**3. Unified velocity model of NIED**

To theoretically reproduce the observed hSAF and vSAF from GIT we need a velocity structure at each site from the seismological bedrock to the surface because they are the spectral ratios with respect to the outcrop spectra on the seismological bedrock. Note again that the seismological bedrock here is the surface of the crust on which we can assume no site amplification, whose S-wave velocity should be equal to or higher than 3 km/s. We have been delineating velocity structures in the deeper- and shallower-parts separately, primarily because we need to use different methods to explore the velocity structures in different depths. The boundary between them is the so-called engineering bedrock, whose Vs would be in between 350 m/s to 450 m/s. This is so beause we are using plenty of borehole information to constrain velocity structures in the shallower-parts, for which engineers need to gather information for their construction works. They usually need information only down to the layer with Vs in between 350 m/s to 450 m/s. However, it is apparent that higher-mode contributions of reverberated S- and P-waves within the whole basin above the seismological bedrock should show up in the frequency range higher than the fundamental peak frequency [15]. Therefore, we need a unified velocity model that integrates both shallower- and deeper-parts.

To that end, the National Research Institute for Earth Science and Disaster Resilience (NIED) has developed a unified velocity model (UVM) by integrating shallower- and deeper-parts of the structures above the seismological bedrock in the Kanto and Tokai regions [16–18]. The procedure to develop the model was based on the "concept of creating a subsurface structure model" released by the

**123**

**Figure 2.**

*S-Wave Site Amplification Factors from Observed Ground Motions in Japan: Validation…*

government agency, the Headquarters for Earthquake Research Promotion (HERP) [19]. Details of the procedure can be found in the papers referenced above, but the

An initial model of the shallow structure from the ground surface to the engineering bedrock was created based on existing studies and continuously collected SPT values in the boring data. Meanwhile, an initial model of the deep structure from the engineering bedrock down to the seismological bedrock was created based on the velocity models developed in existing studies by HERP. Then, an initial UVM was created by connecting them at the engineering bedrock. Next, the initial UVM was adjusted by using S-wave velocity structures at the strong-motion sites in the regions and those of spatially uniform and dense array microtremor explorations conducted as a part of Japan's national Strategic Innovation Promotion project. Finally, the adjusted UVM was verified by using earthquake data at the strong-

Examples of the important features of the resultant UVM are shown in **Figures 2** and **3**. **Figure 2** shows the depth to the seismological bedrock with the Vs of 3.1 km/s (Z3.1) in the Tokai region, while **Figure 3** shows the depth to the engineering bedrock with the Vs of 0.35 km/s (Z0.35), which is the interface between the shallower- and deeper-parts of the UVM in the Tokai region [18]. A

Please note that in the following sections when we calculate theoretical onedimensional (1D) S-wave SAF, we use the following Q values for intrinsic and

similar map can be seen for the Kanto region in Senna et al. [16, 17].

*Depth contour in meters to the seismological bedrock with Vs of 3.1 km/s after [18].*

*DOI: http://dx.doi.org/10.5772/intechopen.95478*

motion sites of NIED.

scattering attenuation:

following is a brief description of the procedure.

### *S-Wave Site Amplification Factors from Observed Ground Motions in Japan: Validation… DOI: http://dx.doi.org/10.5772/intechopen.95478*

government agency, the Headquarters for Earthquake Research Promotion (HERP) [19]. Details of the procedure can be found in the papers referenced above, but the following is a brief description of the procedure.

An initial model of the shallow structure from the ground surface to the engineering bedrock was created based on existing studies and continuously collected SPT values in the boring data. Meanwhile, an initial model of the deep structure from the engineering bedrock down to the seismological bedrock was created based on the velocity models developed in existing studies by HERP. Then, an initial UVM was created by connecting them at the engineering bedrock. Next, the initial UVM was adjusted by using S-wave velocity structures at the strong-motion sites in the regions and those of spatially uniform and dense array microtremor explorations conducted as a part of Japan's national Strategic Innovation Promotion project. Finally, the adjusted UVM was verified by using earthquake data at the strongmotion sites of NIED.

Examples of the important features of the resultant UVM are shown in **Figures 2** and **3**. **Figure 2** shows the depth to the seismological bedrock with the Vs of 3.1 km/s (Z3.1) in the Tokai region, while **Figure 3** shows the depth to the engineering bedrock with the Vs of 0.35 km/s (Z0.35), which is the interface between the shallower- and deeper-parts of the UVM in the Tokai region [18]. A similar map can be seen for the Kanto region in Senna et al. [16, 17].

Please note that in the following sections when we calculate theoretical onedimensional (1D) S-wave SAF, we use the following Q values for intrinsic and scattering attenuation:

**Figure 2.**

*Depth contour in meters to the seismological bedrock with Vs of 3.1 km/s after [18].*

*Earthquakes - From Tectonics to Buildings*

**3. Unified velocity model of NIED**

*JMA Shindokei network by GIT after [7].*

**Figure 1.**

To theoretically reproduce the observed hSAF and vSAF from GIT we need a velocity structure at each site from the seismological bedrock to the surface because they are the spectral ratios with respect to the outcrop spectra on the seismological bedrock. Note again that the seismological bedrock here is the surface of the crust on which we can assume no site amplification, whose S-wave velocity should be equal to or higher than 3 km/s. We have been delineating velocity structures in the deeper- and shallower-parts separately, primarily because we need to use different methods to explore the velocity structures in different depths. The boundary between them is the so-called engineering bedrock, whose Vs would be in between 350 m/s to 450 m/s. This is so beause we are using plenty of borehole information to constrain velocity structures in the shallower-parts, for which engineers need to gather information for their construction works. They usually need information only down to the layer with Vs in between 350 m/s to 450 m/s. However, it is apparent that higher-mode contributions of reverberated S- and P-waves within the whole basin above the seismological bedrock should show up in the frequency range higher than the fundamental peak frequency [15]. Therefore, we need a unified

*Observed horizontal site amplification factor, hSAF, extracted from strong motions at K-NET, KiK-net, and* 

velocity model that integrates both shallower- and deeper-parts.

To that end, the National Research Institute for Earth Science and Disaster Resilience (NIED) has developed a unified velocity model (UVM) by integrating shallower- and deeper-parts of the structures above the seismological bedrock in the Kanto and Tokai regions [16–18]. The procedure to develop the model was based on the "concept of creating a subsurface structure model" released by the

**122**

**Figure 3.** *Depth contour of the engineering bedrock with Vs of 350 m/s after [18].*

$$\begin{array}{l} Q = Q\_0 \ast \mathbf{0.5} \qquad \text{if } f \le \mathbf{0.5Hz} \\ Q = Q\_0 \ast f \quad \text{if } \mathbf{0.5Hz} \mathbf{<} f < \mathbf{5Hz} \\ Q = Q\_0 \ast \mathbf{5} \qquad \text{if } \mathbf{5Hz} \le f \end{array} \tag{1}$$

**125**

**Table 1.**

*S-Wave Site Amplification Factors from Observed Ground Motions in Japan: Validation…*

 2400 900 2050 180 2400 950 2100 190 2500 1000 2100 200 2500 1100 2150 220 2600 1200 2150 240 2700 1300 2200 260 3000 1400 2250 280 3200 1500 2250 300 3400 1600 2300 320 3500 1700 2300 340 3600 1800 2350 360 3700 1900 2350 380 3800 2000 2400 400 4000 2100 2400 420 5000 2700 2500 540 4600 2900 2550 580 5500 3100 2600 620 5500 3200 2650 640

**) Q0**

**No. Vp (m/s) Vs (m/s) Density (kg/m3**

Vs of Ldt, then we insert three layers with a gentle gradient of increasing Vs. **Table 1** shows the parameters of layers assumed commonly in the deeper part. The bedrock S-wave velocity of UVM, 3,200 m/s, is close enough to that of the hypothesized

As mentioned in the previous section, the UVM of NIED is considered to be the most reliable velocity model for the strong motion simulation because the UVM combines all the available geophysical information to date related to the velocity structures from the ground surface to the seismological bedrock as densely sampled as possible. However, the actual S-wave SAF at a specific site, as shown in **Figure 1**, can be significantly different from the theoretical one calculated by a simple 1D S-wave multi-reflection theory for a stack of layers [20–22]. To see the difference, we plot in **Figure 4** comparisons of the 1D theoretical hSAF with the observed hSAF at the same four sites in **Figure 1**. We calculate the theoretical hSAF as the 1D soil response on the surface of a combined velocity structure of the shallower- and deeper-parts with respect to the outcrop of the seismological bedrock motion (i.e., twice of the input at the bottom of the deeper part). Except for the site SZOH31, the theory tends to

The major reasons for discrepancy are twofold; one is due to an inevitable inaccuracy of the derived velocity structure, and the other is due to a too simplistic assumption of the 1D horizontally-flat layered model. In the former there are two possible causes; one is the inaccuracy of the referenced values to delineate the

bedrock of 3,450 m/s in GIT so we can compare both SAFs directly.

**4. Observed and theoretical SAF**

*Assumed layer profiles for the deeper part of UVM after [18].*

underestimate the observation.

*DOI: http://dx.doi.org/10.5772/intechopen.95478*

The *Q*0 values for the deeper part are listed in **Table 1**, while those for the shallower part we assume *Q*0 = Vs/10.

To connect the bottommost layer of the shallower part Lsb with the topmost layer of the deeper part Ldt, we prioritize the shallower part if the depth Lsb is deeper than Ldt. If there is a gap between the two depths and Vs of Lsb is equal to or larger than Vs of Ldt, we extend Lsb down to Ldt. If Vs of Lsb is much smaller than



*S-Wave Site Amplification Factors from Observed Ground Motions in Japan: Validation… DOI: http://dx.doi.org/10.5772/intechopen.95478*

#### **Table 1.**

*Earthquakes - From Tectonics to Buildings*

0 0 0

*Depth contour of the engineering bedrock with Vs of 350 m/s after [18].*

part we assume *Q*0 = Vs/10.

**Figure 3.**

0.5

<

5Hz

(1)

**) Q0**

if 0.5Hz i

The *Q*0 values for the deeper part are listed in **Table 1**, while those for the shallower

5 f 5Hz

To connect the bottommost layer of the shallower part Lsb with the topmost layer of the deeper part Ldt, we prioritize the shallower part if the depth Lsb is deeper than Ldt. If there is a gap between the two depths and Vs of Lsb is equal to or larger than Vs of Ldt, we extend Lsb down to Ldt. If Vs of Lsb is much smaller than

 1600 350 1850 70 1600 400 1850 80 1700 450 1900 90 1800 500 1900 100 1800 550 1900 110 2000 600 1900 120 2000 650 1950 130 2100 700 2000 140 2100 750 2000 150 2200 800 2000 160 2300 850 2050 170

=∗ ≤ = ∗ < =∗ ≤

*QQ . f QQ f f Q Q f*

**No. Vp (m/s) Vs (m/s) Density (kg/m3**

05 if Hz

**124**

*Assumed layer profiles for the deeper part of UVM after [18].*

Vs of Ldt, then we insert three layers with a gentle gradient of increasing Vs. **Table 1** shows the parameters of layers assumed commonly in the deeper part. The bedrock S-wave velocity of UVM, 3,200 m/s, is close enough to that of the hypothesized bedrock of 3,450 m/s in GIT so we can compare both SAFs directly.
