Shear-wave velocity map for Pohang Basin,South Korea,based on the P-wave seismogram method

Abstract

The 2017 M_L 5.4 Pohang, South Korea earthquake, which caused severe damage, raised attention to the importance of the characterization of the Pohang Basin structure and the identification of active faults. We utilize 366 seismograms recorded at 102 densely deployed seismic stations during 18 microearthquakes with M_L ranging from −1.11 to 1.70. We perform the P-wave seismogram method, which has been validated in various regions using seismograms for moderate earthquakes, to estimate the time-averaged shear-wave velocity from the surface to a depth of 30 m (V_S₃₀) in and around the basin. We perform multichannel analyses of surface waves at the 32 sites to obtain shear-wave velocity (V_S) profiles, and additionally acquire the 22 existing V_S profiles within the study area. We observe that velocity seismograms predict V_S₃₀ more accurately than acceleration and displacement seismograms do. The estimated V_S₃₀ values are in good agreement with the measured V_S₃₀ values. Approximately 93.3% of the measured and estimated V_S₃₀ values are located between the ±100% difference lines. We propose a V_S₃₀ map for the vicinity of the Pohang Basin by geospatially interpolating both the measured and estimated V_S₃₀ values, which is consistent with the local topography and geology. The proposed map exhibits areas with strong V_S₃₀ contrast within the Quaternary and Tertiary sediments, which might be attributed to the existence of faults.

Keywords

Microearthquake P-wave seismogram Pohang Basin fault

Introduction

It is well known that the ground response to earthquake loadings is affected by subsurface ground conditions (e.g. reduction of ground rigidity, shallow bedrock resulting in resonance excitations, and basin edges resulting in surface wave interactions), which are often represented by the shear-wave velocity (V_S). In particular, the time-averaged shear-wave velocity to a depth of 30 m (V_S₃₀) is used as a significant parameter for ground motion amplification due to the reduction in soil rigidity. In addition, abrupt changes in V_S profiles can be used to identify faults (e.g. Catchings et al., 2017, 2020; Duffy et al., 2014; Ivanov et al., 2006; Lee et al., 2010; Martínez-Pagán et al., 2018). The V_S profiles have traditionally been measured through invasive active-source borehole-based field tests such as suspension PS (s-PS) logging, downhole, and crosshole tests. Recently, non-invasive field tests, such as multichannel analysis of surface waves (MASW) and spectral analysis of surface waves, are widely used to measure V_S profiles using both active and passive sources. However, these field tests are constrained by the condition of the test site, as well as excessive time and cost.

As an alternative, Ni et al. (2014) proposed a method to estimate the average shear-wave velocity to a certain depth of z (V_SZ) using the ratio of the radial to the vertical component of the initial portion of P-wave seismograms. Kim et al. (2016) validated this method using ground motions recorded at 31 seismic stations in Central and Eastern North America. Miao et al. (2018) and Kang et al. (2020) also validated the method using 298 and 630 Japanese Kiban-Kyoshin network stations, respectively. In particular, Kang et al. (2020) proposed a V_S₃₀ map for the whole of Japan based on the P-wave seismogram method. Recently, Kim et al. (2020b) validated the method using ground motions recorded at 50 seismic stations in South Korea. All of these studies used seismograms from moderate earthquakes with moment magnitudes ranging from 2 to 5.5.

Despite the importance of V_S profiles or V_S₃₀, such properties have not been well studied for most of the basins in Korea. On 15 November 2017, an earthquake with an M_L of 5.4 occurred in the basin of Pohang City, South Korea, followed by 101 aftershocks with M_L≥2 (Korea Meteorological Administration (KMA) earthquake data service; http://www.weather.go.kr/weather/earthquake_volcano/domesticlist.jsp). This earthquake has been gaining significant attention from geoscience and engineering communities for the following reasons: (1) it was one of the largest earthquakes triggered by enhanced geothermal systems (e.g. Ellsworth et al., 2019; Grigoli et al., 2018; Kim et al., 2018b; Lee et al., 2019; Yeo et al., 2020), and (2) it caused severe damage to buildings and grounds because of its shallow focal depth (i.e. 4 km) (Ji et al., 2021; e.g. Kang et al., 2019a, 2019b; Kim et al., 2018a, 2020a). This earthquake occurred in the middle of the Pohang Basin, which is considered to have been generated by filling with unconsolidated marine sediments from the late Early Miocene to the late Middle Miocene (Song et al., 2015). However, the details of the structure and properties of the basin have not been well studied thus far.

To facilitate the prediction of ground motion intensities and identification of faults within the basin, this study acquired 1078 ground motion recordings during 19 small earthquakes with an M_L ranging from −1.11 to 1.7 from the dense recording array (i.e. 122 temporary seismic stations and one regular seismic station within a radius of 14.4 km). This study estimates V_S₃₀ at 102 stations using the P-wave seismogram method and proposes a V_S₃₀ map for the Pohang Basin. It observes the zones with sharp V_S₃₀ contrasts that are consistent with the known faults within the basin, which proves the validity of the proposed method for identifying new faults. In addition to its potential to support seismic hazard mitigation, this study is distinct from previous studies for the following reasons: (1) it utilizes records from the densely located array to create a local V_S₃₀ map and (2) it validates the use of microearthquake records for the P-wave seismogram method.

Methodology

Figure 1 shows a flow chart of the entire procedure for V_S₃₀ estimation. Signal processing, P-wave arrival time detection, and local baseline correction are described in the subsequent section, followed by another section for the methodology for V_S₃₀ estimation using P-wave seismograms.

Figure 1.

Flow chart for the procedures of V_S₃₀ estimation.

Signal processing

We followed the signal-processing procedure proposed by Kang et al. (2020). The raw seismograms were calibrated by conducting the baseline corrections. A Butterworth filter with band-pass (i.e. lower and higher cutoff frequencies of 5 and 30 Hz, respectively) was applied to the calibrated data to remove any background noise. The two horizontal components (i.e. east-west (EW) and north-south (NS)) of the records were rotated in the radial direction from the epicenter to the station. An algorithm called P_PHASE P_ICKER (Kalkan, 2016) was used to detect the arrival time of the P-wave. When the algorithm could not detect the arrival time, we selected them by visual inspection. To calibrate the radial and vertical components more accurately, we applied the local baseline correction within a time window of 0.5 s $(Δ t^{LBC})$ prior to the P-wave arrival. The maximum values of the P-wave initial pulse for the vertical component $({\overset{\cdot}{U}}_{Z})$ and those for the radial component $({\overset{\cdot}{U}}_{R})$ were selected simultaneously. Signal-to-noise ratios (SNRs) were measured to compare the signal (i.e. ${\overset{\cdot}{U}}_{R}$ and ${\overset{\cdot}{U}}_{Z}$ ) with the maximum noise within a time window of 0.5 s $(Δ t^{SNR})$ prior to the P-wave arrival. Finally, we used the ratios of ${\overset{\cdot}{U}}_{R}$ to ${\overset{\cdot}{U}}_{Z}$ to estimate V_S_Z and V_S₃₀, as described in the previous section. Examples of signal processing and V_S₃₀ estimation are presented in section “Vs measurements.”

P-wave seismogram method

Aki and Richards (2002) expressed the displacement of the P-wave in the radial, tangential, and vertical directions (U_R, U_T, and U_Z) on a free surface as follows:

[U_{R}, U_{T}, U_{Z}] = \frac{\bar{P} \cdot [R, T, Z] \exp [i ω (px - t)]}{{(\frac{1}{{V_{S}}^{2}} - 2 p^{2})}^{2} + 4 p^{2} \frac{cosi}{V_{P}} \frac{cosj}{V_{S}}}

(1a)

R = \frac{4 V_{p} \cdot p}{{V_{S}}^{2}} \frac{cosi}{V_{P}} \frac{cosj}{V_{S}}

(1b)

T = 0

(1c)

Z = \frac{2 V_{P}}{{V_{S}}^{2}} \frac{cosi}{V_{P}} (\frac{1}{{V_{S}}^{2}} - 2 p^{2})

(1d)

where $\bar{P}$ is the amplitude of the P-wave, V_P and V_S are the compression and shear-wave velocities of the subsurface, respectively, i is the incidence angle of the P-wave at the free surface, j is the reflected angle of the shear vertical (SV) wave from the free surface, and p is the ray parameter.

The ratios of the radial to vertical components of acceleration, velocity, and displacement $({\overset{\cdot\cdot}{U}}_{R} / {\overset{\cdot\cdot}{U}}_{Z}, {\overset{\cdot}{U}}_{R} / {\overset{\cdot}{U}}_{Z}, and U_{R} / U_{Z})$ are derived from Equation 1 as follows:

\frac{{\overset{\cdot\cdot}{U}}_{R}}{{\overset{\cdot\cdot}{U}}_{Z}} = \frac{{\overset{\cdot}{U}}_{R}}{{\overset{\cdot}{U}}_{Z}} = \frac{U_{R}}{U_{Z}} = \frac{2 V_{S} \cdot p \cdot cosj}{1 - 2 p^{2} \cdot {V_{S}}^{2}}

(2)

Note that the ratios for acceleration, velocity, and displacement are the same. The equation for velocity can be further arranged as follows:

V_{S} = \frac{cosj \pm \sqrt{co s^{2} j + 2 {(\frac{{\overset{\cdot}{U}}_{R}}{{\overset{\cdot}{U}}_{Z}})}^{2}}}{- 2 p (\frac{{\overset{\cdot}{U}}_{R}}{{\overset{\cdot}{U}}_{Z}})}

(3)

In this equation, V_S is dependent on j, p, and the ratio of the radial to vertical components.

The ray parameter p can be expressed as follows:

p = \frac{sini}{V_{P}} = \frac{sinj}{V_{S}}

(4)

The ray parameter p can be estimated by assuming two simplified layers from the known crustal velocity model (Kim et al., 2016). In this study, we used a crustal velocity model for Korea (Hong et al., 2017) (refer to Kim et al. (2020b) for more details on ray parameter estimation for Korea). Equation 4 can be rearranged for j as follows:

j = si n^{- 1} (p \cdot V_{S})

(5)

Once the ray parameter p is determined, the angle j and the shear-wave velocity can be determined by iteratively calculating Equations 3 and 5 (Kim et al., 2016). Kim et al. (2016) considered the shear-wave velocity in Equation 3 as equivalent to the time-averaged velocity for a depth z (V_SZ), where z can be estimated as the product of the pulse duration of the source time function and shear-wave velocity $(τ_{p} \times V_{SZ})$ . They used $τ_{p}$ of 0.1 s for magnitudes ranging from 3 to 4 (Harrington and Brodsky, 2009; Kanamori and Brodsky, 2001; Kikuchi and Ishida, 1993; Singh et al., 2000). In this study, we utilized the seismograms from considerably smaller earthquakes (i.e. magnitudes ranging from −1.11 to 1.7) whose $τ_{p}$ values have not been studied thus far. Therefore, we measured the pulse durations of the initial vertical components of the surface P-wave seismograms used in this study to determine the depth extent.

Note that the estimated V_SZ values correspond to various depth extents. To obtain average V_S values for the same depth extent, we converted V_SZ to V_S₃₀, which is also a key variable for ground motion amplification. Kim et al. (2020b) confirmed that the empirical correlations between V_SZ and V_S₃₀ for non-glaciated regions in Central and Eastern North America proposed by Kim et al. (2016) match with the data from Korea better than those by Boore et al. (2011) do. Furthermore, we checked that the residuals between the measured and estimated V_S₃₀ in this study were smaller when using the correlations for non-glaciated regions by Kim et al. (2016) than those by Boore et al. (2011). Therefore, we selected the correlations for non-glaciated regions in Central and Eastern North America proposed by Kim et al. (2016).

Kim et al. (2016) suggested that the P-wave pulse and pre-event noise are distinguished more clearly when using velocity seismograms than when using acceleration or displacement seismograms. In this study, we validated the use of all three types of seismograms (i.e. acceleration, velocity, and displacement).

Ground motion records

The Geophysics Research Group at Pusan National University deployed 69 and 60 temporary seismic stations in the Pohang Basin from 3 to 25 January 2019 (Phase 1) and from 27 January to 14 February 2019 (Phase 2), respectively. During these phases, 1078 ground motions were recorded at these stations and a regular seismic station operated by the KMA during 19 earthquakes. The locations of the stations and earthquake epicenters are illustrated in Figure 2a. Kang et al. (2020) reported that the P-wave seismogram method can yield over-prediction if the take-off angle is smaller than 40°. This is because the vertical components of P-wave seismograms are mostly stronger than the radial components, regardless of site conditions, if the waves are generated below the stations. Therefore, we did not consider records with such small take-off angles. In addition, we only considered seismograms with SNRs greater than unity, where the P-wave arrival times were distinguishable. This data selection resulted in 370 seismograms recorded at 50 stations operated during Phase 1 and 60 stations during Phase 2 for 18 earthquakes. The locations of the stations considered in this study are presented in Figure 2b.

Figure 2.

Locations of (a) seismic stations and temporary seismic stations deployed in the two phases (Phase 1: 3 to 25 January and Phase 2: 27 January to 14 February in 2019) and earthquake epicenters, and (b) the stations used in this study, as well as V_S measurements.

We applied the short-term-average/long-term-average (STA/LTA) algorithm to continuous seismic data to detect potential earthquake signals (e.g. Allen, 1982; Thompson and Reyes, 2017). Potential signals with more than five triggers were manually reviewed, and 19 earthquake signals were confirmed. The P-wave and S-wave arrival times were measured from the earthquake signals. We used the Hypoellipse earthquake location software (Lahr, 1999) with a simplified one-dimensional (1D) velocity model proposed by the Geological Society of Korea (2019) to obtain the initial locations of the 19 earthquakes. We refined the earthquake hypocenters using the HypoDD method (Waldhauser and Ellsworth, 2000). Source parameters (earthquake location, time, and magnitude) of 6 out of 19 earthquakes were reported by the KMA (2018), which is responsible for routine earthquake monitoring and reporting in Korea. The magnitudes of the 13 earthquakes were determined using a simple relationship between the amplitudes of the known and unknown events:

M_{unknown} = M_{known} + \log_{10} (\frac{A_{unknown}}{A_{known}})

(6)

where $M$ and $A$ denote the magnitude and amplitude, respectively. Note that the $M_{known}$ values are taken from the KMA earthquake catalog. We assume that the event pairs with high cross-correlation values are close enough to each other to estimate the relative magnitude (or $M_{unknown}$ ) using amplitude ratios (Kim et al., 2020c).

The 18 earthquakes have local magnitudes ranging from −1.1 to 1.7 and focal depths ranging from 4.4 to 5.9 km, as illustrated in Figure 3a and b. The recorded ground motions have epicentral distances of approximately 3–16 km (Figure 3c), and the maximum values of the peak ground velocity (PGV) in the radial and vertical directions are smaller than 0.1 cm/s (Figure 3d). Figure 4a presents a histogram of the SNR for the 370 selected seismograms.

Figure 3.

Histograms of (a) earthquake magnitudes, (b) focal depths, (c) epicentral distances, and (d) maximum PGV values of ground motions in the EW and NS directions used in this study.

Figure 4.

Histograms of (a) pulse durations and (b) signal-to-noise ratios of 370 recordings.

V _S30 estimation

Examples

All 370 ground motions recorded at 103 stations were used to estimate the V_S₃₀. The estimation process is illustrated in the following figures: Figure 5a depicts the velocity time series in the vertical and radial directions recorded at B042 station (the location is presented in Figure 2b) during the earthquake that occurred at around 04:43 on 7 January 2019. Figure 5b depicts a blowup of the arrival time of the P-wave, which is approximately 2.665 s. We applied a local baseline correction within a time window $(Δ t^{LBC})$ of 0.5 s to the velocity time series. We determined the maximum noise of ground motion in the vertical direction within a time window $(Δ t^{SNR})$ of 0.5 s and estimated the SNR to be approximately 3.11. Figure 5c illustrates the initial pulse of the P-wave. We determined that the ${\overset{\cdot}{U}}_{R}$ and ${\overset{\cdot}{U}}_{Z}$ values were approximately 6.26 × 10⁻⁶ and 1.56 × 10⁻⁵ cm/s, respectively, at the time $(t^{{\overset{\cdot}{U}}_{Z}})$ of 2.71 s. In this case, the ratio of ${\overset{\cdot}{U}}_{R}$ to ${\overset{\cdot}{U}}_{Z}$ is approximately 0.402. Using this ratio, the V_S_Z value was calculated to be 1275 m/s.

Figure 5.

(a) Velocity time series recorded at B042 station on 7 January 2019, in the radial and vertical directions, and ((b) and (c)) velocity time series near the arrival time of the P-wave.

We measured the time width of one wavelength from the initial pulse of the P-wave $(τ_{p})$ , as illustrated in Figure 5c. Figure 4b depicts a histogram of the pulse durations for all 370 records used in this study. The pulse durations between 0.03 and 0.05 s are dominant, and the mean pulse duration is 0.0475 s. Therefore, we determined the depth extent (z) for V_SZ by multiplying V_SZ with the mean pulse duration (i.e. as V_SZ × 0.0475 s). Subsequently, we converted V_SZ to V_S₃₀ using the correlations for non-glaciated regions proposed by Kim et al. (2016). The final V_S₃₀ was estimated to be approximately 1062 m/s. The V_S₃₀ measured at a site located approximately 73 m from this station was approximately 750 m/s.

Figure 6a depicts the velocity time series in the vertical and radial directions recorded at the B091 station during the earthquake that occurred at 23:22 on 2 February 2019. Figure 6b presents a magnified view of the arrival time of the P-wave, which is approximately 3.16 s. We applied a baseline correction of a time window of 0.5 s to the velocity time series once again. We estimated the SNR to be 8.38 for this seismogram. Figure 6c illustrates the initial pulse of the P-wave. The ${\overset{\cdot}{U}}_{R}$ and ${\overset{\cdot}{U}}_{Z}$ are approximately 5.80 × 10⁻⁶ and 7.38 × 10⁻⁵ cm/s, respectively, and the ${\overset{\cdot}{U}}_{R} / {\overset{\cdot}{U}}_{Z}$ ratio is approximately 0.079. The V_SZ and V_S₃₀ values were estimated to be approximately 234 and 315 m/s, respectively. The V_S₃₀ measured at a site located approximately 40 m from this station was approximately 497 m/s.

Figure 6.

(a) Velocity time series recorded at B091 station on 2 February 2019, in the radial and vertical directions, and ((b) and (c)) velocity time series near the arrival time of the P-wave.

From these two examples, the differences between the measured and estimated V_S₃₀ are approximately 34.4% and 44.8%, respectively. These differences may be attributed to the various assumptions (such as V_SZ-V_S₃₀ correlations and ray parameter estimation), epistemic uncertainty in recording seismograms, and aleatoric uncertainty of ground motions and subsurface soil layers. A systematic evaluation of the differences using all data is presented in the next section.

All estimates

The seismograms were recorded as velocity time series. However, we calculated the acceleration and displacement time series by differentiating and integrating the velocity time series, respectively, and used them to estimate the V_S₃₀ values. The V_S₃₀ values estimated from the acceleration, velocity, and displacement seismograms are denoted as $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ . Unlike previous studies (e.g. Kang et al., 2020; Kim et al., 2016; Miao et al., 2018), we used ground motions recorded at temporary stations during earthquakes with moment magnitudes of 2 or smaller (microearthquakes). Therefore, there could be some erroneous recordings that would result in unrealistic V_S₃₀ estimates and large differences in estimates from acceleration, velocity, and displacement seismograms. We excluded a few V_S₃₀ estimates that are lower than 50 m/s, which reduced the number of recordings to 366, 366, and 364 for acceleration, velocity, and displacement seismograms, respectively.

Figure 7a depicts the $V_{S 30}^{acc}$ values against the $V_{S 30}^{vel}$ values for 362 individual seismograms. A total of 71.5% (259/362) of the $V_{S 30}^{acc}$ and $V_{S 30}^{vel}$ values are between ±100% of each other (i.e. between 2:1 and 1:2 lines). We averaged the estimates from each of the 102 stations. Approximately 89.2% (91/102) of the averaged $V_{S 30}^{acc}$ and $V_{S 30}^{vel}$ values are located between ±100% of each other, as depicted in Figure 7b. Figure 7c depicts the $V_{S 30}^{dis}$ values against the $V_{S 30}^{vel}$ values for individual seismograms. Approximately 68.9% (248/360) of the $V_{S 30}^{dis}$ and $V_{S 30}^{vel}$ values are between ±00% of each other. Approximately 81.4% (83/102) of the averaged $V_{S 30}^{dis}$ and $V_{S 30}^{vel}$ values are located between ±100% of each other, as is evident in Figure 7d. It can be noted that there are fair correlations among $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ values.

Figure 7.

V_S ₃₀ values estimated using velocity seismograms $(V_{S 30}^{vel})$ versus those using acceleration seismograms $(V_{S 30}^{acc})$ (a) for all 362 individual seismograms and (b) those averaged for the 102 stations. $V_{S 30}^{dis}$ versus $V_{S 30}^{vel}$ (c) for 360 individual seismograms and (d) those averaged for 102 stations.

As depicted in Equation 2, the values of ${\overset{\cdot\cdot}{U}}_{R} / {\overset{\cdot\cdot}{U}}_{Z}, {\overset{\cdot}{U}}_{R} / {\overset{\cdot}{U}}_{Z}, and U_{R} / U_{Z}$ are theoretically identical, and therefore expect to result in the same V_S estimates. However, the real seismograms resulted in different estimates, mainly because of the precision of picking P-wave arrival times and ${\overset{\cdot}{U}}_{R}$ and ${\overset{\cdot}{U}}_{Z}$ values. Kim et al. (2016) reported that the initial pulses of the P-wave can be more clearly distinguished from the velocity seismograms than from the acceleration and displacement seismograms. This would be because of that the acceleration and displacement seismograms are dominated by relatively high- and low-frequency motions, respectively, which are more difficult to be distinguished from noise than intermediate motions in velocity seismograms. We compared the SNRs of the acceleration and displacement seismograms (SNR^acc and SNR^dis) with those of the velocity seismograms (SNR^vel) for the 362 and 360 records, respectively, used in this study, as illustrated in Figure 8. For approximately 63.8% (231/362) and 56.9% (205/360) of the records, SNR^vel values are greater than SNR^acc and SNR^dis values, respectively, which is consistent with the findings of Kim et al. (2016). In addition, we calculated the residuals between V_S₃₀ estimated by the acceleration or displacement time series and those estimated by the velocity time series as follows:

Re s^{acc - vel} = \ln (V_{S 30}^{acc}) - \ln (V_{S 30}^{vel})

(7a)

Re s^{dis - vel} = \ln (V_{S 30}^{dis}) - \ln (V_{S 30}^{vel})

(7b)

Figure 8.

Signal-to-noise ratios of (a) acceleration seismograms (SNR^acc) versus SNR^vel, and (b) SNR^dis versus SNR^vel.

We investigated the correlations between these residuals and various other factors. The residuals are not affected by earthquake magnitude, epicentral distance, take-off angle, or geology. However, the absolute values of $Re s^{acc - vel}$ increase with decreasing SNR^acc and SNR^vel, as shown in Figure 9a and b, respectively. In addition, the $Re s^{acc - vel}$ is dependent on the depth z, as shown in Figure 9c and d, raising the necessity of developing more precise correlations between V_S_Z and V_S₃₀ based on region-specific data. Similar trends were observed for $Re s^{dis - vel}$ . The estimates from the three types of seismograms are further evaluated in section “Final V_S₃₀ MAP.”

Figure 9.

Residuals between $V_{S 30}^{acc}$ and $V_{S 30}^{vel}$ versus (a) SNR^acc, (b) SNR^vel, (c) depth z for $V_{SZ}^{acc}$ , and (d) depth z for $V_{SZ}^{vel}$ .

We used a simplified two-layer crustal velocity model for the ray parameter estimation. Using the original crustal model (Hong et al., 2017) decreased the ray parameter estimates for the 366 velocity seismograms by approximately 6.04% on average and increased the V_S₃₀ estimates by approximately 5.48% on average, which resulted in an increase in the bias for the V_S₃₀ estimates. Therefore, we maintained the use of a simplified crustal velocity model.

Validation of V_S30 estimations

Vs measurements

We performed MASW at 32 sites to obtain V_S profiles. We used 24 geophones with a natural frequency of 4.5 Hz, spaced at intervals of 1 or 2 m depending on the site conditions. We performed active tests using a hammer with an offset of 10 m on both sides of the array (with a total of ten stacks) and a passive test using ambient noise for each site. We used ParkSEIS v2.0 (ParkSEIS, 2016) which automatically creates a dispersion image from the recorded surface waves and determines the shear-wave velocity through the inversion process. We extracted the fundamental-mode dispersion curves by picking the phase velocities at the highest sum of amplitudes from the dispersion images and conducted inversion analyses using the algorithm proposed by Xia et al. (1999) to obtain the V_S profiles. In addition, we collected the existing V_S profile data: seven V_S profiles measured by downhole tests, five V_S profiles obtained by s-PS logging, and ten V_S profiles obtained by MASW from various sources (i.e. the Geotechnical Information DB System (http://www.geoinfo.or.kr), Pohang Office of Education, National Disaster Management Research Institute (NDMI, 2018), and Korea Institute of Civil Engineering and Building Technology). The locations of all 54 V_S measurements used in this study are presented in Figure 2b, and the measured values are summarized in Table 1. Among the 54 profiles, 17 V_S profiles exceed a depth of 30 m. For those shallower than 30 m, the time-averaged shear-wave velocity values from the surface to the maximum possible depth were calculated and converted to V_S₃₀ values using the empirical correlation for non-glaciated regions proposed by Kim et al. (2016).

Table 1.

Measured V_S₃₀ values at the 54 sites, as well as the estimated V_S₃₀ values (i.e. $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ ) at the stations closer than 200 m from the measurement sites

Number	Latitude(°N)	Longitude(°E)	Method	$V_{S 30}^{mea}$ (m/s)	$V_{S 30}^{acc}$ (m/s)	$V_{S 30}^{vel}$ (m/s)	$V_{S 30}^{dis}$ (m/s)
1	36.1012	129.3458	MASW	250	NA	NA	NA
2	36.0962	129.3498	MASW	191	NA	NA	NA
3	36.0955	129.3373	MASW	415	320	682	709
4	36.1297	129.3884	MASW	139	NA	NA	NA
5	36.1052	129.3057	MASW	550	NA	NA	NA
6	36.1386	129.3283	MASW	355	NA	NA	NA
7	36.0814	129.3284	MASW	461	590	505	360
8	36.1062	129.3897	MASW	246	NA	NA	NA
9	36.0851	129.3465	MASW	206	NA	NA	NA
10	36.1536	129.3466	MASW	750	670	839	722
11	36.1276	129.3268	MASW	410	NA	NA	NA
12	36.1144	129.3714	MASW	147	NA	NA	NA
13	36.1025	129.3999	MASW	451	NA	NA	NA
14	36.1124	129.3789	MASW	164	NA	NA	NA
15	36.0517	129.377	Downhole test	412	NA	NA	NA
16	36.0664	129.3691	Downhole test	289	1231	628	893
17	36.0862	129.2784	MASW	588	NA	NA	NA
18	36.1165	129.336	MASW	414	NA	NA	NA
19	36.0798	129.2565	MASW	414	706	886	1226
20	36.1174	129.2576	MASW	421	314	265	680
21	36.0817	129.385	Downhole test	473	NA	NA	NA
22	36.0333	129.3802	Downhole test	167	NA	NA	NA
23	36.0521	129.3689	Downhole test	488	NA	NA	NA
24	36.0322	129.38	s-PS logging	176	NA	NA	NA
25	36.0319	129.38	MASW	246	NA	NA	NA
26	36.0427	129.3742	s-PS logging	176	NA	NA	NA
27	36.0428	129.3749	MASW	237	NA	NA	NA
28	36.1202	129.3539	MASW	247	NA	NA	NA
29	36.1675	129.3459	MASW	375	641	544	514
30	36.0537	129.3017	MASW	497	272	306	555
31	36.0789	129.3924	Downhole test	458	NA	NA	NA
32	36.1114	129.3115	MASW	385	NA	NA	NA
33	36.1104	129.3078	s-PS logging	436	NA	NA	NA
34	36.0599	129.3569	MASW	671	233	322	74
35	36.1167	129.3669	s-PS logging	284	NA	NA	NA
36	36.1121	129.3575	MASW	223	NA	NA	NA
37	36.1183	129.3617	s-PS logging	240	NA	NA	NA
38	36.1071	129.3571	MASW	269	NA	NA	NA
39	36.1172	129.3636	MASW	169	NA	NA	NA
40	36.1089	129.3685	MASW	144	NA	NA	NA
41	36.1062	129.3606	MASW	158	NA	NA	NA
42	36.1123	129.3663	MASW	275	NA	NA	NA
43	36.1362	129.3705	MASW	314	608	596	791
44	36.1353	129.3637	MASW	253	NA	NA	NA
45	36.0503	129.2672	MASW	679	326	484	802
46	36.0334	129.3352	MASW	603	451	689	447
47	36.1157	129.3145	Downhole test	492	NA	NA	NA
48	36.1258	129.3677	MASW	185	NA	NA	NA
49	36.1312	129.3578	MASW	477	NA	NA	NA
50	36.1263	129.3583	MASW	147	NA	NA	NA
51	36.1284	129.2382	MASW	626	508	467	811
52	36.1191	129.4041	MASW	158	NA	NA	NA
53	36.0787	129.366	MASW	555	641	807	1264
54	36.1932	129.3703	MASW	661	1252	1049	1093

MASW: multichannel analysis of surface waves; NA: not available.

Comparison between measured and estimated V_S₃₀ values

We compared the estimated $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ values with the V_S₃₀ values measured at locations closer than 200 m from the seismic stations $(V_{S 30}^{mea})$ , as summarized in Table 1. The residuals between the measured and estimated V_S₃₀ values can be computed as follows:

Residual = \ln (V_{S 30}^{mea}) - \ln (V_{S 30}^{acc}, V_{S 30}^{vel}, or V_{S 30}^{dis})

(8)

A positive residual implies that V_S₃₀ is underestimated, and a negative residual implies that V_S₃₀ is overestimated. Figure 10a to c shows the residuals for $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ , respectively, against $V_{S 30}^{mea}$ . The standard deviation values of residuals $(σ_{res})$ for $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ are approximately 0.65, 0.49, and 0.81, respectively, indicating that $V_{S 30}^{Vel}$ is more accurate than $V_{S 30}^{acc}$ and $V_{S 30}^{dis}$ . We calculated the bias values for the residuals. The bias values for $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ are approximately −0.043, −0.127, and −0.236, respectively. Previous studies (i.e. Kang et al., 2020; Kim et al., 2016, 2020b; Miao et al., 2018) also reported negative biases ranging from −0.48 to −0.073. These negative biases may be associated with the amplification of horizontal waves propagating near-surface soft soil deposits, which is not taken into account in the P-wave seismogram method. It is worth noting that negative residuals are prevalent at the soft sites (V_S₃₀ < 450 m/s), as shown in Figure 10.

Figure 10.

Residual for (a) acceleration $(V_{S 30}^{acc})$ , (b) velocity $(V_{S 30}^{vel})$ , and (c) displacement seismograms $(V_{S 30}^{dis})$ .

We investigated correlations of the residuals with various parameters such as magnitude, epicentral distance, take-off angle, geology, SNR, and depth z. We found that the residuals for acceleration, velocity, and displacement commonly did not show any trend with magnitude, epicentral distance, take-off angle, and geology, but exhibited specific trends with SNR and depth z. Kim et al. (2016), Kang et al. (2020), and Kim et al. (2020b) used seismograms with an SNR of 2 or higher, in addition, Kim et al. (2016) and Miao et al. (2018) used their own empirical correlations between V_SZ and V_S_30. However, we used seismograms with an SNR of unity or higher to utilize as much data as possible, as well as the correlation between V_SZ and V_S₃₀ derived from another region. Furthermore, the highly irregular basin structure in the study area might play a role in reducing precision. These factors resulted in the $σ_{res}$ value from velocity seismograms (i.e. 0.49) greater than the standard deviation values of between-site residuals reported by Kim et al. (2016), Miao et al. (2018), Kang et al. (2020), and Kim et al. (2020b) (i.e. 0.39, 0.34, 0.41, and 0.37, respectively).

We added the bias values to $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ for the final estimates. Figure 11 depicts the $V_{S 30}^{mea}$ versus bias-corrected V_S₃₀ estimates. In total, approximately 80% (12/15) of the $V_{S 30}^{mea}$ and $V_{S 30}^{acc}$ values are located between ±100% of each other (Figure 11a). Approximately 93.3% (14/15) of the $V_{S 30}^{mea}$ and $V_{S 30}^{vel}$ values, and 80% (12/15) of the $V_{S 30}^{mea}$ and $V_{S 30}^{dis}$ values are located between ±100% of each other (Figures 11b and c). Based on these validation results, we considered that velocity seismograms were the most appropriate for estimating the V_S₃₀ values.

Figure 11.

Measured V_S30 versus estimated V_S30 (bias-corrected) using (a) acceleration $(V_{S 30}^{acc})$ , (b) velocity $(V_{S 30}^{vel})$ , and (c) displacement seismograms $(V_{S 30}^{dis})$ .

Final V_S₃₀ map

We produced a final V_S₃₀ map by geospatially interpolating the 102 bias-corrected $V_{S 30}^{vel}$ and 54 $V_{S 30}^{mea}$ using the inverse distance weighting (IDW) method, as depicted in Figure 12a. We also attempted using the kriging method but concluded that the map generated by the IDW represented the V_S₃₀ values at the stations better than those generated by the kriging method. To validate the V_S₃₀ values of the final V_S₃₀ map, we performed a landstreamer test along the A-A’ line, as illustrated in Figure 12a. We performed 28 MASW tests along the test line with a length of approximately 6 km, except for a certain section where the test was not available due to obstacles such as a bridge and intersection. We used weight-drops as the active source, as illustrated in Figure 13a, as well as ambient noise as a passive source. Figure 13b depicts the V_S₃₀ values of the final V_S₃₀ map and the landstreamer test along the A-A’ line. It is worth noting that the V_S₃₀ values from the final map agree well with those from the landstreamer test.

Figure 12.

Maps of (a) the final V_S₃₀, (b) topography, and (c) geology produced by Sohn and Son (2004) in the study area. The faults identified (solid lines) and estimated (dashed lines) by Song et al. (2015) are presented in (a). The line for land streamer test (A-A’) is presented in (a).

Figure 13.

(a) Photograph of landstreamer test with a weight drop, and (b) V_S₃₀ values by the landstreamer test and the final V_S₃₀ map along the landstreamer test line (A-A’) presented in Figure 12a.

As an additional validation effort, we compared this V_S₃₀ map with the topography and geology maps produced by Sohn and Son (2004). It can be noted that V_S₃₀ values are smaller than 300 m/s in the center of the basin and flatlands, whereas they are greater than 500 m/s in the mountainous regions outside the basin (see Figure 12b). The center of the basin consists of Quaternary deposits, where the V_S₃₀ values are small, as illustrated in Figure 12c. It can be observed that the V_S₃₀ values are generally greater than 600 m/s in the Cretaceous and Tertiary volcanic rocks. The V_S₃₀ was estimated to be approximately 300–760 m/s in Cretaceous granites (Khgr). Tertiary sedimentary deposits (Te1, Te2, and Te3) generally have small V_S₃₀ values but also have values greater than 500 m/s. In particular, within the Te2 unit, zones with sharp V_S₃₀ contrasts can be observed. It is possible that these contrast zones are attributed to the existence of the estimated faults. Song et al. (2015) identified faults in and around the Miocene Pohang Basin, as depicted in Figure 12a. The zigzag-shaped western border fault (F7) of the basin was traced through fieldwork, while the intrabasinal faults (F1–F6) are buried faults that were estimated based on the depth distribution of the basin floor using deep drilling borehole data. The estimated faults (F1, F4, and F5) in Figure 12a roughly coincide with the V_S₃₀ contrast. Therefore, this V_S₃₀ map can be used to select trench locations for detailed fault investigations.

Conclusions

The 2017 M_L 5.4 Pohang earthquake triggered by the enhanced geothermal system highlighted the importance of identifying faults and characterizing the Miocene Pohang Basin in Korea. The Geophysics Research Group, Pusan National University, deployed 122 temporary stations within the basin from 3 January to 14 February 2019, which can be considered a significantly dense array. We utilized 366 velocity seismograms recorded by these temporary stations as well as from one regular station during the 18 microearthquakes with M_L ranging from −1.11 to 1.7 and estimated V_S₃₀ at the 102 stations in Pohang City based on the P-wave seismogram method. We only used seismogram data with an SNR > 1 and take-off angle > 40°.

Following the standardized procedure, we rotated the velocity seismograms recorded in the two horizontal directions to the radial direction, identified the maximum value of the initial pulse ( ${\overset{\cdot}{U}}_{R}$ and ${\overset{\cdot}{U}}_{Z}$ ), and estimated the value of V_S_Z, which was then converted to V_S₃₀ using empirical correlations. For the depth extent of the V_SZ, we measured the pulse durations of the initial vertical components of the surface P-wave seismograms used in this study. In addition to the velocity seismograms, we estimated V_S₃₀ using acceleration and displacement seismograms. There were fair correlations between $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ , and the SNR values of 63.8% and 56.9% of the seismograms showed that the SNR^vel values were greater than SNR^acc and SNR^dis, respectively.

To validate the V_S₃₀ estimates, we acquired 22 existing V_S profiles and performed MASW to obtain an additional 32 V_S profiles within the study area. We averaged the V_S₃₀ estimates from multiple seismograms for each of the 102 stations and calculated the residuals between the measured and estimated V_S₃₀ values using the 15 V_S profile data measured at locations closer than 200 m from the seismic stations. The $σ_{res}$ values for $V_{S 30}^{acc}$ , $V_{S 30}^{vel}$ , and $V_{S 30}^{dis}$ are approximately 0.65, 0.49, and 0.81, respectively. Therefore, we concluded that $V_{S 30}^{vel}$ is the most appropriate. We also calculated the bias of residuals, which was used to correct $V_{S 30}^{vel}$ . Finally, using 54 $V_{S 30}^{mea}$ values and 102 bias-corrected $V_{S 30}^{vel}$ values, we proposed a V_S₃₀ map for the Pohang Basin. We observed that the proposed V_S₃₀ map was consistent with the V_S₃₀ values measured from the landstreamer test, local topography, and geology. In particular, it was possible to identify zones with sharp V_S₃₀ contrast, which might be associated with the existence of faults.

Through this study, we demonstrated that seismograms from microearthquakes can be used for the P-wave seismogram method to estimate V_S. We also proposed a V_S₃₀ map for the basin using the seismograms recorded at densely deployed seismic stations. The findings of this study and the proposed V_S₃₀ map are expected to be useful for seismic hazard assessments and fault investigations in Korea and elsewhere.

Footnotes

Acknowledgements

The authors thank Hwanwoo Seo, Jaesung Kim, and Dr. Sinhang Kang for supporting the MASW landstreamer tests. The authors thank the two anonymous reviewers for their insightful feedback, which helped to significantly improve the manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the KMA Research Development Program (KMI2018-02810-2).

ORCID iD

Junyoung Lee

Byungmin Kim

References

Aki

Richards

(2002) Quantitative Seismology. Herndon, VA: University Science Books.

Allen

(1982) Automatic phase pickers: Their present use and future prospects. Bulletin of the Seismological Society of America 72(6): S225–S242.

Boore

Thompson

Cadet

(2011) Regional correlations of V_S₃₀ and velocities averaged over depths less than and greater than 30 meters. Bulletin of the Seismological Society of America 101(6): 3046–3059.

Catchings

Goldman

Trench

Buga

Chan

Criley

Strayer

(2017) Shallow-depth location and geometry of the Piedmont Reverse splay of the Hayward Fault, Oakland, California. Open-file report no. 2016-1123. Available at: https://pubs.usgs.gov/of/2016/1123/ofr20161123.pdf (accessed 15 March 2021).

Catchings

Hernandez

Goldman

Chan

Sickler

Olson

Criley

(2020) 2018 US Geological Survey–California Geological Survey fault-imaging surveys across the Hollywood and Santa Monica faults, Los Angeles County, California. Open-file report no. 2020-1049. Available at: https://pubs.usgs.gov/of/2020/1049/ofr20201049.pdf (accessed 15 March 2021).

Duffy

Campbell

Finnemore

Gomez

(2014) Defining fault avoidance zones and associated geotechnical properties using MASW: A case study on the Springfield Fault, New Zealand. Engineering Geology 183: 216–229.

Ellsworth

Giardini

Townend

Shimamoto

(2019) Triggering of the Pohang, Korea, Earthquake (Mw 5.5) by enhanced geothermal system stimulation. Seismological Research Letters 90(5): 1844–1858.

Geological Society of Korea (2019) Summary Report of the Korean Government Commission on Relations Between the 2017 Pohang Earthquake and EGS Project. Seoul, South Korea: Geological Society of Korea, p. 127.

Grigoli

Cesca

Rinaldi

Manconi

López-Comino

Clinton

Westaway

Cauzzi

Dahm

Wiemer

(2018) The November 2017 M_W 5.5 Pohang earthquake: A possible case of induced seismicity in South Korea. Science 360: 1003–1006.

10.

Harrington

Brodsky

(2009) Source duration scales with magnitude differently for earthquakes on the San Andreas fault and on secondary faults in Parkfield, California. Bulletin of the Seismological Society of America 99(4): 2323–2334.

11.

Hong

T-K

Lee

Kim

Hahm

I-K

Woo

Park

(2017) The 12 September 2016 M_L5.8 midcrustal earthquake in the Korean Peninsula and its seismic implications. Geophysical Research Letters 44(7): 3131–3138.

12.

Ivanov

Miller

Lacombe

Johnson

Lane

Jr (2006) Delineating a shallow fault zone and dipping bedrock strata using multichannal analysis of surface waves with a land streamer. Geophysics 71(5): A39–A42.

13.

Kim

(2021) Evaluation of liquefaction potentials based on shear wave velocities in Pohang City, South Korea. International Journal of Geo-Engineering 12(1): 3.

14.

Kalkan

(2016) An automatic P-phase arrival-time picker. Bulletin of the Seismological Society of America 106(3): 971–986.

15.

Kanamori

Brodsky

(2001) The physics of earthquakes. Physics Today 54(6): 34–40.

16.

Kang

Kim

Bae

Lee

Kim

(2019a) Earthquake-induced ground deformations in the low-seismicity region: A case of the 2017 M5.4 Pohang, South Korea, earthquake. Earthquake Spectra 35(3): 1235–1260.

17.

Kang

Kim

Cho

Lee

Kim

Bae

Sun

(2019b) Ground-motion amplifications in small-size hills: Case study of Gokgang-ri, South Korea, during the 2017 M_L 5.4 Pohang earthquake sequence. Bulletin of the Seismological Society of America 109(6): 2626–2643.

18.

Kang

Kim

Park

H-J

Lee

(2020) Automated procedure for estimating V_S30 utilizing P-wave seismograms and its application to Japan. Engineering Geology 264: 105388.

19.

Kikuchi

Ishida

(1993) Source retrieval for deep local earthquakes with broadband records. Bulletin of the Seismological Society of America 83(6): 1855–1870.

20.

Kim

Hashash

YMA

Rathje

Stewart

Somerville

Kottke

Silva

Campbell

(2016) Subsurface shear wave velocity characterization using P-wave seismograms in Central and Eastern North America. Earthquake Spectra 32(1): 143–169.

21.

Kim

Lee

Y-J

Kang

Yun

N-R

Lee

(2020a) Building damage caused by the 2017 M5.4 Pohang, South Korea, earthquake, and effects of ground conditions. Journal of Earthquake Engineering. Epub ahead of print 16 July. DOI: 10.1080/13632469.2020.1785585.

22.

Kim

H-S

Sun

C-G

Cho

H-I

(2018a) Geospatial assessment of the post-earthquake hazard of the 2017 Pohang earthquake considering seismic site effects. ISPRS International Journal of Geo-Information 7(9): 375.

23.

Kim

Cho

H-I

(2020b) Shear wave velocity estimation in Korea using P-wave seismograms. KSCE Journal of Civil Engineering 24(12): 3650–3658.

24.

Kim

K-H

Ree

J-H

Kim

Kang

Seo

(2018b) Assessing whether the 2017 M_w 5.4 Pohang earthquake in South Korea was an induced event. Science 360(6392): 1007–1009.

25.

Kim

K-H

Seo

Han

Kwon

Kang

Ree

J-H

Kim

Liu

(2020c) The 2017 M_L 5.4 Pohang earthquake sequence, Korea, recorded by a dense seismic network. Tectonophysics 774: 228306.

26.

Korea Meteorological Administration (KMA) (2018) 2017 Earthquake Bulletin. Available at: https://www.kma.go.kr/communication/webzine/earthquakeyearly.jsp

27.

Lahr

(1999) HYPOELLIPSE: A computer program for determining local earthquake hypocentral parameters, magnitude, and first-motion pattern (Y2K compliant version). Open-file report no. 99-23. Reston, VA: US Geological Survey (accessed 15 March 2021).

28.

Lee

K-K

Ellsworth

Giardini

Townend

Shimamoto

Yeo

I-W

Kang

T-S

Rhie

Sheen

D-H

Chang

Woo

J-U

Langenbruch

(2019) Managing injection-induced seismic risks. Science 364(6442): 730–732.

29.

Lee

Callahan

Shelly

Iqbal

Kribbs

(2010) MASW survey identifies causes of sink activity along I-476 (Blue Route), Montgomery County, Pennsylvania. In: Proceedings of the GeoFlorida 2010, Orlando, FL, 20–24 February, pp. 1350–1359. Reston, VA: ASCE.

30.

Martínez-Pagán

Navarro

Pérez-Cuevas

Alcalá

García-Jerez

Vidal

(2018) Shear-wave velocity structure from MASW and SPAC methods: The case of Adra town, SE Spain. Near Surface Geophysics 16(3): 356–371.

31.

Miao

Shi

Wang

S-Y

(2018) Estimating near-surface shear wave velocity using the P-wave seismograms method in Japan. Earthquake Spectra 34(4): 1955–1971.

32.

National Disaster Management Research Institute (NDMI) (2018) Liquefaction Hazard Evaluation in Pohang, South Korea. Ulsan, South Korea: NDMI.

33.

Somerville

(2014) Estimating subsurface shear velocity with radial to vertical ratio of local P waves. Seismological Research Letters 85: 82–90.

34.

ParkSEIS (2016) ParkSEIS, version 2.0. Shelton, CT: Park Seismic LLC.

35.

Singh

Pacheco

Ordaz

Kostoglodov

(2000) Source time function and duration of Mexican earthquakes. Bulletin of the Seismological Society of America 90(2): 468–482.

36.

Sohn

Son

(2004) Synrift stratigraphic geometry in a transfer zone coarse-grained delta complex, Miocene Pohang Basin, SE Korea. Sedimentology 51(6): 1387–1408.

37.

Song

Son

Sohn

Han

Shinn

Kim

(2015) A study on potential geologic facility sites for carbon dioxide storage in the Miocene Pohang Basin, SE Korea. Journal of the Geological Society of Korea 51: 53–66.

38.

Thompson

Reyes

(2017) GISMO—A seismic data analysis toolbox for MATLAB (software). Available at: http://geoscience-community-codes.github.io/GISMO/(accessed 29 March 2021).

39.

Waldhauser

Ellsworth

(2000) A double-difference earthquake location algorithm: Method and application to the Northern Hayward Fault, California. Bulletin of the Seismological Society of America 90(6): 1535–1368.

40.

Xia

Miller

Park

(1999) Estimation of near-surface shear-wave velocity by inversion of Rayleigh waves. Geophysics 64(3): 691–700.

41.

Yeo

I-W

Brown

MRM

Lee

(2020) Causal mechanism of injection-induced earthquakes through the M_w 5.5 Pohang earthquake case study. Nature Communications 11(1): 2614.

Shear-wave velocity map for Pohang Basin,South Korea,based on the P-wave seismogram method

Abstract

Keywords

Introduction

Methodology

Signal processing

P-wave seismogram method

Ground motion records

V S30 estimation

Examples

All estimates

Validation of VS30 estimations

Vs measurements

Comparison between measured and estimated VS30 values

Final VS30 map

Conclusions

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

ORCID iD

References

V _S30 estimation

Validation of V_S30 estimations

Comparison between measured and estimated V_S₃₀ values

Final V_S₃₀ map