Structure of the Gonghe Sedimentary Basin in the northeastern Tibetan Plateau: evidence from teleseismic P waves recorded by a dense seismic array

SUMMARY

The Gonghe Basin in the northeast Tibetan Plateau presents significant potential for hot dry rock (HDR) geothermal resources. A 1990 M_w 6.4 earthquake in the basin furthers the need for an improved understanding of its sedimentary structure. In this study, we utilize data from a dense seismic array of 88 short-period seismometers deployed at an interstation spacing of approximately 3 km to scrutinize the sedimentary structure of the Gonghe Basin. By analysing teleseismic P waveforms, we identify P-to-S converted waves (Ps wave) originating from the sedimentary basement. We then determine the delay time between the Ps waves and the direct P waves (P wave) through waveform cross-correlation. By integrating this delay time with empirical velocity structure models, HDR borehole data and results from teleseismic receiver function analysis, we derive a sediment thickness model of the Gonghe Basin for the Qabqa geothermal area. Our findings reveal a gradual increase in sediment thickness from around 500 m in the east to approximately 3000 m in the west, which is consistent with other geophysical surveys and borehole data. The thick sediments in the basin could potentially serve as an excellent thermal storage cover for HDR. The strong ground motion simulation using our sediment thickness model shows that thick sediments can amplify seismic waves, increasing the risk of seismic hazards. Moreover, our study indicates that the clear Ps waves can be effectively extracted to construct a dependable sediment thickness model using teleseismic P waves recorded by a short-period dense seismic array.

1 INTRODUCTION

The Gonghe Basin, located in the northeast Tibetan Plateau and situated at the junction of the Qilian, Qaidam and Songpan-Ganzi massifs, is a rhombic graben basin that has been evolving since the Middle Cenozoic era (Fig. 1). During the Middle and Late Mesozoic periods, the basin margin experienced rupture, resulting in gradual uplift of the surrounding mountain ranges, while the central region of the basin continued to subside. The basin is bounded by several faults, including the South Margin of Qinghai Lake-Baoji, South Margin of Qinghai Nanshan, Wayuxiangka-Guinan and Wahongshan-Wenquan faults (Tapponnier & Molnar 1977; Hetzel 2013). The basin's sediments comprise mainly Quaternary alluviums and fluvial–lacustrine accumulations as well as Palaeoproterozoic–Neogene lacustrine units (Craddock et al. 2011, 2014; Lu et al. 2012). The basement of the basin is composed of Triassic metamorphic rocks and intrusive granites (Weinert et al. 2020, 2021).

The Gonghe Basin is renowned for its abundant reservoir of hot dry rock (HDR) geothermal resources (Feng et al. 2018; Gao et al. 2018). It is estimated that approximately 3000 km² of the basin is enriched with HDR. Notably, within the Qabqa geothermal area, six HDR boreholes, namely DR9, DR6, DR4, DR3, GR1 and GR2, have been drilled (Fig. 1). The average geothermal gradient in HDR borehole GR1 for example, reached 8.8 °C/100 m below 3366 m, whereas the temperature at the bottom of the borehole (3705 m) was high at 236 °C (Zhang et al. 2018; Zhang et al. 2020; Lin et al. 2021). The Gonghe Basin is the most promising HDR development area in China today. Consequently, exploring the spatial distribution characteristics of granites, subsurface temperatures and deep heat sources is crucial for estimating the total HDR resources and assessing their commercial value.

During the subsidence of the sedimentary basin, the sediments undergo progressive compaction and dehydration due to pressure, forming a denser, low-permeability barrier layer. When the burial depth of the mudstone is less than 1500 m, it exhibits lower diagenetic maturity and higher porosity, resulting in weaker sealing capability. Optimal sealing capability is obtained in the depth range of 1500–3200 m. However, when burial depth exceeds 3200 m, the mudstone undergoes higher diagenetic maturity, increased brittleness and increased fracturing under elevated formation pressures, subsequently reducing sealing capability. This layer serves as an effective shield against the migration and diffusion of deep-seated hydrocarbons, thus promoting the accumulation and concentration of basin oil and gas (e.g. Ni et al. 2014a). Additionally, the sedimentary layer acts as a thermal barrier, effectively insulating and preserving the deep-seated heat flow, with important subsequent implications for geothermal resource exploration and development (e.g. Chen et al. 2000; Fuchs 2018; Schütz et al. 2018; Wang et al. 2018; Gascuel et al. 2020; Li et al. 2020). Therefore, accurately determining the depth extent of the sediments in the Gonghe Basin is vital for exploring and developing HDR resources.

Moreover, on 26 April 1990, the Gonghe Basin experienced a M_w 6.4 earthquake (Xu & Chen 1997; Hao et al. 2012; Xie et al. 2018). As seismic waves propagate through the sedimentary basin, they undergo multiple reflections and refractions that resonate with the ultra-soft sediment. This resonance leads to significant ground motion and potentially severe structural damage to buildings (e.g. Chen & Nábelek 1988; Taborda & Bielak 2013; Isbiliroglu et al. 2015; O'Kane & Copley 2020). Consequently, acquiring a comprehensive understanding of the sediment structure is important for assessing seismic hazards in the Gonghe Basin.

Numerous passive seismic methods can be used to constrain the depth and wave speed structure of sedimentary basins. These techniques include ambient noise tomography (e.g. Shapiro et al. 2005; Huang et al. 2010; Lin et al. 2013; Li et al. 2016; Jia & Clayton 2021), horizontal-to-vertical spectral ratio (HVSR) analysis (e.g. Civico et al. 2017) and analysis of converted waves from local earthquakes (e.g. Hough 1990; Li et al. 2014; Ni et al. 2014b; Bao et al. 2018, 2019). P-to-S converted waves (Ps wave) and S-to-P converted waves (Sp wave) from the sedimentary basement and teleseismic receiver function analysis can also be used to investigate the sediment structures (e.g. Hough 1990; Li et al. 2014; Ni et al. 2014b; Yu et al. 2015; Bao & Niu 2017; Jiang et al. 2021). For example, using Ps wave and Sp wave waveform fitting and Sp/S spectral ratio methods, Chen et al. (1996), Langston (2003) and Chiu et al. (2016) studied the thickness and S-wave velocity of sediment in the Mississippi embayment basin, USA.

In this study, we examine the seismograms of the teleseismic earthquakes to identify Ps converted wave energy from the basement of the basin. The delay time between the first-arriving P wave and the Ps conversion is determined via cross-correlation analysis. Subsequently, we establish a correlation between the delay time and the sediment thickness by using the empirical formula of wave velocity variation with depth. Through this process, we obtain the sediment thickness in the Qabqa geothermal area within the Gonghe basin. To validate our results, we compare them with those obtained using the teleseismic receiver function analysis. We use the sediment thickness model to simulate strong ground motion and assess seismic hazards. Additionally, we use a synthetic modelling approach to examine the relationship between the delay time and the sediment parameters and structure, including the seismic ray parameter, sediment thickness, velocity and inclination. We also thoroughly discuss the factors that influence the accuracy of sediment thickness estimation using the delay time approach.

2 DATA AND PRE-PROCESSING

In this study, we utilized data from a short-period dense seismic array of 88 three-component portable seismometers (resonant frequency 4.5 Hz) with a frequency band range of 0.2–100 Hz. These seismometers were deployed in the Qabqa geothermal area in the Gonghe Basin, Qinghai Province, China (Fig. 1). To ensure high-quality imaging of sedimentary structures in the study area, the array was designed to cover the east-central part of the Gonghe Basin with an interstation spacing of approximately 3 km (Fig. 2a). During the deployment of the array (6 August to 13 September 2018), 51 earthquakes of M_w > 5.0 in the epicentral distance range of 30–95° were recorded (Fig. 2b). This data set of high-quality teleseismic waveforms serves as the foundation of our investigation into the sedimentary structures of the Qabqa geothermal area.

To prepare the raw teleseismic waveforms for further analysis, a series of pre-processing steps were performed. After instrument response removal, seismograms were demeaned, detrended and bandpass filtered with corner frequencies of 0.2 and 5 Hz. The bandpass filter range was determined via examination of the spectral characteristics of the seismograms (Fig. S1). Seismograms were then transformed into the L-Q-T (P, SV and SH) coordinate system following the approach of Kennett (1991) (Fig. 3).

3 DISTRIBUTION OF THE DELAY TIME

3.1 Teleseismic Ps-wave method

When a seismic P-wave propagates through an interface between the sediment and the crystalline basement at an oblique angle, it undergoes partial transmission as a P wave (the direct P wave) and partial conversion to an S wave (P-to-S converted wave). Because the sediment velocity is relatively low and the teleseismic P wave typically has a small incidence angle, most P-wave energy is recorded in the L-component. In contrast, the Ps-wave energy, as the S wave, is recorded in the Q-component. By analysing SV P-wave receiver function (SV PRF) waveforms, the coherent Ps wave can be identified (Fig. 4).

In the absence of seismic anisotropy, the delay time between the direct P wave and the later-arriving basement Ps conversion will depend on the thickness and velocity structure of the basin sediments and the ray parameter. For a horizontally layered medium, the delay time between the P waves and the Ps waves can be given as (Zhu & Kanamori 2000):

where |${\rm{\Delta }}{t_{Ps}}( {{\rm{\Delta }}{t_{Ps}} = {t_{Ps}} - {t_P}} )$| is the delay time between the P waves and the Ps waves. h is the sediment thickness. |${V_P}$| and |${V_S}$|⁠, respectively represents the average velocity of P and S waves in the sediments. p is the seismic wave ray parameter. Given the small incidence angle of the P waves, it can be treated as nearly vertical incidence. Consequently, the contribution of p can be neglected, resulting in an approximation of |${\rm{\Delta }}{t_{Ps}}$| as:

Therefore, after obtaining |${\rm{\Delta }}{t_{Ps}}$|⁠, given |${V_P}$| and |${V_S}$| of the study area, h of the study area can be derived as:

3.2 The delay time |$\Delta {t_{Ps}}$| from the waveform cross-correlation method

3.2.1 Methodology

As the P waves and the Ps waves are both generated by the incident P wave, they should be identical in waveform in ideal conditions. Nevertheless, due to the high-attenuation nature of the sediment, some waveform changes may occur. Despite this, the waveforms of the P and Ps waves still exhibit significant similarities. Consequently, the waveform cross-correlation method can be used to accurately determine the delay time |$\Delta {t_{Ps}}$| between the P waves and the Ps waves.

After selecting teleseismic waveforms with a high signal-to-noise ratio (Table S1) and manually picking first arrivals, we can estimate the delay time |$\Delta {t_{Ps}}$| from the maximum cross-correlation coefficient. The correlation functions of the L-component and Q-component can be calculated by:

where |${\rm Cor}( \tau )$| is normalized cross-correlation coefficient in the range of [−1, 1]. |${\widehat {{\rm Cor}}_{QL}}( \tau )$| is the correlation functions of the L-component (⁠|$L( t )$|⁠) and Q-component (⁠|$Q( t )$|⁠). |${\rm{*}}$| denotes the conjugate. |$t,\ N$| are the waveform length and finite sample numbers. |$\tau $| is the sliding time window length.

3.2.2 Synthetic tests

In order to explore the effectiveness of waveform cross-correlation analysis in distinguishing between thin and thick sedimentary layers, we devised two synthetic models. Both models comprise two horizontal layers: an upper sedimentary layer atop a homogeneous half-space. The first model comprises a relatively thin sedimentary layer (⁠|$h = 0.2\ {\rm km},\ {V_P} = 2.36\ {\rm km\,s}^{-1},\ {V_s} = 0.87\ {\rm km\,s}^{-1},\ p = 0.062\ {\rm s\,km}^{-1}$|⁠). The theoretical calculation shows that |$\Delta {t_{Ps}}$| of the synthetic model is 0.14 s. The second model comprises a relatively thick sedimentary layer (⁠|$h = 3.0\ {\rm km},\ {V_P} = 4.04\ {\rm km\,s}^{-1},\ {V_s} = 2.31\ {\rm km\,s}^{-1},\ p = 0.062\ {\rm s\,km}^{-1}$|⁠). The theoretical calculation shows that |$\Delta {t_{Ps}}$| of the synthetic model is 0.57 s.

To determine the optimal frequency band filtering, we conducted a series of experiments applying different bandpass filters to the synthetic seismic waveforms of the L-component and Q-component. Specifically, we used bandpass filters of 0.2–0.5 Hz, 0.2–2 Hz and 0.2–5 Hz, acting simultaneously both on the L and Q components. To accurately measure the delay time of Ps waves lagging behind P waves, we calculated the cross-correlation coefficient with |$\tau = - 2\sim 6s,{\rm{\ \ }}t = 1.5s$|⁠. Our results demonstrate that the waveform cross-correlation method is best suited for the 0.2–5 Hz frequency range, offering excellent resolution for both thin and thick layers (Figs 5, 6).

3.2.3 Results

After evaluating the preprocessed waveforms and determining the theoretical delay time |$\Delta {t_{Ps}}{\boldsymbol{\ }}$| under geological conditions, we estimated the cross-correlation coefficients with |$\tau = - 0.3\sim 6{\rm{\ }}s,{\rm{\ \ }}t = 1.5{\rm{\ }}s$|⁠. Through a rigorous selection of the waveform cross-correlation results, we eventually selected 64 stations’ Ps and P waves to obtain the spatial distribution of the delay time |$\Delta {t_{Ps}}$| of the dense seismic array in the Gonghe Basin (Fig. 7). The delay time |$\Delta {t_{Ps}}$| gradually increases from east to west and ranges from 0.36 to 1.02 s. For most stations, the standard deviation of the delay time |$\Delta {t_{Ps}}$| is less than 0.25 s.

3.3 The delay time |$\Delta {t_{Ps}}$| from the teleseismic receiver function method

3.3.1 Methodology

Receiver functions—time-series, computed from three-component seismograms—record the relative response of Earth structure below a seismograph station. The presence of the sedimentary layer affects P-wave receiver functions, causing the initial P-wave peak to partially represent the Ps wave. This initial peak exhibits a significant time delay compared to the P wave arriving at zero time, and this delay increases with sediment thickness. Therefore, by measuring the arrival time of the initial wave peak of the P-wave receiver functions, we can calculate the sediment thickness (e.g. Zheng et al. 2005; Borah et al. 2015; Agostinetti et al. 2018; Liu et al. 2018; Cunningham & Lekic 2020; Li et al. 2021).

P-wave receiver functions are obtained by deconvolving the waveforms of the Q-component by the L-component. In this study, we utilized the iterative time-domain deconvolution method of Ligorria & Ammon (1999) to calculate the P-wave receiver functions. Additionally, we applied a low-pass Gaussian filter to the receiver functions to remove high-frequency noise. The Gaussian filter in the frequency domain can be expressed as,

where |$\omega $| is the angular frequency. |$\alpha $| is the Gaussian factor, which represents the width of the Gaussian.

To assess the effectiveness of the receiver function method for estimating sediment thickness, we used the same models as in the earlier trials. We used a Gaussian factor of 10 for these simulations. According to the results of our analysis, sedimentary layers of varying thickness, including both thin and thick layers, could be resolved very well using the filtering frequency band of 0.2–5 Hz (Figs 8, 9).

3.3.2 Parameter tests

A Gaussian filter is applied to the receiver functions to remove high-frequency noise. As the width of the Gaussian factor is increased, more noise remains in the receiver functions. Therefore, there is a trade-off between the vertical resolution of the true strata and noise signal artefacts when imaging subsurface structures (Ward & Lin 2017). The ideal Gaussian factor depends on the data and research goals. We established that a Gaussian factor of 5.0 was the ideal compromise between removing high-frequency noise and guaranteeing that the Ps and PpPs phases did not overlap by examining seven alternative Gaussian factors ranging from 1.0 to 10.0 (maximum cut-off frequencies 0.50–5.0 Hz; Figs 10, 11).

3.3.3 Results

We calculated the P-wave receiver functions for each station using the iterative time-domain deconvolution method with a Gaussian factor of 5.0. These receiver functions were then stacked linearly. After quality control, we selected 62 stations’ stacked receiver functions to estimate the Ps-wave arrival times from the initial wave peak with a maximum search range of 1.5 s. The calculated delay time |$\Delta {t_{Ps}}$| ranges from 0.30 to 1.04 s and increases from east to west. For most stations, the standard deviation of the delay time |$\Delta {t_{Ps}}$| is less than 0.25 s (Fig. 12).

4 RESULTS AND DISCUSSION

4.1 Sediment thickness results

HDR borehole data from the Gonghe basin has revealed the Qabqa geothermal area's two-layered geological structure. The shallow layer comprises Cenozoic sediments, while the deep layer comprises Middle and Late Triassic granites. HDR boreholes DR9, DR6, DR4, DR3, GR1 and GR2 have sediment thicknesses of 1600, 1700, 1340, 1402, 1350 and 940 m. The core geological data of borehole GR1 reveals that the shallow sedimentary layer is a ∼ 500-m-thick Quaternary stratum of sand, gravel and siltstone. And the bottom half of the sedimentary layer is an ∼850-m-thick Palaeoproterozoic–Neoproterozoic layer composed primarily of mudstone and sandy conglomerate (Zhang et al. 2020). Based on the lithological data of the basin, we assigned the average velocity of |${V_s} = 1.40\ {\rm km\,s}^{-1}$| and |${V_p} = 3.80\ {\rm km\,s}^{-1}$| for the sediment layer (Jia et al. 2019; Wang et al. 2019). Using eq. (3), we calculated the distribution of sediment thickness beneath the short-period dense seismic array in the Gonghe Basin (Figs 14a and b).

Since wave velocity varies with depth, sediments with different thicknesses have varied average wave velocities. Thicker sediments tend to have higher average wave velocities than thinner ones (Brocher 2008). We built a new wave velocity model based on an empirical equation that characterizes the change in wave velocity with depth to integrate this variability (Table 1, Fig. 13a). By applying published connections between wave velocity and sediment thickness (Brocher 2005, 2008; Bao et al. 2018), we developed the association between the delay time and sediment thickness (Fig. 13b). We were able to convert delay time to sediment thickness:

where h is the sediment thickness, and |${\rm{\Delta }}{t_{Ps}}$| is the delay time between the Ps and P waves.

The study area's sediment thickness distribution range is 500–3000 m after time-depth conversion (Figs 14c and d). Interestingly, this matches HDR borehole data (Table 2). Sediment thickness correlated with the research area's surface topography, increasing from east to west. The eastern region, located in the basin's interior, has increased sediment deposition, in contrast to the western region, which is exposed mountain ranges. The sediment structures from the two approaches matched well, with sediment thickness increasing from ∼500 m in the east to ∼3000 m in the west. The thickness discrepancies between the two methods are typically less than 0.5 km, with the maximum thickness reaching 510 m at GH825 (Fig. 15, Table S2).

4.2 Factors influencing the accuracy of sediment thickness

The accuracy of obtaining the sediment thickness from Ps waves is affected by the following factors: the error in determining the delay time |${\rm{\Delta }}{t_{Ps}}$|⁠; the confidence with which the P-wave and S-wave speed structure of the sediment is known; and the difference between the actual sediment model and the theoretical model chosen for calculating the sediment thickness.

The determination error of the delay time |${\rm{\Delta }}{t_{Ps}}$| depends largely on the precision of identifying the arrival time of the Ps wave. The waveform cross-correlation method estimates this delay time based on the waveform similarity between P and Ps waves. The delay time is the sliding time window length with the maximum cross-correlation coefficient. Precise P- and Ps-wave arrival times are not required. Due to multiple waves and noise, the sliding time window length may not match the true delay time. Therefore, it is crucial to select an appropriate waveform length that captures the characteristics of both the P wave and Ps wave on the LQ components. This ensures the reliability of the waveform cross-correlation results and reduces the influence of multiple waves on the measurements. Additionally, excluding seismic records with a low signal-to-noise ratio can further minimize errors in determining the delay time |${\rm{\Delta }}{t_{Ps}}$|⁠.

Errors in P- and S-wave velocity values significantly affect the absolute sediment thickness. The wave velocity often increases with depth. In regions with large variability in sediment thickness, relying exclusively on average velocity may result in sediment thickness predictions that are either underestimated or exaggerated. The relative inaccuracy of the interface diminishes as the sediment thickness grows, but the absolute error increases. It is critical to account for the depth-dependent variation in wave velocity in order to achieve appropriate results.

The sediment thickness might be affected by theoretical model inaccuracies. Simplifying multilayer strata with two layers or ignoring interface inclination might cause calculation errors. The magnitude of error resulting from neglecting the multi-layered nature of the medium depends on the epicentral distance. The inaccuracy is insignificant for teleseismic earthquakes with epicentral distance |$\Delta \ge {30^{\circ}} $| (Fig. S2). The degree of depth estimation error caused by mistaking Ps waves from an inclined interface for a horizontal interface increases with interface inclination (Li et al. 2019). When the interface inclination |${\rm dip} \le {10^{\circ} } $|⁠, its effect is generally negligible (Fig. S3). Therefore, the sediment thickness obtained from teleseismic Ps waves can be approximated using eq. (3) for almost horizontal strata.

4.3 Comparison with other geophysical surveys

Many geophysical surveys have shown that the Qabqa geothermal area in the Gonghe Basin has a maximum sediment thickness of ∼3000 m (Wang et al. 2010, 2019 ; Jia et al. 2019). The deep seismic sounding (DSS) P-wave velocity profile (Jia et al. 2019) and receiver function S-wave velocity profile (Wang et al. 2019) show a gradual decrease in P- and S-wave velocity in the Gonghe Basin from east to west, indicating an increase in sediment thickness. This discovery is consistent with the trend of increasing sediment thickness from east to west from the waveform cross-correlation analysis (Figs 16b–e).

Sediment thickness distributions were plotted along three profiles (CC', DD' and EE’) crossing the HDR boreholes to compare the result to HDR borehole data (Figs 16f–h). Due to being close to seismic stations, HDR borehole DR4 has the best sediment thickness agreement at 0.06 km. However, seismic stations weakly limit HDR boreholes DR3, GR1, GR2, DR6 and DR9, resulting in larger discrepancies. These boreholes have thickness differences of 0.16, 0.10, −0.55, 0.12 and −0.29 km.

4.4 Strong ground motion simulation

In thick sedimentary layers, a significant impedance contrast between the sediment and bedrock causes a low fundamental frequency and a high amplitude peak, amplifying ground motions. Large structures, including bridges, high-rise buildings and industrial facilities, are at risk from low-frequency ground shaking (Cunningham & Lekic 2020). The seismic ambient noise horizontal-to-vertical spectral ratio (HVSR) is usually used to identify sediment layers’ fundamental frequency and amplitude peak, indicating the S-wave resonant frequency. We used the procedures given by Bao et al. (2019) to calculate the Qabqa geothermal area's HVSR (Figs 17b and c). There is a definite relationship between the sediment thickness and the fundamental frequency as well as the amplitude peak. Specifically, as the sediment layer thickness grows, the fundamental frequency lowers (Fig. 18a) while the amplitude peak increases (Fig. 18b).

Ground motion in sedimentary basins is complicated by intrabasin surface waves, basin edge effects and basin focusing effects. We simulated strong ground motion from the 1990 M_w 6.4 earthquake using our Gonghe Basin sediment thickness model to examine the significance of the sediment thickness model for seismic hazard assessment. The 3-D seismic wave curve grid finite difference method was used for the simulation (Zhang et al. 2012; Sun et al. 2018). In the calculations, surface topography data from the GTOPO30 DEM was used. Based on the depth-dependent seismic wave velocity model (Table 1), we modeled the earthquake as a point source at 32 km deep and simulated low-frequency strong ground motion (1 Hz and below). Our simulation results show that asymmetric PGV distributions extend to the west and northwest from the earthquake epicentre (Fig. 17d). The PGV is amplified by thicker sediments in the northwest (Fig. 18c). Therefore, our sediment thickness model provides valuable insights for ground motion simulation and seismic hazard quantification in the Gonghe Basin.

6 CONCLUSIONS

In this study, we deployed a dense seismic array of 88 short-period three-component seismometers in the Gonghe Basin in the northeast Tibetan Plateau. Using teleseismic waveforms recorded by the array, we extracted the delay time between the Ps waves from the sediment basement and the P waves by waveform cross-correlation analysis. Based on the delay time, empirical sediment velocity structure model, and HDR borehole data, we inferred a reliable sediment thickness model for the Qabqa geothermal area in the basin. Additionally, we compared this model with those derived from teleseismic receiver function analysis. We find that the sediment thickness of the Qabqa geothermal area of the Gonghe Basin gradually increases from ∼500 m in the east to ∼3000 m in the west. The thick sediment layers provide an ideal thermal storage cover for HDR. Furthermore, our strong ground motion simulations utilizing the sediment thickness model revealed that thick sedimentary layers may amplify seismic waves, causing significant seismic hazards.

Through theoretical analysis and data testing, we find that waveform cross-correlation analysis of teleseismic earthquakes recorded at the short-period dense seismic array can constrain the delay time between the Ps and P waves to construct sediment thickness. During the one-month observation period, the array can capture at least one or two teleseismic earthquakes of M_w > 5 per day with high signal-to-noise ratios. This is sufficient to detect sediment structure without relying on local earthquakes, which are typically difficult to gather due to time and space constraints. Thus, by deploying a dense array in the targeted sedimentary basin area, direct identification of teleseismic converted waves from the sedimentary basement is possible, providing valuable insights into the sediment structure.

SUPPORTING INFORMATION

Please note: Oxford University Press is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the paper.

ACKNOWLEDGEMENTS

We thank the editor and reviewers for their valuable comments that improve the paper. All figures in this study are plotted by the Generic Mapping Tools version 6 (Wessel et al. 2019). The GTOPO30 DEM from United States Geological Survey (https://www.usgs.gov/centers/eros/science/usgs-eros-archive-digital-elevation-global-30-arc-second-elevation-gtopo30). This work is supported by the National Natural Science Foundation of China (41974064 and 42174076) and China Geological Survey (DD20190131).

DATA AVAILABILITY

The data sets are available at: https://doi.org/10.6084/m9.figshare.20479851. The authors declare that there is no conflict of interest regarding the publication of this manuscript.

References