of 2.4 cm. Some thin sections were broken into irregular shapes during grinding; however, they did not affect our measurements.

3.2. Determination of the Horizontal Distance between Sampling Sites and Faults

All the samples were plotted to the corresponding digital elevation model data (Figures 1(b) and 1(c)). Then, we drew a line that was nearly perpendicular to the fault trace and made all the samples approach the line as uniformly as possible. Third, we used the coupe function of GMT [31] to plot the samples to the vertical plane that corresponded to the line, which helped us determine the horizontal distance between the sampling sites and the fault.

3.3. Measurement of Microfractures

To obtain the length and number of microfractures per area (online supplementary Tables S1–S8), we digitized all the photos of the thin sections (online supplementary Figure S1). First, we used light illumination toward the back of the thin section, which can help us better identify microfractures in the front of the thin section. Second, we used controlling points to make orthographic projections of the photos. Third, we interpreted microfractures with polylines and mapped the shape of thin sections with polygons. All the polylines and polygons were imported into CAD software to calculate the total length of the microfractures and the related areas, which yielded the length and number of microfractures per area. Some thin sections from the samples at the Jushui site (e.g., JS-02, JS-03, JS-05, JS-09, JS-10, JS-14, and JS-15; online supplementary Figure S2) show some stylolite-related fissures and pores. Considering that (1) these fissures and pores were affected by chemical weathering and might not truly reflect fault deformation, and (2) the samples were within the two sides of the fault, we did not count them during the measurement.

3.4. Statistical Analysis

Because we have three thin sections at each plane, we could check their reliability. We used the following rule to determine which thin section is useful for further statistical analysis: (1) if three measured data are all concentrated, all the data are useful; (2) if two of the three data are similar and the other one is dominantly different, we do not use the different data; and (3) if all the data are scattered, we would make some extra thin sections to determine which ones are more concentrated. For example, HTS from sample B-05 at the Qingping site yielded three groups of microfracture lengths per area at 0.88, 2.16, and 0.54 cm/cm2 (online supplementary Table S2). We suggest that the second value (2.16 cm/cm2) is an outlier. In other words, we do not use it for further analysis. All the measured data were further analyzed using the power law decay model (F = F0× r−n, where F0 is the fault constant [the fracture density at unit distance from the fault], r is the distance from the fault, and n is an exponent describing the decay), which might better reflect perturbations and the related mechanism of propagating crack tips [14].

3.5. Finite Element Modeling

Considering that microfractures are possibly related to long-term transient coseismic loading due to faulting movement [1-3, 5, 6], the coseismic deformational amount could be considered a proxy and provide insight for understanding the mechanism of the microfracture density pattern. Here, we used the finite element software PyLith [32] and built a simplified model according to the fault geometry at the two sites to reveal coseismic deformation during one seismic event. We meshed the fault geometry that is embedded in a homogeneous, elastic crust in three dimensions (50 km in the X direction, 50 km in the Y direction, and 15 km in the Z direction). Other parameters for the modeling are listed in Table 1.

4.1. Microfracture Density at the Qingping Site

The Mianyuan River flows southward and produces a deep-cut exposure on the northern side of the river (Figure 3). Fifteen samples were collected within the approximately 800 m long transect across the fault (Figures 3 and 4). The 2008 Wenchuan earthquake produced a coseismic vertical offset of ∼3.8 m on the first terrace to the south of the Mianyuan River (Figures 3 and 5). To the north of the river, the fault with a dip angle of ∼60° ruptured the bedrock transect that mainly consists of Precambrian, Permian, and Devonian limestone and sandstone (Figure 1(b)). Samples B-06 and B-09 collected in the footwall of the fault were sandstone, whereas the other samples were limestone.

We focused on two key parameters: microfracture length per area (MLPA) and microfracture number per area (MNPA; Figure 6, online supplementary Tables S1–S4). Specifically, MLPA from HTSs in the hanging wall shows a function at F = 237.03 × r–1.03, whereas MLPA in the footwall presents a function at F = 7.80 × r–0.46 (Figure 6(a)). The two functions indicate that the convergence trend is much steeper for the hanging wall than for the footwall. Moreover, the nearest sampling site in the hanging wall from the fault is ∼61 m away and yields the maximum MLPA at ∼4.3 cm/cm2, whereas the closest sampling site in the footwall away from the fault is shorter at ∼29 m but yields a much smaller maximum MLPA at ∼2.4 cm/cm2 (Figure 6(a)). Similarly, MNPA from HTS in the hanging wall shows a function at F = 243.51 × r–1.06 that is steeper in convergence trend than that of the footwall characterized by a function at F = 6.46 × r–0.44 (Figure 6(b)). The corresponding maximum MNPA in the hanging wall is ∼4.7 #/cm2, whereas the maximum MNPA in the footwall is smaller at ∼2 #/cm2.

For microfractures observed in vertical thin sections VTS1, we suggest that MLPA in the hanging wall shows a function at F = 414.27 × r–1.16, whereas MLPA in the footwall presents a function at F = 5.62 × r–0.38 that is gentler in convergence trend (Figure 6(c)). In addition, the maximum MLPA in the hanging wall is ∼4.4 cm/cm2, whereas the maximum MLPA in the footwall is smaller at ∼2.4 cm/cm2 (Figure 6(c)). Similarly, MNPA from VTS1 in the hanging wall shows a function at F = 367.05 × r–1.17 that is steeper in convergence trend than that of the footwall characterized by a function at F = 5.64 × r–0.38 (Figure 6(d)). The corresponding maximum MNPA in the hanging wall is ∼4.6 #/cm2, which is greater than that in the footwall at ∼2.4 #/cm2.

For microfractures from vertical thin sections VTS2, we suggest that MLPA in the hanging wall shows a function at F = 40.41 × r–0.58, whereas MLPA in the footwall presents a function at F = 11.52 × r–0.51 that is also gentler in convergence trend (Figure 6(e)). In addition, the maximum MLPA in the hanging wall is ∼6.1 cm/cm2, which is greater than that in the footwall at ∼4.0 cm/cm2 (Figure 6(e)). Similarly, MNPA from VTS1 in the hanging wall shows a function at F = 43.85 × r–0.60 that is steeper in convergence trend than that of the footwall characterized by a function at F = 11.08 × r–0.47 (Figure 6(f)). The corresponding maximum MNPA in the hanging wall is ∼5.9 #/cm2, whereas the maximum MNPA in the footwall is smaller at ∼3.6 #/cm2.

Based on the aforementioned results, we suggest that the development of microfractures in the horizontal and vertical planes within the hanging wall is much stronger than that of the footwall, which shows an asymmetrical pattern of microfracture density across the fault.

4.2. Microfracture Density at the Jushui Site

Liu-Zeng et al. [26] and Chen et al. [33] did not find the coseismic offset and observed discontinuously weakly deformed fissures or sand emissions at the Jushui site. Therefore, they suggested that the coseismic surface ruptures on the JGF ended at the site. Moreover, we did not observe a bedrock fault at the site, which indicates that the fault might be a hidden thrust fault and did not rupture to the surface. To better compare the microfractures across the hidden fault, we used the discontinuous surface deformation as the reference to define the part to its west side as the hanging wall and the part to its east side as the footwall (Figures 7 and 8).

Due to the downcutting of the Ganhezi River, a bedrock transect is well exposed at the site and mainly consists of Triassic limestone, sandstone, and mudstone (Figure 1(c)). Seventeen samples within an approximately 900 m long transect were collected within the two sides of the fault (Figures 7 and 8), of which three mudstone samples (JS-07, JS-08, and JS-11) in the hanging wall were too soft and weakly cemented to prepare for thin sections (online supplementary Table S5). Specifically, the MLPA from the HTS in the hanging wall and the footwall shows a function at F = 22.10 × r–0.73 and F = 14.02 × r–0.62, which indicates a similar convergence trend within the two sides (Figure 9(a)). The nearest sampling sites away from the fault in the hanging wall and footwall are similar at ∼39 and ∼45 m, which yield an identical maximum MLPA at ∼1.7 cm/cm2 for the two sides (Figure 9(a)). In addition, MNPA from HTS in the hanging wall shows a function at F = 8.07 × r–0.56 that is also comparable with the footwall characterized by a function at F = 8.69 × r–0.54 (Figure 9(b)). The maximum values (∼1.1 #/cm2) of the MNPA in the hanging wall also appear to be close to those of the footwall (∼1.3 #/cm2). Therefore, we suggest that HTSs within two sides of the fault show a comparable microfracture density pattern.

However, the MLPA from vertical plane VTS1 in the hanging wall shows a function at F = 40.92 × r–0.83, whereas the footwall presents a function at F = 4.22 × r–0.42 (Figure 9(c)). The two functions indicate that the convergence trend is steeper in the hanging wall than in the footwall. Moreover, the maximum MLPA is ∼2 cm/cm2 in the hanging wall, whereas the corresponding values in the footwall are slightly smaller at ∼0.9 cm/cm2 (Figure 9(c)). In addition, MNPA in the hanging wall shows a function at F = 50.77 × r–0.93 that is steeper in convergence trend than that of the footwall characterized by a function at F = 1.76 × r–0.28 (Figure 9(d)). The maximum MNPA in the hanging wall is ∼2 #/cm2 compared with that of the footwall at ∼0.7 #/cm2.

For microfractures observed in vertical planes VTS2, MLPA in the hanging wall shows a function at F = 306.03 × r–1.23, whereas the footwall presents a function at F = 3.80 × r–0.24 that is also gentler in convergence trend (Figure 9(e)). Moreover, the maximum MLPA is ∼2.8 cm/cm2, whereas the corresponding values in the footwall are smaller at ∼2.0 cm/cm2 (Figure 9(e)). In addition, MNPA in the hanging wall shows a function at F = 59.89 × r–0.87 that is steeper in convergence trend than that of the footwall characterized by a function at F = 3.17 × r–0.25 (Figure 9(f)). The maximum MNPA in the hanging wall is greater at ∼3.0 #/cm2 than that of the footwall at ∼1.9 #/cm2.

Therefore, we suggest that two vertical planes VTS1 and VTS2 within two sides of the fault show an asymmetrical microfracture density pattern, indicating that microfractures in the hanging wall are more intense than those in the footwall.

4.3. Deformational Fields Modeled at the Two Sites

We used a coseismic offset of 3.8 m and a dip angle of 60° to model deformational fields at the Qingping site. First, we chose the line below the surface by 1 km in the modeling to compare deformational differences between the two sides of the fault (Figure 10). Specifically, the deformational amount in the hanging wall increased from approximately 0.9 to 2.6 m. After passing the fault, the deformational amount sharply dropped to 1.6 m and decreased to 0.7 m. The convergence trend of the deformational amount in the hanging wall is evidently steeper than that of the footwall. Second, all deformational fields in the hanging wall are more intense than those in the footwall, including the coseismic deformational amount and the related influenced width away from the fault.

For the modeling at the Jushui site, we set the hidden fault with the same dip angle (60°; Figure 11). Considering that (1) the range-front fault of the Longmen Shan was buried at a depth of ∼2 km [34] and developed coseismic fissures or sand emissions during the 2008 Wenchuan earthquake [35] and (2) the nearest coseismic rupture to the south of the Jushui site is approximately 5 km away with an offset of ∼0.9 m [33], we suggest that parameters for a burial depth of 2 km and deformational amount of 0.5 m at the Jushui site are reasonable for building the fault geometry model. Similarly, the line below the surface by 1 km in the modeling also shows deformational differences within the two sides of the fault (Figure 11). Specifically, the deformational amount in the hanging wall increased from approximately 0.13 to 0.26 m with a slightly steeper convergence trend. After passing the fault, the deformational amount decreases to 0.08 m with a gentler convergence trend.

Moreover, considering that the modeling parameters such as coseismic offset, dipping angles, and buried depth of the faults at the two sites might have uncertainties compared with the true complex structures of the Longmen Shan Fault, we further set different parameter groups to explore deformational fields using scenarios (1) offset at 2 m; (2) dipping angles at 30° and 60°; (3) buried depth at 2 and 4 km (Figure 12). Specifically, greater offset indicates that the convergence trend in the hanging wall is much steeper than that in the footwall based on the comparisons between Figures 10 and 12. Similarly, larger dipping angles might show smaller difference in the convergence trend between the hanging wall and footwall (Figures 12(a) and 12(b) and Figures 12(c) and 12(d)). Furthermore, greater depth of buried fault could decrease the difference in the convergence trend between the hanging wall and footwall (Figures 12(c) and 12(e) and Figures 12(d) and 12(f)).

Compared with the modeling for the two sites, we suggest that the deformational fields show evidently asymmetrical patterns across the faults, which is evidenced by the fact that the deformational fields, including the coseismic deformational amount and the related influenced width in the hanging wall, are greater than those in the footwall (Figures 10-12). Moreover, the deformational amount at the Qingping site is approximately much greater (approximately ten times) than that at the Jushui site. The influenced width at the Qingping site is also much wider than that at the Jushui site. Therefore, deformational fields from the modeling further indicate that the asymmetrical pattern is more dominant at the Qingping site than at the Jushui site.

5.1. Comparisons of Microfracture Density between Qingping and Jushui Sites

Most of the measured data at the two sites were fitted well by the power law decay model (F = F0 × r–n), which was consistent with observations reported by Savage and Brodsky [14] from outcrop studies. Although the microfracture density in VTS2 in the footwall at the Jushui site has a lower fitting (R2 < 0.5; Figures 9(e) and 9(f)), the pattern still shows a gradual decrease away from the fault core. Considering that the deformational width of the fault zone also plays an important role in comparing the intensity of microfractures within the two sides of faults [12], we transform the functions into linear fitting models using logarithmic coordinates (Figures 13 and 14). However, determining the background microfracture density is difficult and uncertain due to the complex tectonic and lithological characteristics of the Longmen Shan thrust fault. Here, based on the measured data, we used 1 cm/cm2 and 1 #/cm2 as a simple index to define a dominant deformational zone width (DZW) for the two sites.

Specifically, the dominant DZWs for MLPA and MNPA within HTS, VTS1, and VTS2 in the hanging wall at the Qingping site are ~205, ~175, ~179, ~158, ~593, and ~524 m, whereas the corresponding values in the footwall are smaller at ~84, ~72, ~98, ~100, ~126, and ~173 m (Figure 13). For the dominant DZW for MLPA and MNPA within HTS, VTS1, and VTS2 in the hanging wall at the Jushui site, the values are ~72, ~43, ~87, ~69, ~106, and ~109 m; the corresponding values within the footwall are ~73, ~55, ~31, ~7, ~244, and ~104 m (Figure 14). To better compare the differences in microfracture density within the two sides of the two faults, we summarize all the data in Table 2. It is evident that the development of microfractures is more intense within the three planes in the hanging wall than in the footwall at the Qingping site, which shows an asymmetrical pattern of microfracture density across the fault. Moreover, the maximum MLPA, MNPA, and DZW from VTS2 appear to be greater than those from HTS and VTS1, which indicates that the development of microfractures might not be uniform on a three-dimensional scale. This difference could be explained by the fact that the VTS2 plane is mostly related to the maximum principal stress due to compressive deformation. However, the data at the Jushui site show a complex microfracture density across the fault. Specifically, the maximum MLPA and MNPA and dominant DZW from the HTS within the two sides of the fault appear to be similar. However, the data from VTS1 show greater values in the maximum MLPA and MNPA and a dominant DZW in the hanging wall than in the footwall. Although the linear fitting in VTS2 in the footwall at the Jushui site has some uncertainties in the dominant DZW, the maximum MLPA and MNPA are greater in the hanging wall than in the footwall. Therefore, the development of microfractures in vertical planes and horizontal planes at the Jushui site appears to be different. Vertical planes show an asymmetrical microfracture density pattern highlighting more intensely developed microfractures in the hanging wall than in the footwall, whereas horizontal planes are basically consistent in the microfracture density patterns. Moreover, the maximum MLPA, MNPA, and dominant DZW from VTS2 appear to be greater than those from HTS and VTS1, which is consistent with that revealed at the Qingping site.

Furthermore, the values of the maximum MLPA, MNPA, and dominant DZW at the Qingping site are much greater than those at the Jushui site (Table 2), which indicates that the intensity of microfractures is stronger at the Qingping site than at the Jushui site. This difference might be related to the maturity degrees of the fault, which indicates the development status of microfractures in fault damage zones [15]. For example, the scale and coseismic offsets were much greater for the YBF than for the JGF [26-28]. In addition, the fault at the Qingping site was exposed to the surface, whereas the fault at the Jushui site was not ruptured to the surface, indicating that the YBF is more mature than the JGF [20, 36].

5.2. Mechanisms of an Asymmetrical Pattern in Microfracture Density

Previous studies suggested that microfracture density in the fault damage zone might be influenced by the amount of slip across the fault, the size of the fault, lithology, rupture processes, and movement history [8, 13, 16]. The two sites show similar characteristics in lithology, rupture processes, and movement history. For example, the strata at the two sites mainly consist of limestone and sandstone (Figures 1(b) and 1(c)); the two faults were simultaneously ruptured during past paleoearthquakes [29] and have been deformed by thrust movement since at least the middle Miocene [36]. The other two factors, such as greater slip and fault size, might lead to more intensively developed microfracture density [13]. The Qingping site on the YBF has been more intensely deformed than that of the Jushui site on the JGF, including the coseismic offset and its scale, which is consistent with the MLPA, MNPA, and dominant DZW at the two sites (Table 2).

However, none of the aforementioned factors could well explain the asymmetrical pattern in microfracture density at the two sites. The amount of coseismic deformation could be considered a proxy to reflect microfracture density that is possibly related to long-term transient coseismic loading [1-3, 5, 6], which could help us understand the mechanism of the pattern in microfracture density. The modeling of coseismic deformation at the two sites shows a steep convergence trend in the hanging wall and a gentle convergence trend in the footwall (Figures 10 and 11), which is consistent with the asymmetrical decay curves of the measurements from thin sections at the two sites. The modeling also shows that the coseismic deformational amount at the Qingping site is much greater (approximately ten times on average) than that at the Jushui site. This indicates that the microfracture density at the Qingping site would be more intensely developed than that at the Jushui site, which is also evidenced by the maximum MLPA and MNPA and dominant DZW parameters compared with the two sites (Table 2). In other words, an asymmetrical microfracture density should be more evident at the Qingping site than at the Jushui site, which might explain why HTS at the Jushui site does not evidently show an asymmetrical pattern. However, the asymmetrical pattern in microfractures from VTS1 to VTS2 at the Jushui site provides us with another clue to further explore the development of microfractures. Although the fault is hidden at the Jushui site, the modeling (Figure 11) clearly shows that the potentially deformational interface above the buried point (with a depth between 0 and 2 km) of the fault is nearly vertically distributed, which is consistent with an estimation of distance from the fault for the sampling sites by field investigations. Moreover, due to the effect of thrust deformation at the fault tip, the rock mass buried within 0 and 2 km in the hanging wall successively accumulates positive compressional stress, which might in turn form locally distributed stress in the vertical direction [37]. This might result in the enhancement of the development of more microfractures within the vertical planes rather than the horizontal plane in the hanging wall (Figure 15). In other words, even if a hidden thrust fault is not exposed to the surface, its influence on the development of microfractures might be variable in vertical planes within the two sides of the fault but is likely to be much weaker than that of an exposed thrust fault. Furthermore, Oglesby et al. [17] used two-dimensional dynamic simulations of earthquakes on dipping faults and showed that shear energy could be trapped to form radiated waves in the hanging wall and amplify ground motion due to the asymmetric fault geometry. Wen et al. [38] summarized sixty-four strong ground motion records and found that the accelerations (Peak Ground Acceleration (PGA) and Peak Ground Velocity (PGV)) on the hanging wall were observed much larger than that on the footwall during the 2008 Wenchuan earthquake. In other words, dynamic seismic behavior might further enhance the asymmetrical effect of microfractures within the two sides of the faults. Therefore, compared with the other factors, we suggest that fault geometry might exert a first-order control on the asymmetrical microfracture density in the fault damage zone within two sides of the faults (Figure 15). This behavior is particularly evident for an exposed thrust fault such as the Qingping site, whereas a hidden thrust fault such as the Jushui site might show a quasiasymmetrical microfracture density due to its weaker deformation that was affected by the deeply buried fault structure.

5.3. Implications for Understanding the Hanging-Wall Effect for Evaluating Seismic Hazards on Active Thrust Faults

Previous studies have mostly suggested that microfracture density decreases gradually away from the fault [11, 12, 14, 15], which is consistent with our results. However, our results further show that the microfracture density in active thrust fault damage zones is more complex than previously thought and is characterized by an asymmetrical pattern due to fault geometry. In other words, microfractures in the hanging wall are likely to be more intense with a wider range than those in the footwall for active thrust faults. Moreover, the asymmetrical pattern for active thrust faults appears to be consistent with the seismic effect of the hanging wall, which highlights a larger ground seismic motion in the hanging wall than in the footwall, as evidenced from the worldwide strong motion dataset of earthquakes [39]. Furthermore, coseismic landslides on active thrust faults appear to be more focused on the hanging wall than on the footwall. For example, more and larger-scale coseismic landslides consistently occurred in the hanging wall of the seismogenic thrust faults associated with the 2008 Wenchuan Mw 7.9 earthquake [40, 41], the 1999 Chichi Mw 7.6 earthquake [42], and some earthquakes in Japan [43]. In addition, the 2022 Mengyuan M6.9 earthquake, associated with a strike-slip fault with a thrust component, also resulted in asymmetrical coseismic destruction of the tunnel across the fault [44]. All these seismic-related hazards, including coseismic motions, landslides, and rock mass deformation, show an asymmetrical pattern, which appears to be consistent with the asymmetrical microfracture density for active thrust faults. We suggest that more intensely developed microfractures contribute to a weaker rock mass strength and a higher flow permeability that is more likely to produce more severe geologic hazards [45]. If so, the risk of geological hazards in the hanging wall should be much higher than that in the footwall. Moreover, exposed thrust faults with larger offsets and greater scales can further dominantly enhance this potential risk, according to the comparisons between the Qingping and Jushui sites. The abovementioned behavior is helpful for understanding the mechanical characteristics of rock masses controlled by faulting, which can guide us to better assess geologic hazards related to active thrust fault deformation.

Revealing the characteristics of microfracture density in fault damage zones can help us better understand the deformational behavior of active faults and the mitigation of seismic hazards. We measured microfractures at the Qingping and Jushui sites across the seismogenic thrust faults associated with the 2008 Wenchuan Mw 7.9 earthquake in the Longmen Shan along the eastern margin of the Tibetan Plateau. The results showed that microfractures were more intense in the hanging wall than in the footwall at the two sites. Moreover, the intensity of microfractures, including length and number per area and dominant DZW, is greater at the Qingping site than at the Jushui site. These characteristics are consistent with the coseismic deformational modeling, which indicates that fault geometry might exert first-order control of the asymmetrical microfracture density pattern for active thrust faults.

This work was financially supported by the 2nd Tibetan Plateau Scientific Expedition and Research (2019QZKK0901) and the National Science Foundation of China (Grant no. 42372241, 42072244, and 41672207). Great thanks to Shikuo Chen for his insightful discussion.

The authors declare no conflict of interest.

Datasets that present all the raw photos and interpreted microfractures at Qingping and Jushui sites are available at https://doi.org/10.5281/zenodo.7672495.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0).

Supplementary data