Spatial Pattern of Late Quaternary Shortening Rate in the Longmen Shan Foreland, Eastern Margin of the Tibetan Plateau

Longmen Shan is located along the eastern margin of the Tibetan Plateau. In this belt, with the eastward expansion of the plateau, strong oblique shortening happens between the Songpan-Ganzi terrane and Yangtze Craton (i.e., the Sichuan Basin in China) [1, 2]. The dominant mechanism of developing such high topography in the Longmen Shan has been debated continuously. More than six tectonic models that are capable of explaining highly fragmentary geologic and geophysical observations in the thrust belt have been proposed [3, 4]. Some models emphasized the fault-controlled deformation, such as the upper-crustal shortening with a serious of detachments [5–7], the simple-shear or pure-shear shortening and thickening of the lithosphere [3, 4, 8], and the reactivation of preexisting weak zones in the heterogeneous crust [9, 10]. Other models emphasized the large-scale deformation of the crust, such as the channel flow in the mid and lower crust [11, 12], the indentation of the Yangtze Craton [13], and the westward under thrusting of the Yangtze Craton accommodated by distributed shortening in the whole plateau [14]. Among them, the models of upper-crustal shortening and channel flow are treated as end-member models to reveal the fundamental orogenic dynamics. To assess these models, exhumation history [15], tectonic kinematics [10], and deep structure observations [16] have been further studied. Moreover, the Quaternary shortening rate of the Longmen Shan is also significant to the different tectonic models, and the surficial deformation pattern can provide a direct approach to the driving mechanisms. Take the two end-member models, for example; a low crustal shortening rate corresponds to the channel flow model. On the contrast, the upper-crustal shortening model requires a considerable shortening rate. Therefore, this article focused on the Late Quaternary crustal shortening rate and shortening pattern of the foreland area in the southern Longmen Shan.

After the 2008 Wenchuan earthquake (⁠$MS=8.0$⁠, China Earthquake Networks Center) occurred in the northern range of the Longmen Shan, coseismic data and surface ruptures revealed a distinctive structural pattern, comprised of three primary active faults, including the Wenchuan-Maowen fault, Beichuan-Yingxiu fault, and the Jianyou-Guanxian fault [17, 18]. However, when stretching southward, the pattern becomes a broad thrust zone comprised of six faults, including the Zhonggang fault, Gengda-Longdong fault, Yanjing-Wulong fault, Xiaoguanzi fault, Shuangshi-Dachuan fault, and the Xinkaidian fault (Figure 1). Even during the following 2013 Lushan earthquake (⁠$MS=7.0$⁠, China Earthquake Networks Center), which occurred in the southern range, there were no surface ruptures that could reasonably indicate the overall configuration (Figure 1) [19]. A fold-generated foreland basin is a distinctive feature of the southern range of the Longmen Shan (Figure 1). Unlike the flat Chengdu Plain in front of the northern Longmen Shan, there are two anticlines in the southern section between the Longmen Shan and the boundary of the Cenozoic foreland basin (i.e., the Longquanshan). The tectonic activity of Xiongpo and Sansuchang anticlines is not well constrained [20, 21]. This is mainly because of lush vegetation and intensive weathering.

Tectonic fluvial terraces are substantial features because they can reflect characteristics of mountain uplift and surface deformation [25, 26]. By means of detailed chronology and measurement data, the Quaternary crustal deformation pattern and shortening rate can be obtained through longitudinal profile of river terrace [27–29]. The Qingyi River cuts across the foreland basin of the southern Longmen Shan in a southeastern trend and developed six-level terraces (Figure 1). To avoid elevation uncertainties, a high-precision mobile global positioning satellite (GPS) station with net-based, real-time kinematics (RTKs) was used in the field measurement, its horizontal precision was ±5 cm, and its vertical precision was ±8 cm. A longitudinal profile of Qingyi River terraces was established, which showed Late Quaternary deformation pattern of the foreland area in southern Longmen Shan. Combined with detailed chronology data, including ¹⁴C, OSL, and ESR, the regional crustal shortening rate was further calculated. Finally, a kinetic model was established to estimate the deformation pattern and activity distribution of the southern Longmen Shan. In addition, the along-strike variation of the Longmen Shan was further discussed. The main contribution of this work is constraining the Late Quaternary shortening rate of the foreland in southern Longmen Shan.

The Longmen Shan has gradually become the eastern most boundary of the Tibetan Plateau since the collision of the Indian Plate and the Eurasian Plate in 50-60 Ma (Figure 1). The Longmen Shan thrust belt is known for its steep topography and frequent geological disasters [30, 31]. Several studies have been conducted on parts of the mountain belt to show its Quaternary activity [32]. The tectonic of foreland region has been proved to be active during the Quaternary period [21, 33, 34]. Studies on the deep tectonics of the region have indicated that the foreland topography is dominated by thin-skinned structures [33, 35, 36]. Seismic profiles show that branches of the frontal thrust fault have extended into the foreland basin, which suggests the potential for large earthquakes. It can also be seen from the history of large earthquakes between 780 B.C.E. and 2020 C.E. (⁠$MS≥5.0$⁠) that the hypocenters in front of the southern Longmen Shan correspond to the tectonic pattern (Figure 1). In this study, we aim to determine the Late Quaternary activity of the foreland basin in southern Longmen Shan.

As the Cenozoic foreland basin, the Chengdu Plain is located in front of the Longmen Shan. To the east, the Longquanshan anticline is the eastern margin of the range–foreland structure (Figure 1). The northern and southern segments of the active Quaternary structure of the Longmen Shan differ in both their ranges and foreland regions. For instance, in the northern segment, there are three main active thrust faults; from NW to SE, these are Wenchuan-Maowen fault (WMF), Yingxiu-Beichuan fault (BYF), and Guanxian-Anxian fault (JGF, Figure 1) [37]. However, as the three faults stretch southwestward, they grade into at least six active faults, which exhibit minimal evidence of fault activity (Figure 1). Additionally, the detachment fault of the southern Longmen Shan has further extended into the foreland, generating multiple folds, namely, the Xiongpo anticline, the Sansuchang anticline, and the Longquanshan anticline (Figure 2), as well as multiple active faults, including the Datang fault (DTF) and the Pujiang-Xinjin fault (PXF; Figure 2). The Xiongpo anticline was estimated to have experienced an average shortening rate of 0.2 mm/yr since 40–25 Ma. The Xiongpo anticline acted as the deformation front of the Longmen Shan thrust belt between 40 and 25 Ma and 15 and 10 Ma driven by the eastward growth of the Tibetan Plateau [38]. The Datang fault includes a series of steep, east-dipping reverse faults, which roots in a 5–6 km deep basal detachment. It is a younger fault than the northeast-southwest-trending structures in the region. The Pleistocene slip rate on the Datang fault was defined as 0.4-1.2 mm/yr [39].

The southern foreland region has good continuity in its Mesozoic strata, but Cenozoic strata are not well preserved. More specifically, Paleogene sediments are distributed in the north of the Xiongpo anticline only, and Neogene sediments are only distributed on several small massifs along the frontal Longmen Shan (Figures 2 and 3). Additionally, instead of Holocene alluvial sediments, two ancient, abandoned alluvial fans (Q_p) are widely developed and well preserved around the Xiongpo and Sansuchang anticlines (Figure 2(a)). They are also the evidence of strong uplifting of the foreland during the Quaternary period. The GPS velocity field, with a reference of Eurasia, shows the kinematics of the southern Longmen Shan and its foreland (Figure 1) [24]. It can be seen that the foreland region seems to have a similar velocity as the southern Longmen Shan.

A corresponding swath profile was plotted to analyze average topography along the Qingyi River (Figure 3(a), with extent marked in Figure 1). Applied source data was the DEM generated from digital map in the scale of 1 : 50000. Moreover, we drew a geological cross-section and amended it by field surveying to understand the relationship between the surface geology and regional structures (Figure 3). Note that the swath profile (Figures 1 and 3(a)), geological section (Figure 3(b); profile S-S$′$ in Figure 2), and the seismic profiles (Figure 3(c); profiles A-A$′$ and B-B$′$ in Figure 1) were not strictly corresponded. We selected the position of geological profile to reveal more stratum information. We considered the diver position and adjusted them to each other in the final analysis. In view of the topographic features, as the boundary of the foreland basin, the Longquanshan is the longest anticline in front of the Longmen Shan thrust belt, which stretches for ~200 km (Figure 1). However, in the southern foreland, the Xiongpo anticline is more prominent than the Longquanshan, which can be clearly seen from a swath profile (Figure 3(a)). Additionally, the Sansuchang anticline is relatively low, and its core has been strongly eroded, showing relatively weak activity. The deformation of river terrace shows a trend similar to that of the topography.

We investigated the deformation of Qingyi River terraces to reveal the activity of the southern foreland basin. The Qingyi River is the largest river in the southern Longmen Shan. It flows through the range from the county of Baoxing upstream to Ya’an City downstream (Figure 2(a)). Finally, it reaches the trunk of the Yangtze River in Leshan City, along with the Min River and Dadu River (Figure 2(a)). Accordingly, the Qingyi River is a reasonable feature for studying the Quaternary activity of the foreland area in southern Longmen Shan.

Using the deformation of terraces to interpret active tectonics, precision of longitudinal profile is one of the most important factors. Under the conditions of kilometer-scale surveying, RTKs can provide accurate data and have been widely applied. RTK surveying is usually based on Bluetooth or wireless network technologies, which have a distance limit of ~5 km. However, in the measurement of the 150 km long Qingyi River, traditional RTKs cannot provide continuous and accurate mapping results. To solve this problem, we used a net-based RTK approach, using the Sichuan Global Navigation Satellite System (SCGNSS). The SCGNSS was constructed by the Chinese government following the Wenchuan M_S 8.0 earthquake in 2008 in order to replenish the destroyed surveying facilities. It is now composed of 60 satellite stations (Figure 2(b)), which can provide real-time modifying functions to the survey work across the eastern margin of the Tibetan Plateau. Additionally, there are three stations in the research area: Qionglai station, Mingshan station, and Leshan station (Figure 2(b)). Thus, we can achieve an altitudinal survey precision of ±8 cm with the SCGNSS across the entire Qingyi River basin. During the study, we measured all terrace surfaces and their straths along the Qingyi River. The terraces in different levels were investigated in the field. Every measured terrace was estimated to be originally developed. The overlap layers including flood sediments, depositions of barrier lake, or sediments formed by human activities were all excluded. All data were measured under the coordinate system of the UTM Northern Hemisphere 48. Then, the data were projected along a 120-degree axis vertical to regional structures in contrast with topography and seismic profiles. The measured points were marked on the geological map (Figure 2).

The sequence of river terraces is the base of establishing the longitudinal profiles. In order to define the sequence of river terraces, we initially classified the terraces according to the heights above the riverbed and the marker layer. However, because of the heavy weathering caused by the humid climate, the terraces developed in the study area are fragmentary. We encountered difficulties in accurately distinguishing among terraces in multiple levels. Meanwhile, during the field survey, we found that the matrix of terrace sediments obviously varied from one another. The weathering of old terraces seemed to be more serious than the young ones, which may be related with the climate. Overall, the different weathering can help to classify the terraces in different ages. Therefore, we applied bulk element testing to help accounting for the degrees of weathering of different-level terraces. Sedimentary samples belonging to each terrace were collected, and terraces with thick, silty, top deposits were excluded, as these may be young overbank deposits. To avoid excessive erosion, the samples were collected at a depth of ~2 m beneath the terrace surface. Bulk and trace element analyses were then applied to reveal the degree of weathering. Testing was completed at the Key Laboratory of Crustal Dynamics of the China Earthquake Administration. The classification of terraces is a detailed process of fundamental work. Hence, the results and analyses of bulk and trace elements were shown in Supplemental material E.

The age of the terraces studied varied from thousands of years to the Middle Pleistocene. Therefore, multiple dating methods were used. Carbon isotope dating (¹⁴C) was applied to terrace T1 (youngest); optical stimulated luminescence dating (OSL) was applied to terraces T1, T2, and T3, and electron spin resonance dating (ESR) was used for terraces T4, T5, and T6 (oldest). The samples were collected in the sand or clay layer close to the top fluvial gravels, as we attempted to approach the age of the morphological surface of the terrace. Totally, four samples of terrace T1 (including two ¹⁴C and two OSL samples), six samples of terrace T2 (OSL samples), two samples of terraces T3 (OSL samples), three samples of T4 (ESR samples), one sample of T5 (ESR sample), and one sample of T6 (ESR sample) were collected and dated in the laboratory. Carbon isotope samples were dated at the School of Archaeology and Museology, Peking University; OSL samples were dated at the Key Laboratory of Crustal Dynamics, China Earthquake Administration, and ESR samples were dated at the State Key Laboratory of Earthquake Dynamics, China Earthquake Administration.

To calculate the regional crustal shortening rate, a model with a clear tectonic structure must first be established. We combined the surficial terrace deformation with the deep structures to interpret tectonic pattern and calculate regional crustal shortening rate. The longitudinal profile of terraces’ elevations was transformed into terraces’ heights above the riverbed to show deformation in detail and correlate to deep structures. We applied the Monte Carlo simulation to estimate uncertainties of the increments in evaluating the terrace ages and calculating of crustal shortening rates. In our simulations, the probability distributions of each input parameter are sampled randomly for 50000 trials to generate a frequency distribution. Reported values reflect the mode of the output histogram and the associated 95% confidence intervals. Input parameters contain the excess area, detachment depth, and terrace age. Output parameters contain the total shortening length and shortening rate. Details of the Monte Carlo simulation are shown in Supplemental material D.

In the foreland region of the Longmen Shan, six terraces are preserved formed during different stages of the Quaternary period along the Qingyi River; in age-ascending order, they are T1, T2, T3, T4, T5, and T6. Among them, terraces T1–T4 are distributed along the entire length of the river. Terraces T5 and T6 are only developed around the Ya’an region. Here, we take the Ya’an Valley (i.e., Ya’an City) and Hongya Valley (i.e., Hongya Town) as examples to introduce the terrace details.

In the Ya’an Valley, the heights of the leading edges of the terraces above the modern riverbed are ~4.9–8.5 m for T1, ~15.2–18.6 m for T2, ~48.0–59.9 m for T3, ~88.1–92.1 m for T4, ~130.1–150.8 m for T5, and~204.3–209.1 m for T6 (Figures 4(a) and 4(b)). Terrace T6 is preserved only at the top of Longgang Mountain, southwest of Ya’an City (Figure 4(a), profile 6). The visible sedimentary strata area is 5.6 m thick, and the thickness of the gravels above the bedrock strath is 11.2 m (Figure 4(c)). The terrace sediments are comprised of quartzite and mudstone gravels, with diameters ranging from 8 to 35 cm. The matrix between the gravels is clay. Terrace T5 is preserved along the Qingyi River between Ya’an City and the town of Hongya, mainly north of the Xiongpo anticline (Figure 4(a), profile 5). The thickness of the sediments in profile is 11.5 m (Figure 4(c)). Gravels in the outcrop are comprised of quartzite, sandstone, and mudstone, with a small amount of magmatic rocks. The matrix between the gravels has been weathered into clays (Figure 4(c)). Terrace T4 is mainly developed on the piedmont, south of the Ya’an Valley (Figure 4(a), profile 4). In profile 4, a 3.8 m thick brownish red sandy clay layer mixed with gravel can be seen (Figure 4(d)). The gravels are mainly comprised of sandstone and lesser amounts of quartzite, and the matrix is weathered to a brownish red clay (Figure 4(d)). Terrace T3 is distributed mainly from Ya’an City to the town of Jiajiang (Figure 4(a), profile3). In profile 3 (Figure 4(d)), the gravel layer is 18.5 m thick. The matrix between the gravels is brown, clayey sand (Figure 4(d)). Terraces T2 and T1 are distributed along the entire Qingyi River Valley. In profile 2 (Figure 4), the matrix is comprised of reddish brown sand and silt. Additionally, the gravels in T1 and T2 are barely weathered, and the matrix holds the original structure, with coarse particle sizes.

In the Hongya Valley, the heights of the leading edges of the terraces above the modern riverbed are ~6.5–8.6 m for T1, ~21.4–25.6 m for T2, ~73.9–81.5 m for T3, and ~127.2–132.3 m for T4 (Figures 5(a) and 5(b)). Terraces T1-T4 in the Hongya Valley are all basin-wide distributed. Terrace T4 is distributed around the highest hills along the river. The thickness of the gravels above the bedrock strath of T4 is 20.5 m. The sediments are comprised of quartzite and mudstone gravels, with diameters ranging from 10 to 30 cm. The matrix between the gravels is clay (Figure 5(c)). Terrace T3 is distributed more limited, and its gravel layer is 16.5 m thick. Gravels in the outcrop are comprised of quartzite, sandstone, and mudstone. The matrix between the gravels is brown, clayey sand (Figure 5(c)). Terraces T2 and T1 are distributed along the entire Hongya Valley. Gravels in the outcrop are mainly comprised of sandstone and mudstone. The matrix is comprised of reddish brown sand and silt (Figure 5(c)).

Terraces of Qingyi River are discontinuously distributed in the study area. Hence, we acquired the altitudes of riverbed and terraces T1–T6 via mobile surveying. All the riverbed and terrace points were measured individually with a mobile GPS based on network RTKs (see Section 3.1). Considering that there were many villages on the terraces, we tried to measure the initial terrace surfaces and corresponding straths. To reduce the influence of river bends, all the measurement data were projected to the tectonic line P-P$′$⁠, which is vertical to the strikes of the Longmen Shan and foreland anticlines (Figures 2 and 6(b)). In view of the plotted longitudinal profile (Figure 6(a)), the deformation of Xiongpo anticline was reflected by terraces in all levels. It can be seen that the basin-wide terraces T1–T4 show distortions to differing extents around the anticlines. Terraces T6 and T5 have been seriously eroded, but in view of the trend of T5’s surface, the Xiongpo anticline may have had the same activity before the period of T4, which promoted the damage of high terraces around the anticline (Figure 6(a)). Near the Xiongpo and Longquanshan anticlines, a new local terrace, T1b, has developed because of recent uplift, but T1b is absent around the Sansuchang anticline (Figure 6(a)). Around the Longquanshan area, terraces T4 and T3 were all eroded, and only terraces T2 and T1 can reflect tectonic activities (Figure 6(b)). The original RTK measurement data are listed in the Supplemental material F.

Multiple dating methods were employed to test terrace samples, which are distributed in the town of Lushan, upstream of the Qingyi River, Ya’an City in the midstream, and the town of Hongya downstream (Figure 7). Carbon isotope dating (¹⁴C) was applied to terrace T1; optical stimulated luminescence dating (OSL) was applied to terraces T1, T2, and T3, and electron spin resonance dating (ESR) was used for terraces T4, T5, and T6. To approach the age of the morphological surface of the terrace, the samples were collected in the sand or clay layer close to the top fluvial gravels as far as possible. All chronology samples were collected in newly excavated outcrops to avoid of pollution. The details of the outcrops that used to collect dating samples are listed in supplemental material A.

Original dating results are listed in Table 1, and the radiocarbon results were calibrated with the software package OxCal [41]. The detailed calibrated results are shown in Supplemental material B.

Detailed chemical data of the samples are shown in Table 2, including the contents of U, Th, and K, the testing grain size, environmental dose, and the equivalent dose of the samples. The simplified multiple aliquot regenerative (SMAR) dose protocol [42] was applied. Additionally, the OSL signals and sensitivity-corrected OSL of each sample are shown in Supplemental material C.

The chemical data during sample testing are listed in Table 3.

The revised test results are listed in Figure 8 and Table 4. Through the results, it can be seen that the terraces are approximately isochronous (Figure 8 and Table 4), especially terraces T1, T2, T3, and T4. Although the Qingyi River is reported to have migrated for several times [43], the river channel should has not changed since the development of terrace T4. Accordingly, it should be reasonable to use the deformation of terraces at the same level to solve tectonic problems. Notably, among the ages of terraces T2 and T3, ages of the terraces in the mountain area of Ya’an and Lushan differ from that in the campaign area. Through the field survey, we consider that the formation of terrace in the mountain area is steady and continuous. The main misleading factor is the overlap layers caused by large floods, which may lead to excessively young ages. In the downstream area, the river channel sways within a specific range and leads to complex depositions. So, we decided to exclude excessive age results of terrace T2 from the calculation of the rate of shortening. As to terrace T3, we did not get sufficient results to evaluate distribution characteristic of the ages. Compared with the ages of terraces T2 and T4, we consider that the result of $73.1±9.7$ ka was extremely young and was not applied in the final calculation of shortening rates. In the evaluation of the terrace ages, we applied the Monte Carlo simulation method to avoid the increase of error. In the simulations, the probability distributions of each ages were sampled randomly for 50000 trials to generate a frequency distribution.

To calculate the regional crustal shortening rate, a model with a clear tectonic structure must first be established. We combined the surficial terrace deformation with the deep structures to interpret tectonic pattern and calculate regional crustal shortening rate (Figure 9). The longitudinal profile of terraces’ elevations was transformed into terraces’ heights above the riverbed to show deformation in detail (Figure 9). It can be seen that the deformation pattern of river terrace is influenced by anticline uplifting (Figure 8(b)), especially those of the Xiongpo and Longquanshan anticlines. In addition, there is a deficiency of Paleogene strata between these anticlines, which are locally preserved in front of the Longmen Shan (Figure 2). This may indicate continuous activity in the foreland region. According to seismic profiles (Figure 9(b)), previous studies have proposed that two imbricated, blind thrust faults are developed beneath the front range of the Longmen Shan [35, 36, 40], which has extended into the Sichuan Basin and would affect the densely populated area in the basin (Figure 9(b)). In this study, we also found that the anticlines, which are generated by the upper blind thrust fault, were active during the Late Quaternary (Figure 9(b)).

Deformation of strath surfaces has been recommended to calculate excess area rather than the deposit surface in rapidly uplifting areas, such as the Himalayas [26]. In such areas, both the amount of river incision and river terrace deformation are accounted for in shortening rate calculations because uplifting and tilting are considered to be the dominant patterns of tectonic activity. Additionally, because of the narrow valley and floods, thick silty overlying strata, interpreted as overbank deposits, appeared frequently. Sometimes, these silty deposits could be over 10 m thick. Consequently, terrace surfaces no longer preserved their original heights. Meanwhile, when dealing with fault displacement or fold deformation, well-preserved terrace surfaces were also considered in analyses of deformation [52]. In the Qingyi River basin, terrace straths were not exposed continuously along the river on account of heavy erosion and dense vegetation. A majority of observed straths exhibited lower elevations because of the covering of eroded gravel and vegetation. Meanwhile, the river valleys in the foreland area are so broad that terraces have rarely been influenced by overbank floods. The thickness of the gravel deposits was found to be constant in multiple segments along the river (Table 5). The measured terraces had integrated and dualistic fluvial structures and were determined to have not been reconstructed or flooded. Hence, we used the measured terrace surfaces to estimate the excess area of each terraces in different levels. The excess area was deduced by subtracting the conceivable original riverbed from the area of the deformed terraces (Table 6). As terrace T1b is developed only in the core region of the Xiongpo and Longquanshan anticlines (Figure 9(a)), it was not considered in the excess area calculations.

According to seismic reflection data, the three anticlines in the foreland region are derived from a shallow detachment with a depth of 3–4 km [9, 53]. However, in the area of the Xiongpo and Sansuchang anticlines, this detachment layer has probably involved into deformation (Figures 3(c) and 3(d)). In this regard, large uncertainties might arise from the estimation when using single-bed excess area. According to seismic reflection data across the whole Sichuan Basin, there are three detachment levels, which define the base of the thrust sheets in the foreland of southern Longmen Shan: a shallow detachment 3–5 km below the surface, an intermediate detachment 5–7 km below the surface, and a deep detachment approximately 15 km deep [5, 54]. Among them, the shallow one controls the current deformation of surficial anticlines, the middle and deep two detachments probably extend to the eastern mountain belt in Sichuan Basin (Figure 2(c)). In the foothill and foreland regions, referring to regional seismic profiles (Figure 3(d)), the middle detachment owns a depth of 7 to 10 km and is not deformed [9]. Consequently, we took the shallow detachment (3–4 km) as the minimum value, corresponding to the maximum crustal shortening rate, and the middle detachment (7–10 km) as the maximum value, corresponding to the minimum crustal shortening rate. Against each anticline, the depth of the two detachments was interpreted individually. We used the two detachments to estimate crustal shortening rates of Xiongpo and Sansuchang anticlines and the single shallow detachment to estimate the Longquanshan anticline.

The local terraces, T5 and T6, are too restricted to reflect tectonic activity. Hence, terraces T1-T4 in the Xiongpo anticline, T3 in the Sansuchang anticline, and T1-T2 in the Longquanshan anticline were used to estimate the shortening rate. Longitudinal profile of terrace’s height above the riverbed was integrated along the individual curve to get the excess area (Table 6). The shallow detachment depth was interpreted to be 3.7–4.0 km in the Xiongpo anticline, 3.2–3.8 km in the Sansuchang anticline, and 3.4–3.8 in the Longquanshan anticline. The middle detachment depth was interpreted to be 7.0–7.9 km in the Xiongpo anticline and 5.5–5.9 km in the Sansuchang anticline (Table 6). We assess that the shallow detachment in the Longquanshan anticline is not involved into deformation; hence, a single detachment model was performed.

We applied the Monte Carlo simulation to estimate uncertainties of the increments in the calculating process of chronological data and shortening rate. Our simulations followed the approach of studies that focused on typical fold-related models [27–29]. In our simulations, the probability distributions of each input parameter are sampled randomly for 50000 trials to generate a frequency distribution. Reported values reflect the mode of the output histogram and the associated 95% confidence intervals. Input parameters contain the excess area, detachment depth, and terrace age. Output parameters contain the total shortening length and shortening rate. Details of the Monte Carlo simulation are shown in Supplemental material D.

Seven terraces, including terrace T1–T4 in Xiongpo anticline, terrace T3 in Sansuchang anticline, and terraces T1–T2 in Longquanshan anticline, were processed by Monte Carlo simulation for 50000 trials (Figure 10). Then, the overall data were plotted in the summarizing graph to show their characteristics. The Xiongpo and Sansuchang anticline were estimated according to a model with two detachments, and the estimating results were shown as irregular extents in the graph (Figure 10). Moreover, with regard to the Xiongpo and Longquanshan anticline, the trends of the shortening rates that obtained from different level of terraces seem to be continuous and stable (Figure 10). A linear fitting was applied to evaluate the average shortening rate. Consequently, the results showed that the evaluated crustal shortening rates of the Xiongpo anticline, Sansuchang anticline, and Longquanshan anticline are $1.01±0.05$ mm/yr, $0.16±0.02$ mm/yr, and $0.89±0.07$ mm/yr, respectively. We propose that the total crustal shortening rate of the Longmen Shan foreland area is $2.06±0.14$ mm/yr.

From the result of total shortening rate, it can be seen that the Xiongpo anticline accounts for 48%, the Sansuchang anticline accounts for 9%, and the Longquanshan anticline accounts for 43%. The Late Quaternary shortening rate of the southern Longquanshan anticline (0.89 mm/yr) is considerably smaller than that of the northern Longquanshan anticline, which was reported to be 1.47 mm/yr [21]. However, the Xiongpo anticline possesses the largest shortening rate in the foreland area of southern Longmen Shan, which is corresponding to the topography. The extremely small shortening rate of the Sansuchang anticline indicates that the shortening of the foreland is not distributed in-sequence. Considering the serious surficial erosion and low terrain, the activity of the Sansuchang anticline has probably transformed to the Xiongpo anticline. Furthermore, taking the shortening rate of the front range in the southern Longmen Shan (1.14 mm/yr) into consideration [32], the overall shortening rate that controlled by the detachment system is assessed to be 3.18 mm/yr, which is 54% higher than that of the northern Longmen Shan.

The SW segment of the Longmen Shan seems to be in a rapidly shortening period, especially the activity of detachments, which has expanded into the foreland basin and sustained strong activity in the Late Quaternary period. We proposed that the 2013 Lushan earthquake (⁠$MS=7.0$⁠) may rupture the middle detachment, which has extended into the foreland basin and controlled the surficial deformation (Figure 11(c)). The ramp of the detachment probably acted as a locking structure, which was proved to have been uplifting the surface constantly during the Late Quaternary [32]. Seismic reflection and earthquake relocation data further revealed that a back thrust vertical to the ramp of detachment also ruptured in the Lushan earthquake [55–58]. Overall, a kinetic model shows the tectonic shortening pattern of the southern Longmen Shan: the southern Longmen Shan seems to be an actively reversing edge of the plateau, and the foreland area has absorbed the majority of the shortening (Figure 11(c)). The Late Quaternary shortening rate of the detachment system (3.18 mm/yr) is considerable for the mountain building and foreland expansion. Therefore, our work suggests that, compared to the effect of channel flow (Figure 11(b)), crustal shortening played a leading role in the current uplift of Longmen Shan (Figure 11(a)).

The along-strike variation is a particular feature of the Longmen Shan. The northern Longmen Shan exhibits a distinctive structural pattern that comprised of three primary active faults. However, the southern segment possesses a broad thrust zone with multiple fault branches and a fold-generated foreland basin (Figure 12). We processed the focal mechanism data to understand the converging characteristics of the belt. Beach balls showing the focal mechanisms of earthquakes (⁠$MS≥4.5$⁠) from 2000 to 2021 were projected on the map (Figure 12). The focal mechanism data were from the Global CMT Project [61]. Each beach ball can be treated as a stereographic projection of the strain ellipse due to the deformation associated with an earthquake. Solid quadrants show orientations of extensional strain, and open quadrants show orientations of compressional strain. In the Longmen Shan, the focal mechanisms in the southern segment can clearly be seen to differ from those in the northern segment (Figure 12). In the northern Longmen Shan, strike-slip is concentrated on the middle region of the belt, as well as the backland; in the frontal region, convergence shows the reversing characteristic that may be generated by a blind fault [9]. In contrast, in the SW, overall convergence is orthogonal to the regional strike and is achieved purely by thrusting.

Previous studies suggested that the reactivation of Triassic structures played an important role in the complex evolution of the Longmen Shan [10]. A “maximum exhumation belt” was proposed to explain the along-strike variation of the Longmen Shan. This belt strikes from the southern Longmen Shan to western Min Shan and was ~15° anticlockwise to the overall Longmen Shan belt. The deviation between the two belts generates the along-strike variation: from south to north along the strike, the reverse slip component on faults, and the uplift rate in the hanging-wall blocks decrease, while the dextral strike-slip component and the corresponding slip rate increase. Our results further verified the deduction of this model. We found that the southern Longmen Shan has an expanding foreland structure, which is dominated by detachment system with a shortening rate of 3.2 mm/yr. On the contrast, the foreland shortening rate in the northern Longmen Shan was limited to 1.47 mm/yr [21], and a part of strain was probably assigned to the transpressional movement of the Beichuan-Yingxiu [62] and Qingchuan faults [63]. Another factor that cannot be ignored is that the upper crust may be decoupled from the middle and lower crust due to the existence of the detachments in the foreland area of the Longmen Shan. Although surface denudation and Moho gradient reflect the trend of plateau expansion, the surficial deformation may reflect a shallower tectonic pattern and a smaller time-scale deformation.

In this article, terrace deformation of the Qingyi River was quantitatively analyzed to estimate the Late Quaternary crustal shortening rate and deformation pattern. A kinetic model was established to explain the deformation pattern and activity distribution of the southern Longmen Shan. Based on our research, the mechanism of orogenic evolution and along-strike variation of the Longmen Shan was further discussed.

By systematic measurement of the terraces in six levels with a net-based RTK approach, we generated a longitudinal profile of Qingyi River across the foreland in the southern Longmen Shan. Among the terraces, terraces T5 and T6 were too restricted to reflect tectonic activity, and terraces T1-T4 in the Xiongpo anticline, T3 in the Sansuchang anticline, and T1-T2 in the Longquanshan anticline were used to estimate the shortening rate. Moreover, by the dating methods of ¹⁴C, OSL, and ESR, the chronology sequence of the river terraces was estimated. Namely, the age of terrace T6 is 300 ka; the age of T5 is 217 ka; the age of T4 is 178 ka, while that of T3 is 97 ka, T2 is 48 ka, and T1 is 7 ka.

Excess area method was applied to calculate shortening rate, and Monte Carlo simulation was used to estimate increments in the calculation. The results indicate that the overall Late Quaternary shortening rate of the foreland in the southern Longmen Shan is 2.06 mm/yr, in which the Xiongpo anticline accounts for 48% (1.01 mm/yr), the Longquanshan anticline accounts for 43% (0.89 mm/yr), and the Sansuchang anticline accounts for 9% (0.16 mm/yr). The Xiongpo anticline seems to be the most active structure in the foreland of the southern Longmen Shan, and the Sansuchang anticline has been declined during the Late Quaternary.

According to the shortening rate, we propose that the special fold-generated foreland of the southern Longmen Shan is an expanding segment of the thrust belt and possesses dispersive activities. The active folds are derived from a thin-skinned structure that was dominated by double detachments. The 2013 Lushan earthquake (⁠$MS=7.0$⁠) may be related to the ramp of the deeper detachment. In addition, the deformation pattern in this article specifies the along-strike kinematic variation of the Longmen Shan: the southern segment is under the compression of in the NW-SE trend and has a broad deformation belt, while the northern segment has a narrow belt and accommodates the transpressional strain to the dextral strike-slip movement.

BYF:

Beichuan-Yingxiu fault

DTF:

Datang fault

GLF:

Gengda-Longdong fault

JGF:

Jiangyou-Guanxian fault

LQSA:

Longquanshan anticline

PXF:

Pujiang-Xinjin fault

RTK:

Real-time kinematic

SSCA:

Sansuchang anticline

WMF:

Wenchuan-Maowen fault

XGZF:

Xiaoguanzi fault

XPA:

Xiongpo anticline

YWF:

Yanjing-Wulong fault.

The underlying data is available in the article tables and the appendant tables.

The authors declare that they have no conflicts of interest.

This project has been supported by the National Natural Science Foundation of China (41802226) and the research grants from the National Institute of Natural Hazards, Ministry of Emergency Management of China (grant number ZDJ 2018-03).