Crustal and uppermost mantle structure of the northeastern Qinghai-Xizang Plateau from joint inversion of surface wave dispersions and receiver functions with P velocity constraints

1.   Introduction

The high topography and thick crust of the Qinghai-Xizang Plateau are the result of Indian-Eurasian continental collision, a process ongoing for approximately 50–60 million years (Molnar and Tapponnier, 1975; Yin and Harrison, 2000). Debates surrounding the uplift and growth of the Qinghai-Xizang Plateau have given rise to three main models. The rigid block model (e.g., Tapponnier et al., 2001) suggests stepwise Qinghai-Xizang Plateau growth, primarily through deformation along strike-slip faults. The continuous deformation model suggests that due to the coupling of the crust and uppermost mantle, stress can transfer within and across blocks, resulting in broad lithospheric deformation (e.g., England and Houseman, 1986; Zhang PZ et al., 2004). The crustal flow model (e.g., Royden et al., 1997; Clark and Royden, 2000) suggests that the mid-lower crust moves southeast around the Sichuan Basin and northeast around the Ordos Block, contributing to the topographic gradient around the Qinghai-Xizang Plateau.

The northeastern Qinghai-Xizang Plateau, due to its strategic location and a history of significant earthquakes, serves as an ideal site for studying Qinghai-Xizang Plateau deformation and uplift. Some receiver function (RF) studies indicate moderate to low crustal v_P/v_S ratios beneath the northeastern Qinghai-Xizang Plateau margin, accompanied by crustal azimuthal anisotropy. These findings align more closely with the whole-lithosphere shortening model (e.g., Pan SZ and Niu FL, 2011; Li XF et al., 2015; Ye Z et al., 2015; Wang Q et al., 2016; Wang WL et al., 2017; Huang ZC et al., 2017). Tomography studies have identified widespread mid-crustal low-velocity zones (LVZs) in the Songpan-Ganzi terrane (SPGZ) (e.g., Bao XW et al., 2013; Li HY et al., 2014; Shi YT et al., 2021; Yang YJ et al., 2012), consistent with the crustal flow model. Additionally, the observed Moho offsets (e.g., Zhu LP and Helmberger, 1998; Zhang ZJ et al., 2011; Ye Z et al., 2015) and the correlation between GPS and SKS splitting (e.g., Li J et al., 2011; León Soto et al., 2012; Chang LJ et al., 2017) support coupled lithospheric-scale deformation (e.g., Tapponnier et al., 2001; Yin A et al., 2008; Ye Z et al., 2015). Images from joint inversion of the RFs and surface wave dispersions indicated that the rigid block model and crustal flow model could be concomitant (e.g., Liu QY et al., 2014). In summary, despite numerous geophysical investigations, the deformation mechanism in this region remains a subject of debate. Well-constrained models encompassing crustal thickness, velocity, and v_P/v_S ratios are crucial for understanding Qinghai-Xizang Plateau uplift and growth.

To mitigate trade-offs between model parameters and inversion result ambiguities, joint inversion of independent datasets is a common practice (e.g., Julià et al., 2000; Bao XW et al., 2015a; Zheng C et al., 2016; Li JT and Song XD, 2021; Li MK et al., 2022; Li YH et al., 2023). However, for the conventional joint inversion of RFs and surface wave dispersions, the Moho depth is significantly influenced by the v_P/v_S ratio in the crust, which is often unconstrained and is normally set to a global average (1.75). Based on the method of Xu ZJ et al. (2013), Li JT et al. (2017) developed a new method that incorporates independently constrained P-wave information into the joint inversion. This method, incorporating an independent v_P model, enables simultaneous inversion of the v_S structure of the crust and uppermost mantle as well as the crustal v_P/v_S ratio. To enhance our understanding of the lithospheric structure in the northeastern Qinghai-Xizang Plateau, we utilized data recorded by 673 temporary stations (ChinArray II) and first implemented the improved H-κ method with azimuthal corrections (H-κ-c) for all the RFs. Subsequently, we applied a nonlinear joint inversion, which incorporated the P-wave velocity model constraint, to the corrected RFs and surface wave dispersions. This joint inversion scheme reduced model ambiguity and enabled us to simultaneously resolve lithospheric S-wave velocity, crustal thickness, and crustal v_P/v_S ratio in the study area. Our seismic images provide detailed insights into the low-velocity anomalies beneath the Qinling Orogen, SPGZ, and Qilian Orogen, supporting the notion of coupled lithospheric-scale deformation in this region.

2.   Data and methods

2.1   Data

Three independent datasets were used in this study: P-wave RFs, dispersion curves of surface wave group and phase velocities, and a P-wave velocity (v_P) model of the crust. These datasets were assembled with the temporary stations in northeastern Qinghai-Xizang Plateau.

RFs were obtained from teleseismic P-waveforms recorded at 673 broadband stations of the China Array (Phase II), deployed on the northeastern Qinghai-Xizang Plateau from September 2013 to April 2016 (Figure 1). To extract the RFs, earthquakes with magnitudes M_W≥5.5 and epicentral distances ranging from 30° to 90° were selected. To generate the RFs, the waveform data were first decimated to 10 samples per second and a 0.02 to 1 Hz band-pass filter was applied. Next, the time-domain iterative deconvolution method (e.g., Ligorría and Ammon, 1999) was employed to obtain P-wave RF waveforms. To account for back-azimuthal variations in the arrival times of Ps phases and crustal multiples caused by crustal anisotropy and dipping interfaces, we utilized the H-κ-c method (e.g., Li JT et al., 2019) to correct the RFs. A low-pass filter with a Gaussian parameter of 2.0 was used to remove high-frequency noise. We also visually inspected all the RFs and removed those without clear first P and/or Ps arrival phases. All the RFs with a clear Ps phase at each station were stacked into one RF, after moveout correction (e.g., Yuan XH et al., 1997) at a reference slowness of 6.4 s/deg using the IASP91 model (e.g., Kennett and Engdahl, 1991) and H-κ-c corrections. Finally, after harmonic correction and stacking, the RFs from the 649 stations were used in the joint inversion.

The group and phase velocity dispersion data (at periods of 10–70 s) at each station site were extracted from the surface wave tomography results for Chinese mainland (e.g., Bao XW et al., 2015b), which were obtained from more than 700,000 dispersion curves recorded at 1, 316 seismic stations, to generate group and phase velocity maps at periods of 10–140 s. Our crustal v_P model was sourced from an updated unified seismic tomography model of the continental China lithosphere, USTCLitho2.0 (e.g., Han SC et al., 2022). The USTClitho2.0 model uses 10 years of body-wave arrival times and surface-wave dispersion data; see details on the v_P model construction in Han SC et al. (2022).

2.2   Method

Our joint inversion followed the scheme developed by Li JT et al. (2017) for the joint inversion of RFs and surface wave dispersion curves (modified from Xu ZJ et al., 2013) with v_P constraints. The joint inversion was divided into two steps. First, the v_S profile at each station site was inverted using only the dispersion data. Next, based on this v_S model and the constraints of crustal v_P values, the fast global searching Neighborhood Algorithm (NA) (e.g., Sambridge, 1999) was used to minimize the misfit of the RFs and dispersions with model damping and smoothing. The objective function is constructed as follows:

where the first term is the RF misfit and the second term is the dispersion misfit. ${D}_{{\rm{obs}}}$ is the data that we are trying to fit, and ${D}_{{\rm{pred}}}$ is the prediction from a random model generated by the NA. Here, ${w}_{1}$ and ${w}_{2}$ are the relative weights of the RF and dispersion misfit, respectively. The weightings were chosen empirically. After many experiments, ${w}_{1}$=1 and ${w}_{2}$=1 were selected for our joint inversion. The third term of Equation 1 represents the model constraints. λ is the damping parameter; m is the model parameter (S-wave velocity or interface depth) and ${m}_{0}$ is the reference model. The fourth term is the smoothing constraint, where φ is the smoothing parameter and L is the Laplacian operator (e.g., Lees and Crosson, 1989). Another advantage of the joint inversion method is the introduction of a sedimentary layer in the parameterization, which significantly improves the waveform fitting of the RFs; omitting this feature from the inversion may result in artifacts in both the shallow and deep crust (e.g., Li JT et al., 2017). We also set a fixed v_P/v_S of 2.0 for the sedimentary layer (e.g., Shen WS et al., 2013) and fixed the mantle v_P/v_S to the global average of 1.8 (e.g., Li JT et al., 2017; Ye Z et al., 2017).

Figure  1.  Distribution of ChinArray II stations (triangles) used in this study, with topography (color), the main crustal blocks (labeled, boundaries indicated by dark blue lines) and major historical earthquakes (M>7) (stars). The green triangles indicate the two example stations used for the joint inversion demonstration in Figure 2. Black lines AA' to DD' indicate the locations of the velocity cross sections in Figure 5.

DownLoad: Full-Size Img PowerPoint

3.   Results

High-resolution images of lithospheric S-wave velocity, crustal thickness, and crustal v_P/v_S ratio were obtained for the northeastern Qinghai-Xizang Plateau from the joint inversion of surface wave dispersions and RFs with P velocity constraints. Before interpreting these results, we inspected them at two representative stations, as shown in Figure 2. The nearly perfect fit of the RFs and surface wave dispersions confirmed the robustness of the joint inversion results. The smoothed crustal P-wave velocities and v_P/v_S variations in Figure 2 also imply that anomalous v_P/v_S ratios originate mainly from v_S anomalies, which confirms the results obtained by Li JT et al. (2017) and Deng YF et al. (2018).

Figure  2.  Examples of joint inversion of RFs and surface wave dispersions with v_P constraints for (a) station 62436 and (b) station 64006 (locations shown in Figure 1). The panels for each station show the best-fitting v_S model from the joint inversion, the input v_P model from USTClitho2.0 (Han SC et al., 2022), the derived v_P/v_S ratio, the observed (red) and predicted (blue) RFs, and the observed (red) and predicted (blue) group and phase dispersion curves, respectively. The arrow indicates the Moho depth obtained from the joint inversion.

DownLoad: Full-Size Img PowerPoint

3.1   Crustal thickness

An initial v_S model, obtained by inverting only the dispersion data, is commonly utilized in joint inversion to speed up the convergence of the NA search (e.g., Xu ZJ et al., 2013). Similarly, before the joint inversion, we corrected all of the RFs using the H-κ-c method (e.g., Li JT et al., 2019) to improve the inversion. The inverted Moho depth is shown in Figure 3a. Crustal thickness exhibits a good spatial correlation with the tectonic blocks in the study area. The Moho depth decreases from the center of the northeastern Qinghai-Xizang Plateau towards the edge. The thickest crust occurs in the western Qilian Orogen, whereas the stable Alxa and Ordos Blocks exhibit a thinner crust. There is an evident contrast between the thick crust in the SPGZ and the thin crust in the Qinling-Dabie Orogen to the east.

Figure  3.  Estimates of crustal thickness from this study. (a) Crustal thickness from RF inversion with the H-κ-c method. Dark blue lines indicate the block boundaries. (b) Same as (a), but for crustal thickness from the joint inversion. (c) Comparison of crustal thickness obtained using the two methods. The red dashed line indicates the same value of crustal thickness from the two methods. Blue dashed lines indicate ±3 km deviation from the same value.

DownLoad: Full-Size Img PowerPoint

The crustal thicknesses estimated using the joint inversion method with v_P constraints are shown in Figure 3b. The crustal thickness varies from ~66 km in the northeastern Qinghai-Xizang Plateau to ~36 km in the Ordos Basin, which is consistent with previous estimates obtained with the RF method (e.g., Li XF et al., 2015; Pan SZ and Niu FL, 2011; Wang Q et al., 2016; Wang XC et al., 2017; Wang WL et al., 2017) and active-source seismic profiles (e.g., Liu MJ et al., 2006; Teng JW et al., 2013; Wang CY et al., 2014). Crustal thickness also exhibits significant lateral variation within each block. There is a remarkable contrast between the western and eastern Qilian terranes: the western Qilian Block features a thick crust (~65 km), whereas the eastern Qilian Block has a thin crust (~45 km). The crust is slightly thicker (~50 km) beneath the western Alxa Block and thinner (~45 km) beneath the eastern Alxa Block. The crustal thickness in the western Qinling terrane decreases from ~55 km in the west to ~40 km in the east.

Comparing the crustal thicknesses estimated by joint inversion with those obtained from RFs and the H-κ-c method (e.g., Li JT et al., 2019) reveals differences of less than 3 km at most of the stations (Figure 3c). The largest differences in crustal thickness occur in the Qilian Mountains, northwestern Ordos Basin, and northern Sichuan Basin, which coincide with block intersections and areas covered by thicker sedimentary layers.

3.2   Three-dimensional S-wave velocity model

A high-resolution 3D S-wave velocity model of the crust and uppermost mantle beneath the northeastern Qinghai-Xizang Plateau was obtained from a joint inversion of the RFs and dispersions with independent v_P model constraints. Below, we examine the S-wave velocity maps at depths of 4, 20, 40, 60, 80, and 100 km, as shown in Figure 4.

Figure  4.  Horizontal slices of the S-wave velocity model (color) at depths of (a) 4 km, (b) 20 km, (c) 40 km, (d) 60 km, (e) 80 km, and (f) 100 km. Dark blue lines represent the main block boundaries.

DownLoad: Full-Size Img PowerPoint

At a shallow depth of 4 km (Figure 4a), regions of high S-wave velocity occur in the Qinling Orogen, southern Qilian Orogen, and Alxa Block, while a low S-wave velocity anomaly lies below areas with thick sedimentary deposits, such as the Ordos Basin and Hetao Basin. Some low-velocity anomalies are also evident along the western Qilian Orogen, coinciding with sedimentary basins. The S-wave velocity is higher in the southeast Ordos Block than in the northwest, indicating a northwest-to-southeast decrease in sedimentary thickness in the Ordos Block. The 20-km depth slice (Figure 4b) reveals significantly low S-wave velocity anomalies in the upper crust of the SPGZ, west of the Qinling Orogen, and below the Qilian Orogen. The Ordos Basin exhibits a high S-wave anomaly, particularly in the northern part of the basin.

At a depth of 40 km (Figure 4c), the Hetao Basin, northern Ordos Basin, and Qinling Orogen show high S-wave velocities, implying a thin crust in these areas, as observed in Figure 3. With relatively low v_S values, the Alxa and southern Ordos Blocks have begun to reach the mantle at this depth. In contrast, the northeastern Qinghai-Xizang Plateau, composed of the SPGZ and the Qilian Orogen, is still in the crust at this depth with a much lower S-wave velocity. The southwestern margin of the Ordos Block shows relatively lower velocity anomalies than the northern Ordos.

At a depth of 60 km, the SPGZ Block, North Qilian Orogen, and western part of the Qinling Orogen in the study area showed low-velocity anomalies (Figure 4d), whereas high S-wave velocities were present in most parts of the Alxa and Ordos Blocks. As shown in Figure 4e, the S-wave velocities at a depth of 80 km are well correlated with the block boundaries. Low S-wave velocities are generally observed in the Qilian Orogen, SPGZ, and western part of the Qinling Orogen. The eastern part of the Qinling Orogen and the entire Ordos and Alxa Blocks exhibit obvious high-velocity anomalies, whereas some low S-wave velocity anomalies remain at the boundary between the Ordos and Alxa Blocks.

At a depth of 100 km (Figure 4f), the Ordos and Sichuan Basins are characterized by high velocities, and the South Qilian Orogen also presents a higher S-wave velocity. In contrast, the SPGZ, western part of the North Qilian Orogen, and western part of the Alxa Block reveal obvious low-velocity anomalies.

To better demonstrate the variations of the S-wave velocity and v_P/v_S ratio with depth, we plotted several vertical slices of the 3-D velocity model, as shown in Figure 5. The locations of the profiles were chosen to be nearly orthogonal to the Qinghai-Xizang Plateau orogeny margin (Figure 1). For the Qilian Orogen, massive low-velocity anomalies in the crust and uppermost mantle are evident in Figure 5a, 5b, and 5c, respectively. From west to east in the western Qilian Orogen, the S-wave velocity increases slightly, and the low-velocity crustal material becomes thinner. The southern Qilian Orogen not only contains low-velocity anomalies in the crust and uppermost mantle but also high v_P/v_S ratios in its mid-lower crust, indicating the presence of weak and ductile materials (Figure 5b).

Figure  5.  Vertical slices of the v_S and v_P/v_S models (color) along selected profiles (a) AA′, (b) BB′, (c) CC′, and (d) DD′. The locations of the profiles are shown in Figure 1. The topography (top) and the Moho depth beneath each station (black crosses) are also plotted. Abbreviations: QLO: Qilian Orogen; WQL: western Qinling; SPGZ: Songpan-Ganzi terrane.

DownLoad: Full-Size Img PowerPoint

For the Alxa Block, S-wave velocities in the mid-lower crust show distinct features east and west of 103°E (Figure 5a and 5b). Beneath the western Alxa Block, S-wave velocity is significantly high, and the Moho is notably deep. The eastern part of the Alxa Block is closer to the stable Ordos Basin and has a stronger middle and lower crust. Overall, the Alxa Block exhibits a low crustal v_P/v_S ratio, except in the northeast, where the mid-lower crust shows relatively high v_P/v_S anomalies. In addition, the crustal thickness of the Alxa Block gradually decreases from west to east and south to north. There is a significant uplift of the Moho interface from the northern Qilian Orogen and east towards the boundary of the Alxa Block (Figure 5a and 5b), which corresponds with the undulation of the topography.

Our joint inversion model reveals a remarkable contrast between the SPGZ and Qinling-Dabie orogens (Figure 5d). Beneath the SPGZ in the west, crustal low-velocity anomalies are accompanied by high v_P/v_S ratios (>1.9) in the mid-lower crust. In contrast, the eastern Qinling-Dabie Orogen is characterized by higher crustal velocities and lower v_P/v_S ratios. Furthermore, a prominent LVZ (v_S < 4.2 km/s) residing in the uppermost mantle is present beneath the SPGZ at depths of 70–100 km (Figure 5d). The high-velocity mass beneath the Qinling-Dabie Orogen indicates material with a high mechanical strength. Based on the apparent differences in velocity and v_P/v_S ratios, the boundary between the two tectonic elements can be confidently confirmed.

3.3   Three-dimensional crustal v_P/v_S

Studies of laboratory data have indicated that the v_P/v_S ratio is related to rock types and can be used to constrain the average bulk composition of the crust (e.g., Christensen, 1996). To shed light on the components of the crust, a 3D crustal v_P/v_S model was derived using joint inversion with P-wave velocity constraints. Figure 6 depicts the characteristics of v_P/v_S ratios at depths of 4, 20, 40, and 60 km.

Figure  6.  Horizontal slices of the v_P/v_S ratios (color) at depths of (a) 4 km, (b) 20 km, (c) 40 km, and (d) 60 km. Dark blue lines represent the main block boundaries. Note that the color bar in (d) presents a different scale to better indicate the variation in v_P/v_S ratios between crust and mantle.

DownLoad: Full-Size Img PowerPoint

At a shallow depth of 4 km (Figure 6a), the v_P/v_S ratio is fairly uniform (~1.7) across most of the study area. High v_P/v_S ratios can be seen below the SPGZ Block, western Qinling Orogen, and southern Ordos Block. Some high v_P/v_S ratios occur along the western Qilian Orogeny, coinciding with sedimentary basins. The upper crust slice (Figure 6b) plotted at a depth of 20 km reveals significantly low v_P/v_S anomalies in the northern Ordos Block, west of the Qilian Orogen, and below the northern Sichuan Basin. The northern Alxa Block shows high v_P/v_S anomalies (>1.8). Uniform v_P/v_S ratios can be observed in the remainder of the study area.

At a depth of 40 km (Figure 6c), v_P/v_S ratios in the lower crust vary significantly between the different blocks. The SPGZ Block, southern Qilian Orogen, and central Alxa Block show high v_P/v_S ratios (>1.9), implying the existence of partial melting in these areas, which correlates well with the low velocities shown in Figure 4c. The southern Ordos Block and Qinling Orogen possess relatively high v_P/v_S ratios (~1.8). Low v_P/v_S ratios (~1.7) were observed in the northern Ordos Block and western Qilian orogeny.

Because the mantle v_P/v_S ratio was fixed at the global average of 1.8 (e.g., Li JT et al., 2017; Ye Z et al., 2017), the horizontal slice at a depth of 60 km (Figure 6d) shows a fairly constant v_P/v_S ratio of 1.8 across the Alxa Block, Ordos Block, and Qinling Orogen. The northeastern Qinghai-Xizang Plateau, composed of the Songpan-Ganzi Block and the Qilian Orogen, also exhibits high v_P/v_S ratios (≥1.8), as most of this material lies in the mantle. High v_P/v_S ratios (>1.8) beneath the western Qilian Orogen imply that the Orogen is still in the crust at this depth, which is consistent with the thick Moho in Figure 3b.

4.   Discussion

4.1   Crustal LVZs in the northeastern Qinghai-Xizang Plateau

Recent seismic tomographic and active source images (e.g., Bao XW et al., 2015a, b; Li HY et al., 2014; Ye Z et al., 2015; Zhang ZJ et al., 2013; Zheng C et al., 2016) have revealed low-velocity anomalies in the crust and uppermost mantle beneath the Qilian Orogen (Figure 5a and 5b). Yang YJ et al. (2012) established that an S-wave velocity below 3.4 km/s at a 30 km depth serves as the criterion for partial melting. These low-velocity anomalies have been interpreted as indicative of partial melting or fluid formation in the Qilian crust. During the uplift process, the eastern and western segments undergo varying forms of deformation (e.g., Li HL et al., 2022). The low-velocity crustal materials in the western Qilian Orogen gradually thin from west to east, accompanied by a slight increase in the shear wave velocity. In contrast, no obvious low-velocity features are discernible in the crust of the eastern Qilian Orogen (e.g., Zheng C et al., 2016; Zhao PP et al., 2021). The low-velocity anomalies in the middle and lower crust below the Qilian Orogen may stem from delamination or thermal erosion of the lithosphere triggered by hot asthenospheric flow (e.g., Deng YF et al., 2018; Li HL et al., 2022).

Our joint inversion model reveals a notable LVZ within the middle to lower crust of the SPGZ (Figure 5d). This substantial intracrustal LVZ gradually thins eastward, concluding at the SPGZ-Qinling boundary. High v_P/v_S ratios (>1.9) indicate the presence of weak materials (partial melts) in the lower crust beneath the SPGZ. The v_S and v_P/v_S structures in the SPGZ may support the existence of partial melting and even the mid-lower crustal flow model (e.g., Royden et al., 2008). Seismic tomography (e.g., Li C and van der Hilst, 2010; Liang XF et al., 2012; Liu M et al., 2004) has revealed a pervasive and continuous upper mantle LVZ at asthenospheric depths from the central and northern Qinghai-Xizang Plateau to the eastern Qinghai-Xizang Plateau, indicating the eastward migration of this asthenospheric flow, a phenomenon corroborated by SKS anisotropy analysis (e.g., León Soto et al., 2012). Combined with previous seismic tomography and RF studies (e.g., Lei JS and Zhao DP, 2016; Zhang ZJ et al., 2010), we infer that delamination or thermal erosion of the lithosphere triggered by hot asthenospheric flow contributed to the observed uppermost mantle LVZ in the SPGZ (Figure 7).

Figure  7.  3D views of the LVZs beneath the northeastern Qinghai-Xizang Plateau and surroundings. (a) Perspective view; (b) side view along latitude; and (c) side view along longitude. Red masses indicate crustal LVZs with v_S < 3.4 km/s, and brown masses represent lithospheric mantle LVZs with v_S < 4.2 km/s. White lines indicate the major geological boundaries. Surface topography is shown in (a).

DownLoad: Full-Size Img PowerPoint

To examine the LVZs and their connectivity, Figure 7 presents a 3D sketch of the low-velocity masses in the crust and upper mantle. The crustal LVZs with v_S < 3.4 km/s and the lithospheric mantle LVZs with v_S < 4.2 km/s are indicated in red and brown, respectively. Low-velocity anomalies are widespread in the crust and the uppermost mantle beneath the northeastern Qinghai-Xizang Plateau. The correlation between crustal and uppermost mantle velocities is evident in the SPGZ and Qilian Orogen (Figure 7a and 7b). The crustal LVZs beneath the SPGZ are massive and have high v_P/v_S ratios, which may imply a weak and ductile mid-lower crust. However, the absence of low-velocity anomalies and intermediate-to-low v_P/v_S values (≤1.8) within the crust of the Qinling Orogen may signify its relative strength. Consequently, these findings contest the notion of the Qinling Orogen serving as a conduit for lower crustal flow escaping from the interior of the Qinghai-Xizang Plateau to the east (e.g., Ye Z et al., 2017). A LVZ gap between 36°N and 37°N (Figure 7b) implies that the strong lithosphere of Kunlun-West Qinling hinders the northward movement of Qinghai-Xizang’s material. Located in the middle of the Qilian Orogen (Figure 7c), this LVZ gap suggests a discrepancy in the lithospheric structure and deformation mechanism between the eastern and western Qilian Orogen (e.g., Li HL et al., 2022).

4.2   Crustal deformation in the northeastern Qinghai-Xizang Plateau

While certain previous tomography studies (e.g., Cheng B et al., 2016; Jiang CX et al., 2014; Li HY et al., 2014) supported the crustal flow model (Royden et al., 1997; Clark and Royden, 2000) in the northeastern Qinghai-Xizang Plateau, wherein crustal thickening arises from the extrusion of ductile mid-to-lower crustal materials from the center to the margin, our joint inversion results are incompatible with the crustal flow model across significant areas of the northeastern Qinghai-Xizang Plateau. Prominent LVZs are mainly present in the mid-lower crust beneath the SPGZ and the Qilian Orogen. Despite exhibiting high v_P/v_S ratios, these LVZs are not interconnected (Figure 7b and 7c). The absence of connected LVZs in the mid-lower crust within much of the northeastern Qinghai-Xizang Plateau indicates that possible partial melts in this region are all localized and isolated. Additional geophysical factors, including high electrical resistivity (e.g., Xiao QB et al., 2016), moderate heat flow (e.g., Hu SB et al., 2000), and low-to-intermediate Poisson’s ratios (e.g., Li YH et al., 2006; Pan SZ and Niu FL, 2011; Wang XC et al., 2017), observed across extensive parts of the northeastern Qinghai-Xizang Plateau, further deviate from the crustal flow model applicable to this region.

The Moho depth serves as a key parameter for characterizing crustal structure and deformation (e.g., Zandt and Ammon, 1995). The substantial lateral variations in the Moho depth (Figure 3a and 3b) indicate the complexity of the crustal structure beneath the northeastern Qinghai-Xizang Plateau and its adjacent regions. The strength contrast across the plateau boundary plays an important role in crustal/lithospheric deformation and topographic evolution (e.g., Chen L et al., 2017; Clark et al., 2005; Sun YJ and Liu M, 2018). The crust and uppermost mantle beneath the Alxa Block exhibit significantly high v_S values (Figure 5a and 5b). Distinct high-velocity mantle lithosphere extending to a depth of ~250 km has recently been revealed by P-wave travel-time tomography (e.g., Guo HL et al., 2017). Based on S-wave RF data, Zhang HS et al. (2012) delineated a notably thick lithosphere in this region. Collectively, these findings indicate that the Alxa Block possesses a robust lithosphere that resists deformation despite substantial compression from the Qilian Orogen. Consequently, a convex front in crustal thickness materializes. In contrast, the Qilian Orogen is marked by low velocities in the lower crust and uppermost mantle (e.g., Guo HL et al., 2017; Li YH et al., 2017; Wang XC et al., 2017) and elevated heat flow (e.g., Hu SB et al., 2000; Tao W and Shen ZK, 2008), yielding a thin and fragile lithosphere that serves as the mechanical foundation for overall lithospheric shortening.

We argue that whole-lithosphere shortening is a more reasonable explanation for crustal thickening in the NE margin of the Qinghai-Xizang Plateau, based on the crustal thickness and strength of the lithosphere, as revealed by the v_S and v_P/v_S ratios. Lithospheric strength plays a pivotal role in the shortening process, with weaker blocks being more susceptible to deform. In response to crustal thickening and deformation, horizontal shear zones may form within the mid-lower crustal LVZs (e.g., Nabelek et al., 2010; Chen L et al., 2019; Bao XW et al., 2020). The scale of the shear zone indicates the stage of the lithospheric deformation and its strength. The large-scale LVZs in the SPGZ and Qilian Orogen reflect the deformation history of these terranes, which have resulted in a weak lithosphere (e.g., Deng YF et al., 2018; Bao XW et al., 2020). Interestingly, the crust’s low velocity closely corresponds with that of the uppermost mantle (Figure 7). Similarly, correlations between crustal and uppermost mantle velocities were also discovered in the eastern Qinghai-Xizang Plateau (e.g., Ye Z et al., 2017), northeastern Qinghai-Xizang Plateau (e.g., Deng YF et al., 2019; Li HL et al., 2022) and southeastern Qinghai-Xizang Plateau (e.g., Liu QY et al., 2014). These correlations also suggest that the crust and mantle are coupled as a lithospheric block, supporting the interpretation of coupled lithospheric-scale deformation in this region (e.g., Deng YF et al., 2018; Xu XM et al., 2018). This is also supported by seismic anisotropy through XKS splitting analysis, which suggests that the lithosphere beneath the NE margin of the Qinghai-Xizang Plateau deforms coherently and approximately parallel to the direction of surface GPS motion (e.g., Chang LJ et al., 2017; Hao M et al., 2019; Xu XM et al., 2018).

5.   Conclusions

In summary, our investigation encompassed a joint inversion of RFs and surface wave dispersions within the northeastern Qinghai-Xizang Plateau and its surrounding areas, alongside an independent v_P model. This approach yielded high-resolution images of v_S, v_P/v_S ratios, and crustal thickness, which unearthed shallow low v_S anomalies that correlate substantially with sedimentary basins, and demonstrated a decrease in crustal thickness outward from the northeastern Qinghai-Xizang Plateau. The Qilian Orogen, SPGZ, and southwestern Ordos Block exhibited elevated average v_P/v_S ratios, while the Alxa Block and interior of the Ordos Block were pervaded by low v_P/v_S ratios. Despite featuring low v_S and high v_P/v_S ratios, potential partial melts within the SPGZ’s middle to lower crust remain confined and do not extend across the Xizang-Qinling boundary. Notably, velocity coherence between the crust and uppermost mantle within the northeastern Qinghai-Xizang Plateau was remarkable, indicative of unified lithospheric-scale deformation in this region.