Quantifying the Pabu Normal Fault Scarp, Southern Tibetan Plateau: Insights into Regional Earthquake Risk

Abstract

The location of the main boundary fault of the Yadong-Gulu Rift (YGR) shifts from the east side in the southern segment to the west side in the northern segment. The Nyemo Graben Group (NGG) connects the southern and northern segments of the YGR and provides clues for understanding the migration of boundary fault locations along the YGR. However, the NGG has received very little attention. In this study, we map the geometry of the Pabu normal fault, which is the boundary fault of the westernmost graben in the NGG, using high-resolution remote sensing images. We then utilized a digital elevation model (DEM) with a spatial resolution of 1 m. Morphometric parameters such as scarp height, width, and slope were obtained from elevation profiles in three typical deformation regions. Our results reveal a fault segment approximately 3 km long that links the southern and northern segments of the Pabu Fault. Each fault segment could be a major segment. Furthermore, based on regional tectonic activity, the Pabu Fault has the potential to produce an earthquake with a magnitude of around M 6.7.

1. Introduction

A series of north–south trending rifts have developed on the Tibetan Plateau, related to the growth and evolution of the plateau [1], leading to the belief that there has been a change in the dynamic mechanisms of the plateau [2,3,4]. The Yadong–Gulu rift (YGR) is a significant tectonic feature among these rift systems [5,6,7]. The main boundary fault of the northern segment of the YGR is situated on the west side of the rift, whereas the main boundary fault of the southern segment is positioned on the east side of the rift. A series of small grabens named Nyemo Graben Group (NGG) [8] links the northern and southern segments of the YGR near Yarlung Zangpo River (Figure 1). Thus, the NGG is pivotal for understanding the migration of boundary fault locations along the YGR as it progresses from north to south. However, the NGG remains less explored compared to the northern and southern segments.

Among the grabens in the NGG, the Angong Graben is particularly notable for its significant seismic activity, which has been the focus of extensive research [6,7,8,11]. Historical and contemporary seismic records reveal four earthquakes with magnitudes of six or higher along the NGG, including the 1264 Chubutse and 1901 Nyemo earthquakes, both registering magnitudes of 6 ¾ [10]. These events are thought to be influenced by the Yangyi or Angong grabens [7,11]. More recent significant earthquakes include the 1992 M_S 6.3 event and the 2008 Yangyi M_S 6.6 event, both occurring in the eastern part of the NGG. In contrast, research on the grabens in the western part of the NGG remains relatively sparse. This lack of information impedes a comprehensive understanding of the region’s tectonic activity and the assessment of future earthquake hazards.

The Pabu normal fault is the boundary fault of the westernmost graben in the NGG (Figure 2). In this study, we map and describe the geometry of the Pabu normal fault. We then utilize a high-resolution digital elevation model (DEM), derived from satellite images, to detect changes in scarp height, which serves as a proxy for vertical displacement. We create their displacement profiles at significant structural deformation regions along the normal fault to further infer fault structural segmentation and the potential existence of secondary linkages. Finally, we assess the potential regional earthquake risk by integrating data from other grabens within the NGG.

2. Geological Setting

The YGR is divided into northern, central, and southern segments, running from north to south [6]. The northern segment, extending from Gulu to Yangbajain, is bounded by NE-striking, E–SE-dipping normal faults [5,12]. In contrast, the southern segment of the YGR is bounded by W-dipping normal faults located between the Yarlung–Zangbo Suture (YZS) and Himalayan (Figure 1). In the region of Nyemo County, the fault splits into multiple branches, creating six smaller, elongated valleys or grabens that run in a north–south direction. These valleys are collectively known as the Nyemo Graben Group. The southern segment of the Yadong–Gulu Rift (YGR) exhibits extension rates of approximately 0.8 ± 0.3 mm/yr during the late Quaternary period [13]. In contrast, extension rates near the Gulu graben at the northern end of the YGR increase significantly to between 3 and 6 mm/yr [12]. Overall, the YGR demonstrates a vertical slip rate ranging from 0.5 to 2.0 mm/yr, with the exception of the northern Gulu rift, which is influenced by the Beng Co dextral fault [5,12]. Geodetic surveys reveal a present-day extension rate of approximately 2 to 10 mm/yr [14,15,16,17]. This variation in extension rates suggests that while different segments of the YGR exhibit similar tectonic characteristics and late Quaternary activity, there are significant regional differences in extension rates across the rift.

Located in the central portion of the Yadong–Gulu Rift and intersecting the Yarlung Tsangpo River, the NGG comprises six relatively smaller half-grabens: the Pabu Graben, Pangang Graben, Angong Graben, Nyemo Graben, Yangyi Graben, and Panton Graben (Figure 1). Among these, the Angong Graben is the most tectonically active. Generally exhibiting a north–south orientation, the NGG trends north–south and extends approximately 60 km in length and about 40 km in width, with a slight dispersion towards the south. Individual grabens within the group vary in width from approximately 1 to 5 km and in length from about 10 to 28 km

The Pabu Graben, a half-graben situated at the western edge of the NGG, extends approximately 25 km in length and 2–3 km in width (see Figure 2). The Pabu normal fault forms the western boundary of the graben and trends roughly northeast–southwest. Based on the fault trace geometry observed in the remote sensing image (Figure 2), three distinct segments can be identified. In the middle section of the Pabu fault, fault segment F2 links the northern segment F1 with the southern segment F3, and has a strike direction distinct from both. The Pabu fault is predominantly a normal fault with notable vertical displacements. Along this fault, various geomorphic surfaces formed during different geological periods, ranging from the late Pleistocene to the Holocene. These vertical displacements have led to the creation of fault scarps. Tributaries flow into the graben from the west, perpendicular to the boundary fault. Holocene alluvial sediments are found along these rivers, while proluvial sediments, dating from the Holocene to the Pleistocene, are primarily located on the graben’s western side (Figure 3).

3. Methods

Understanding the surface geometry and kinematics of active faults is essential for evaluating seismic hazards and analyzing structural impacts. In this study, we utilized satellite imagery to map the surface geometry of active faults along the Pabu graben. This approach yielded detailed information on fault geometry, kinematics, and their effects on landforms. To ensure the accuracy of our remote observations, we complemented these data with field surveys

The GaoFen-7 satellite is China’s first civilian satellite, designed for submeter stereo mapping [18]. It captures panchromatic stereoscopic images with a front-view resolution of 0.8 m, a rear-view resolution of 0.65 m, and multispectral images with a resolution of 2.6 m, all within a 20 km wide field of view. The digital elevation model (DEM) derived from GaoFen-7 has been successfully utilized in research on active tectonics [19] and glacial dynamics [20]. For this study, the GaoFen-7 DEM product (Figure 4a), with a spatial resolution of 1 m, was provided by the High-resolution Remote Sensing Data Center, China Earthquake Administration.

The scarp morphology was quantified using a semi-automated algorithm named SPARTA, based on the fundamental assumption that the fault scarp approximates the position of the fault [21]. The algorithm is designed to calculate the scarp height, width, and slope of a fault scarp from a scarp elevation profile. An elevation profile is positioned perpendicular to the local scarp trend. The precise location of the scarp within the profile is determined by identifying the scarp crest and base using the initial and final values of the scarp profile. These values must meet pre-established threshold values for the slope and the derivative of the slope. Subsequently, a linear regression (using the least-squares method) is applied to the upper and lower original surfaces located away from the scarp. In SPARTA, the scarp height (H) is defined as the difference in elevation between the regression lines at this point (Figure 4b). The slope (α) of the best-fit line through the fault scarp represents the scarp gradient, while the horizontal distance between the fault scarp crest and base is referred to as the scarp width (W). A detailed illustration of the SPARTA can be found in [21]. The three typical tectonic deformation regions were used to calculate the scarp height, width, and slope (Figure 4c).

4. Results

The Pabu fault is prominently visible in remote sensing images and separates the mountain from the graben, creating a distinct topographic difference (Figure 5a and Figure 6a). The river flows across the alluvial fan, forming terraces T1 and T2 along the fault trace (Figure 5a). Terrace T1, in particular, is offset by the Pabu fault, as evidenced by the fault scarp (Figure 5c). The Pabu fault cuts through the granite, resulting in observable fault striations and cataclasite (Figure 6c). Similar to the northern segment F1, fault segment F2 is clearly depicted on the DEM as fault scarps that disrupt various geomorphic surfaces (Figure 2). In contrast, the southern segment F3, is characterized by the discontinuous and pronounced fault scarps (Figure 6b), significantly different from F1 and F2.

Moving from the SW to the NE along the Pabu fault, three significant tectonic deformation regions are selected to analyze fault scarp morphology parameters, including height, width, and slope. Figure 7a illustrates a clear fault trace, except for a segment interrupted by a river, extending from SW to NE along the basin boundary. Figure 7b shows variations in scarp height along the fault trace. Initially, in the first thousand meters from the SW direction, the scarp height starts at a certain level and then increases to 50 m. Subsequently, it gradually decreases to 30 m and remains stable. In general, two bell-shaped profiles with slip maxima at their centers can be discerned. However, beyond the midpoint of this 1000 m segment, a break in scarp continuity occurs, with only a few high scarp height values recorded. Meanwhile, the scarp width shows subtle variations along the strike (Figure 7c). In terms of scarp slope, it displays peak–valley variations, initially decreasing before eventually increasing (Figure 7d).

Figure 8a highlights the prominent line features of tectonic geomorphology resulting from fault activity. In a similar fashion to Figure 7b, Figure 8b reveals several breaks in scarp height along certain segments of the fault trace. At a distance of 750 to 1000 m along the fault trace, the curve displays a bell shape. Furthermore, another similar pattern is observed between 1700 and 1900 m. Meanwhile, scarp width values exhibit relatively more variations along the strike (Figure 8c). In contrast, scarp slope values demonstrate fluctuations, with distinct peaks and troughs (Figure 8d). Figure 8a clearly shows the fault triangle plane and the line landform. Initially, the scarp height increases, then decreases, and subsequently increases again (Figure 9b). Some small breaks in scarp height can be found along the strike. Meanwhile, the scarp width shows a similar variation along the strike as the scarp height (Figure 9c). In addition, the scarp slope presents relatively small variations, with values that generally exceed 20° only starting from 1250 m (Figure 9d). Then, it continues to increase and remains constant along the left fault trace.

5. Discussion

5.1. Spatial Pattern of the Fault Scarp Morphometric Parameters

The along-strike pattern of scarp height for the three typical deformed regions displays a bell-shaped profile with the maximum scarp height near the center of the fault [22,23]. We interpret the Pabu fault as potentially consisting of a single major segment. At least a partially significant tectonic deformation region along the strike exhibits this obvious phenomenon. The local highs and lows in scarp height, or rather breaks in continuous scarp height profile, are considered secondary segments and associated linkage structures [21]. The Pabu fault is 25 km long, with three typical tectonic deformation regions analyzed in this study, each measuring 2 km in length. We should exercise caution when interpreting the breaks in the scarp height profile of the fault, although the scarp width and slope suggest the presence of a fault scarp. First, the fault is of a relatively small scale. Non-tectonic features may lead to the SPARTA failing to identify the fault scarp [21]. For example, the irregularities in the original surface were create by freeze–thaw action or glacial periglacial action in the Pabu graben region. This may cause significant errors in the linear regression fitting between the upper original and lower original surfaces away from the scarp. In addition, the elevation distortion points derived from optical satellite stereo image pairs can also affect the determination of the fault scarp. In the middle part of the Pabu fault, fault segment F2 may function as a transfer fault, which could be another reason for the decrease in scarp height observed in Figure 8. Thus, the second fault segment derived from the scarp height profile should be validated in the field. Because the SPARTA algorithm assumes only a single scarp surface [21], the scarp height thus represents a single event or the cumulative vertical displacement at the surface in this study. These scarps were likely formed by one or more earthquakes. However, we cannot rule out the possibility that variations in scarp morphology also indicate differences in surface fault slip within a single earthquake.

A linear regression (using the least-squares method) is applied to the upper and lower original surfaces away from the scarp in this study (Figure 4). If we manually select the crest and base of the fault scarp, we can define the scarp width. By fitting regression lines along the crest and base, we can determine the slope of the scarp. However, if changes in scarp height are less than 5 m—within the typical algorithm error in SPARTA —these height changes are considered insignificant and are not calculated as real along-fault changes in scarp morphology [21]. The algorithm’s poor performance in this study is likely due to ambiguities in scarp morphology, where non-tectonic features may lead to the misidentification of the fault scarp. Additionally, it is possible that the actual scarp heights are less than 5 m. Both cases may explain why the scarp height profile appears discontinuous, while the scarp width and slope remain continuous.

According to the SPARTA algorithm, the first step is to identify the fault scarp from the high-resolution DEM data [21]. The algorithm is designed to accurately determine the fault scarp by analyzing its slope and width, in comparison to traditional elevation profile methods across the fault line. This approach ensures that the identification of the fault is not confused with non-tectonic features as much as possible. In addition, by combining the spatial patterns of the fault scarp’s height, width, and slope, data breaks at the same location in these profiles can be used to infer the presence of secondary segments. However, due to the limited number of scarp height measurements, we caution against making definitive inferences about the existence of secondary segments based solely on the high-resolution DEM data in this study.

5.2. Uncertainty in the Derived Scarp Height

Since the algorithm assumes a single scarp surface, it treats multi-scarps or composite scarps associated with individual ruptures [24,25] as a single scarp. In other words, the calculated scarp height represents the cumulative vertical displacement at the surface. We acknowledge that interpreting the Paubu Fault’s history of activity is complicated by not knowing whether the scarp formed in one or more events. Generally, when assuming that the fault scarp was formed by a single earthquake event, researchers often use the average scarp height as a proxy for the average coseismic slip [26]. This information is used to calculate the slip–length ratio, which typically ranges between 10⁻⁵ and 10⁻⁴ for a single earthquake [27]. By comparing the average slip–length ratios for the fault, if they fall at or above the upper limit of this typical global range, it may suggest that the scarp height represents a single earthquake event [27]. However, as shown in Figure 7a, Figure 8a, Figure 9a, the profile does not exhibit a continuous distribution of scarp height. Using these data to calculate the slip-length ratio would introduce significant errors. As a result, it may provide only a limited basis for inferring whether the scarp height is the result of a single event or multiple cumulative events. Thus, the dating of the fault scarps along the Pabu fault needs to be conducted in future research.

Based on Figure 2 and Figure 3, it can be concluded that the Pabu graben formed within the Gangdise granite. Therefore, lithological differences in regional topography formation are not the main cause of the spatial patterns of scarp heights. The most likely factors influencing the landscape are freeze–thaw weathering and glacial activity during geological history. However, climatic influences could have had a similar impact on modifying the regional landforms. In other words, while freeze–thaw processes and glacial activity may alter the steepness of scarp landforms, their impact on the spatial patterns of scarp heights is probably limited.

In tectonically active regions, fault propagation processes are often revealed by distinct landforms. A characteristic morphological feature of exposed active faults is the presence of triangular facets. Figure 2 clearly illustrates the morphology of triangular facets along the fault trace. The highest concentration of landslides generally occurs on the faceted spurs adjacent to the fault [28]. In such cases, a single landslide may reduce the scarp height and accumulate debris at the base of the scarp, forming a colluvial wedge. As a result, this can make it difficult to accurately identify the top and base of the fault scarp. However, for landslides to significantly degrade the entire Pabu fault scarp, two preconditions would need to be met: repeated landslide events and the clustering of landslides along the entire, or at least most of, the fault trace. In the high-altitude, cold environment of our study area, relics of numerous landslides may be well preserved. Nevertheless, according to fieldwork observations (Figure 6), the fault scarp does not appear to have undergone significant degradation. Only small gullies or frost-weathered material were observed, which do not impact the scarp’s morphology as seen in remote sensing images and derived DEMs. Therefore, we believe that while landslides may play a role at specific sites, they do not significantly affect the entire or most of the Pabu fault at present.

5.3. Seismic Hazards Implications

The NGG displays distinct tectonic activity characteristics compared to its northern and southern counterparts [7,8]. Specifically, these differences are evident in the relatively small scale of the faults and their dense, parallel distribution (Figure 1). Additionally, recent seismic activity, including the 1901 Nimu M6 ³/₄ earthquake, the 1992 Angong Ms 6.5 earthquake, and the 2008 Yangyi Ms 6.6 earthquake, suggests that each graben within the NGG has the potential to generate significant earthquakes independently [8]. Assessing seismic hazards depends in part on records of earthquake magnitudes, as determined through paleoseismic investigations [29,30]. An empirical mathematical relationship between coseismic surface rupture length (SRL) and earthquake magnitude (M) has been established as follows

M = 5.08 + 1.16 ∗ log(SRL)

based on a statistical analysis of large intracontinental earthquakes globally [31]. The Pabu graben, which delineates the boundary of the active fault, could generate an earthquake with a magnitude of M 6.7 if the entire 25 km length of the fault were to rupture. The height of the scarp may indicate multiple paleoseismic events within the Pabu graben. If that is the case, uncertainties regarding the recurrence intervals of these earthquakes persist. Further paleoseismic trenching is necessary to uncover the paleoseismic history.

6. Conclusions

Based on high-resolution remote sensing images and a derived 1 m resolution DEM, we mapped the trace of the Pabu fault. Drawing on field observations and the spatial distribution of fault scarp heights, we infer that each segment of the Pabu fault corresponds to a distinct major fault segment. This suggests that the Pabu fault consists of a single major segment, which is a roughly N–S fault segment with a 3 km length linking the north and south segments. The second fault segment requires validation through fieldwork. If the entire 25 km length of the Pabu fault were to rupture, it could generate an earthquake with a magnitude of M 6.7.