Figure 4 presents V _se–H scatter plots for the four studied regions, together with histograms of the H and V _se values. The H values of XJ sites are generally <20 m, while the V _se values are at a comparatively high level (250 –450 m/s), indicating that most sites in XJ are covered by thin and hard soil (e.g., sand and silty clay). In contrast, CM sites generally show an approximately uniform distribution of H values in the range of 0 –100 m (average: ∼50 m) and lower V _se values in the range of 150 –300 m/s, indicating deep and soft surficial soil. The V _se–H scatter plots present similar distributions for GG and SY. Generally, the ground surface in both regions is covered by an overlying soil layer that is ∼20 m thick with an average V _se value of ∼250 m/s. However, in comparison with the other three regions, the overlying soil layer is very thin (<5 m) at many more sites in GG. In comparison with those sites with a very deep overlying soil layer (i.e., >80 m) in XJ, CM, and SY, some of the sites in CM are characterized by very low V _se (i.e., <150 m/s), and sites with higher V _se values account for a larger proportion of such sites in SY. The above results fully reflect the major differences in H and V _se at the various sites in the four studied regions.

The potential regional dependency in the V _S profiles was investigated based on the boreholes of the Class-II sites in the four regions. Figure 5 shows the variation of V _S with depth for each Class-II site, as well as the V _S median and the one standard deviation range for each region. Very obvious differences are evident in the V _S profiles of the Class-II sites in the four regions. The median V _S profiles for the four regions are also plotted in Figure 5 for comparison purposes. The sites in the four regions show some similarities in terms of the characteristics of the variation in median V _S with depth, i.e., a sharp increase in V _S in shallow soil layers from the ground surface to a critical depth, and notably slower increase in deeper soil layers beneath the critical depth. The differences in the median V _S profiles for the four regions were investigated. The median V _S profile in XJ shows a much shallower critical depth (∼13 km) than that in the other three regions (∼20 km). The median V _S values (increasing from ∼220; to ∼480 m/s) in the shallow soil layers in XJ are much larger than those (increasing from ∼200; to ∼400 m/s) in the other regions. The median V _S value in XJ appears to exhibit faster increase in shallow soil depths than that in the other regions. The median V _S values of the deeper soil layers in XJ generally show no obvious increase with depth. Such a phenomenon is also observed in many of the deeper soil layers (>60 m) in CM. In GG and SY, the median V _S values of the deeper soil layers maintain continuous increase with depth, i.e., from 400; to 580 m/s and to 700 m/s, respectively. Our results indicate that it is necessary to consider regionally dependent site amplification effects in seismic design.

Although the Chinese seismic code uses site categories to characterize the effects of local site conditions on ground motions, V _S30 is also an important parameter commonly used to characterize local site amplification. The potential regional dependency of V _S30 was further investigated on the basis of the borehole data from the four studied regions. The V _S30 values were estimated for sites with borehole depths of <30 m using the constant extrapolation model (Wang et al., 2022) if a rock layer was situated at the bottom of the borehole, and using the linear extrapolation model if a soil layer was situated at the bottom of the borehole.

Frequency distributions of the V _S30 values in the four studied regions are plotted in Figure 6. The V _S30 distributions in CM, SY, and GG all approximately follow a lognormal distribution with logarithmic means of 257.2, 299.6, and 360.9 m/s, respectively. The range of one standard deviation in CM, SY, and GG is 186.5–354.7, 215.1–417.3, and 243.4–535.0 m/s, respectively, indicating that the near-surface soil layer in the upper 30 m is on average hardest in GG and softest in CM. The V _S30 values at a certain proportion of sites in GG are measured or estimated to be >800 m/s, corresponding to hard soil or rock sites, while barely any such sites are found in CM. However, the near-surface site conditions in CM could be confirmed to be most similar owing to the narrowest dispersion of V _S30 values, while the widest dispersion in GG indicates greatest discrepancy. Different from the other three regions, the V _S30 values of XJ sites do not follow a lognormal distribution. Instead, the V _S30 values for most sites fall within a wide range of 200 –550 m/s, approximately following a uniform distribution with an average of ∼400 m/s.

The V _S30-related empirical relations are often used to characterize site amplification effects, e.g., in predicting ground motions. Moreover, the empirical relations associated with site categories are also used (e.g., Bindi et al., 2011). However, in such cases, the same amplification effects appear associated with sites belonging to the same site category but with different measured V _S30 values. Therefore, we explored the potential regional dependency of V _S30 values measured or estimated at sites belonging to the same site category. Referring to the seismic code of China and other countries and regions, most sites considered in this study were divided into three categories on the basis of the V _S30 values, i.e., C1 for V _S30 = 180 –360 m/s, C2 for V _S30 = 360 –500 m/s, and C3 for V _S30 = 500 –750 m/s.

The median V _S profiles for sites belonging to each category were computed for each of the four regions and illustrated in Figure 7. Note that the median V _S profiles for sites within category C3 in SY and CM are not presented owing to the limited number of samples which is less than 10% of the total number of sites. Substantial regional dependency is also observed in the median V _S profiles of sites within the same category. To clearly express the regional dependency of median V _S profiles of each category, we defined three trends to describe the regional dependency between any two regions considered in this study, i.e., 1) parallel median V _S profiles between two regions that indicate similar V _S variation with depth and greater V _S values in one region in comparison with those in another; 2) intersecting median V _S profiles of two regions that indicate dissimilar V _S variation with depth and greater (lower) V _S in the upper (lower) soil layers in one region in comparison with those in another; and 3) overlapping median V _S profiles at certain depths for both regions indicating consistent V _S values. The three trends between any two regions for each V _S30-based site category are listed in Table 2. For sites belonging to category C1, the median V _S profiles between two regions are mainly manifested as a parallel trend, while a trend of intersection appears for XJ–GG and XJ–SY pairings. However, for sites belonging to categories C2 and C3, the median V _S profiles between two regions are predominantly shown as intersection trends, while only the XJ–CM pairing in category C2 shows as an overlapping trend. It is worth noting that the median V _S profiles for categories C2 show an intersection trend with the other three regions and categories C3 intersect with XJ in GG. The median V _S values of the upper soil layers in GG are relatively small but relatively large in the lower soil layers, reflecting the higher impedance ratio and the corresponding stronger site seismic responses in GG. The trends of the median V _S profiles for the same V _S30-based category between any two regions fully demonstrate the regional dependency of the V _S30 values measured or estimated at sites belonging to the same site category. Even if the V _S30 values among the four regions are similar, the V _S profiles can differ substantially, which could cause huge differences in site seismic responses.