- Full paper
- Open Access
Subsurface velocity structure and site amplification characteristics in Mashiki Town, Kumamoto Prefecture, Japan, inferred from microtremor and aftershock recordings of the 2016 Kumamoto earthquakes
© The Author(s) 2018
- Received: 30 March 2018
- Accepted: 28 June 2018
- Published: 11 July 2018
- 2016 Kumamoto earthquake
- Ground motion
- Site amplification
- Surface wave
- Phase velocity
Field surveys after the mainshock also showed that the cause of the varying damage distribution could be explained by different site conditions of this area. According to the landform classification map for flood control published by the Geospatial Information Authority of Japan, the distribution of damaged buildings corresponds well to the river terrace between R28 and the Akitsu River, whereas almost no buildings collapsed in the floodplain and old river channel along the Akitsu riverside (e.g., Yamada 2017; Yamada et al. 2017a, c). These findings indicate that the building damage pattern is strongly affected by the site conditions associated with local subsurface geology. The results of single-point microtremor observations in this area (e.g., Kagawa et al. 2017; Kawase et al. 2017; Nagao et al. 2017) showed a clear correlation between the patterns of peak frequency in the horizontal-to-vertical (H/V) spectral ratios and the landform characteristics. Yamada et al. (2017c) performed further microtremor observations with miniature arrays at 108 sites in Mashiki and clarified the differences of Rayleigh wave dispersion characteristics between the damaged and undamaged zones. They found that an extremely low S wave velocity (Vs) layer (< 100 m/s) exists at very shallow depth beneath Mashiki. The estimated thickness of this layer is less than several meters in the severely damaged zones, where the dominant frequencies of microtremor H/V spectral ratio are higher than 2 Hz. In contrast, in the floodplain of the Akitsu River where almost no buildings were collapsed, the low-velocity layer is thicker than 10 m and the dominant H/V frequencies were around 1 Hz.
The predominant frequency of soil layers is fundamental information in seismic microzonation and a reasonable indicator that relates the local site condition to earthquake damage. Past observations of damaging earthquakes in Japan showed that the strong ground motions in the frequency range between 0.5 and 1 Hz greatly affect wooden buildings (e.g., Kitagawa and Hiraishi 2004; Sakai et al. 2008). The natural frequency of low-rise wooden buildings is between 2 and 10 Hz (e.g., Sugino et al. 2016a), but the frequency may shift to lower values (around 1 Hz) when they are subjected to strong shaking. Therefore, larger damage was expected along the Akitsu river where the frequencies around 1 Hz are dominant. However, the damage investigations showed very different results with relatively light damage close to the river and more severe damage associated with sites of higher natural frequencies that were located farther from the river. Besides the local differences in site conditions, there are several other possible factors that could account for the damage concentration in Mashiki; the construction years of buildings (e.g., Mori et al. 2017; Sugino et al. 2016b; Yamada 2017), types of foundations in the buildings (e.g., Mori et al. 2017), and/or surface fault ruptures (e.g., Suzuki et al. 2016). There were no strong-motion records of the larger events in the heavily damaged zones so that aftershock data recorded in both the damaged and undamaged zones are used to help understand the difference of seismic response.
Recent studies have tried to explain this unusual site amplification in Mashiki using nonlinear site response analyses, assuming that frequencies between 0.5 and 1 Hz were dominant in the heavily damaged zones and lower frequencies were dominant in the less damaged zone. Yamada et al. (2017c) calculated the seismic response for both damaged and undamaged zones of Mashiki using shallow seismic velocity structure models obtained from microtremor observations with typical nonlinear properties of sand and clay. Shingaki et al. (2017) and Yoshimi et al. (2017) determined dynamic properties of surface soils including volcanic ash clay, using actual soil samples in Mashiki, and Nakagawa et al. (2017) performed time-history response analysis for the mainshock at two sites in Mashiki using the obtained dynamic deformation properties. For analyzing the ground motions of the mainshock, information from the borehole (252 m in depth) of the closest KiK-net strong-motion observation station, KMMH16 (see the location in Fig. 1b) was used. The station is operated by the National Research Institute for Earth Science and Disaster Resilience (NIED) and located about 1 km to the northeast of the severely damaged area. Velocity models obtained from the borehole logging are well constrained for the shallower depths (less than a few tens of meters) but have lower resolution for the deeper structure. Therefore, it is important to evaluate the deep sedimentary structure in the damaged area.
To investigate the differences of ground motion characteristics in the heavily and less damaged zones as well as station KMMH16, we conducted aftershock and microtremor (ambient noise) observations in and around the damaged zones of Mashiki (see the observation locations in Fig. 1b). Figure 1b also shows the locations of aftershock observation stations deployed by Yamanaka et al. (2016) (gray triangles: MK01, MK02, MK05 and MK09) and a seismic intensity station deployed by JMA (white triangle) just after the mainshock. Our aftershock observation sites were installed in the heavily damaged zones, while the existing temporary aftershock stations were located outside of this zone. In this paper, we show the differences of the ground motion characteristics between the contrasting zones and propose a velocity structure model of the entire area for seismic response analysis.
Temporary stations for aftershock observation of the 2016 Kumamoto earthquake
We distributed the sensors in areas with different building damage ratios; sites S1 and S2 were located in a zone with no collapsed buildings near the Akitsu riverside, and the other sites were located in the damaged zones. Figure 1b also shows the ratio of the totally collapsed wooden buildings that correspond to damage grade D5 (story failure) in the damage scale proposed by Okada and Takai (1999). The sensor at site S6 was deployed at a location where 77% of the wooden buildings were collapsed. The distance between site S7 and the station MTO was about 30 m so that we will discuss the ground motion differences between the mainshock observation station and the other temporary stations in the latter part. We excluded the data at site S4 in the following analyses due to a possible problem in the sensor cable, although signals related to earthquake ground motions were recorded.
We also used data from the surface and borehole components of KiK-net station KMMH16 and co-located Hi-net station M.MSIH. We confirmed that the frequency responses from 0.14 to 20 Hz were very consistent for the two instruments in the borehole (see an example in Additional file 1: Fig. S1).
Aftershock data on July 12, 2016
List of the detected events
Origin time (JST)
Characteristics of the site amplifications
We also derived earthquake radial-to-vertical (R/V) spectral ratios using the data of event EQ18 (Mj 4.3 and MW 4.2) that contain surface waves in the coda. After rotation of the two horizontal components (NS and EW) to radial and transverse components with respect to each source-sensor azimuth, we selected 20.48 s segments from 10 s after the S wave arrivals. The spectral ratios of radial-to-vertical motions that would correspond to Rayleigh wave ellipticity were computed at each site. Here, the same smoothing filter as the microtremor H/V calculations above was applied. Figure 5 also shows the derived earthquake R/V ratios at each observation site. They show peak frequencies similar to the H/V spectra and the predominant peaks (ratios of 30 or greater) around 1–2 Hz are also seen at sites S1 and S2. The peaks at 1–2 Hz are also recognized at sites S3, S5, S6 and S8, but the ratios are substantially smaller (between 5 and 20) compared to those in the less damaged zone and other predominant peaks with larger values are seen at the higher frequency range (> 2 Hz).
Besides the predominant peaks at frequencies higher than 1 Hz, there are distinct peaks at 0.4 Hz for all the spectra, especially at S5–S8. This peak can also be observed in the microtremor H/V spectra with smaller amplitudes. The common dominant peaks of the microtremor H/V and earthquake R/V ratios at this lower frequency are possibly related to the response of deep sedimentary layers beneath the area, which will be discussed in a later section.
Rayleigh wave phase velocity
We also estimated the Rayleigh wave phase velocities from aftershock data (event EQ18) based on the semblance analysis of the S-coda wave (Fukumoto et al. 2004), assuming that the coda wave is dominated by surface waves. The phase velocities were estimated at frequencies from 0.2 to 0.9 Hz with an interval of 0.01 Hz, using narrow-band velocity waveforms (0.9 and 1.1 times the target frequency for the low-frequency and the high-frequency cutoffs, respectively) in the vertical component. The length of the time window for the analysis is twice the reciprocal of the target frequency, and the window was shifted every 0.5 s throughout the dominant wave train. For each time window, the apparent velocity as well as incident angle was calculated. The derived apparent velocities with sufficient semblance values (≥ 0.9) were averaged, and the phase velocity for each frequency was determined.
Figure 6a shows the estimated dispersion curve for the Rayleigh wave phase velocity using both microtremor and event data. The error bars indicate that the standard deviations are low for the entire frequency range, except around 0.7 Hz. The estimated dispersion curves of Rayleigh wave phase velocity from the miniature microtremor arrays (Yamada et al. 2017c) at around sites S1 (array A059), S6 (array A026), S7 (array A001) and station KMMH16 (array A000) are also shown in the figure. The estimated phase velocities vary with the sites but are less than 1000 m/s at the target frequencies of the miniature microtremor surveys (> 1.7 Hz), which is consistent with our phase velocity results. For the semblance analysis, we obtained dispersive characteristics of phase velocities in the frequency range between 0.24 and 0.86 Hz. The estimated phase velocities from microtremor and event data also show similar continuous characteristics for the common frequency range, suggesting that our estimations are reasonable.
S wave velocity structure model beneath KiK-net station KMMH16
Seismic velocity structures at KiK-net Mashiki station (KMMH16)
There are several previous velocity structure models in this area. The P–S logging model at KiK-net station KMMH16 has information down to 252 m depth. Goto et al. (2016) tuned the KiK-net logging model (hereinafter called GOTO model; see Table 3) using the surface/borehole spectral ratio of the small event data. They inverted for Vs values of the top five layers to fit the surface/borehole spectral ratio, and used 75% of the Vs in the logging model for the 6th–12th layers due to the limited sensitivity to the deeper structure (Table 3). For deeper structures, the Seismic Hazard Information Station of NIED (J-SHIS: http://www.j-shis.bosai.go.jp) provides a nationwide velocity model based on the borehole logging, reflection and gravity surveys. In the J-SHIS model, the velocity structure beneath the target area consists of six layers; three subsurface layers (600 m/s ≤ Vs ≤ 2100 m/s) along with three upper crustal layers (3100 m/s ≤ Vs ≤ 3400 m/s). Figure 6c shows the theoretical dispersion curves obtained from the KiK-net logging data, GOTO and J-SHIS models, respectively. The KiK-net P–S logging model clearly overestimates the Rayleigh wave velocity for almost the entire frequency range, suggesting the actual Vs values are much lower than those of the logging model. The GOTO model explains the dispersion curve reasonably well at the higher frequency range (3–10 Hz) that corresponds to the target frequency range of miniature microtremor survey at the site (Yamada et al. 2017c), but still underestimates the lower frequency range even after the 75% reduction of Vs. The observed dispersion curve in Fig. 6c suggests that the hard rock layer (Vs > 3000 m/s) as in the J-SHIS model is necessary to explain the low-frequency range of the dispersion curve.
We constructed an initial model for the GA in the following procedure. We directly use the GOTO model for the top five layers, since the theoretical dispersion curve explains the observed dispersion curve at the higher frequency range. We also use the GOTO model as the starting model for the 6th–12th layers in the inversion. Due to the uncertainty of the deeper structure, we use the thicknesses and Vs values of the J-SHIS model for the layers of bedrock (Vs ≥ 2700 m/s). In the preliminary analysis, we found that the Vs = 2700 m/s layer in the P–S logging model (234 m) is too shallow to explain our dispersion curve. Therefore, we assumed that the layer would be deeper than the depth of the borehole. Based on this assumption, we modified the properties of the 13th layer in the GOTO model to those of the 12th layer, and added an extra layer (Vs = 1470 m/s) underneath the borehole with an unknown depth.
In total, we have 18 layers for our initial velocity model: the 1st–5th layers of the GOTO model, 6th–13th layers of the GOTO model for which Vs values will be determined, 14th layer with Vs = 1470 m/s with unknown thickness, and four bedrock layers of the J-SHIS model with Vs = 2700–3400 m/s at the bottom (Table 3). Since there is a trade-off between Vs and thickness of layers, we fix the thickness for the 1st–13th layers and the Vs for the 14th layers just above the bedrock. Other observations such as receiver functions might reduce trade-offs between the velocity and thickness of each layer; however, the small events (M3 and smaller) in our dataset were not appropriate for this method. Therefore, we validate the reliability of the inversion results focusing on the predominant frequency of the theoretical microtremor H/V, as well as the travel-times of body waves between the ground surface and borehole sensor depth in comparison with the GOTO model. The determined parameters for the inversion are Vs values of eight layers and thickness of the 14th layer; the range of the parameter search for Vs is ± 50% of the initial value, and 0–1024 m for the thickness. Since there is no available information on the density profile at station KMMH16, we assigned the parameters based on the relation between Vp and density (Gardner et al. 1974). In the GA optimization, the number of generations that corresponds to the total number of iteration is 500 and the number of population in each generation is 30. Crossover and mutation rates were set to 0.5 and 0.01, respectively. We repeated this optimization process twice by setting the output of the first process as an input of the second process, so that the most optimal values are inside of the range of the parameter search. We used a subroutine program package DISPER80 (Saito 1988) to compute the theoretical dispersion curves for the generated models.
The optimal velocity profile is shown in Fig. 6b and Table 3 with the phase velocities of Rayleigh wave, and theoretical H/V ratios are shown in Fig. 6c, d, respectively. The estimated values for Vs at the 6th–13th layers are substantially smaller than those of the logging model and GOTO model. The estimated depth of the bedrock layer is about 600 m. The comparisons of H/V ratios and phase velocities show that the agreements between the theoretical and estimated phase velocities are substantially improved. Also, the H/V spectral peak around 0.4 Hz is reproduced well by our model, which does not appear in the theoretical H/V spectra of the P–S logging model and GOTO model. Based on the quarter wavelength law, this peak reflects the velocity contrast of the bedrock, suggesting that the bedrock depth can be much deeper than the logging data (234 m).
Earthquake damage and site amplification
As we expected from microtremor geotechnical surveys, our aftershock data show that ground motions around 1 Hz were amplified on the floodplain of the Akitsu River, while the amplification of this frequency range is much less in the damaged zone. This observation is inconsistent with the damage distribution of the mainshock, which shows severe damage away from the river. We need to consider other mechanisms, such as nonlinear site amplification, to explain the strong ground motions in Mashiki.
Comparisons of the velocity model with existing models
Events used for calculation of earthquake R/V ratios at station KMMH16
Origin time (JST)
We would like to compare our velocity model to other existing models. In the nationwide J-SHIS model, the depth of the bottom of the subsurface layers is 602 m, which is consistent with our estimation. Since the shallower part of the structure with low Vs (< 600 m/s) is not included in the J-SHIS model, the shallow velocity was greatly overestimated and the computed dispersion curve of Rayleigh waves beneath Mashiki has a large discrepancy with our observed dispersion curve (Fig. 6c). Also, the peak in the H/V spectrum at 0.4 Hz could not be explained (Fig. 6d). A group in NIED conducted microtremor observations with large-sized arrays after the Kumamoto earthquake and constructed a three-dimensional (3D) structure model of the deep sedimentary layers (Ogawa et al. 2017; Senna et al. 2017a, b). In their model (hereinafter called SIP2017 model), the velocity structure consists of five subsurface layers (350 m/s ≤ Vs ≤ 2100 m/s) and the bedrock layer (Vs = 2700 m/s) that was not included in the J-SHIS model. The velocity profile, theoretical H/V ratio, and dispersion curve of Rayleigh wave phase velocity computed from the SIP2017 model are shown in Fig. 6b–d, respectively. The depth to the bedrock layer in the SIP2017 model is much deeper than the P–S logging model and the consistency between the estimated and theoretical phase velocity characteristics are reasonable in comparison with the J-SHIS model, although the peak frequency of the theoretical H/V is slightly higher than that of our observations. Our velocity model calibrated by the existing Vs profile at the KiK-net KMMH16 site (Goto et al. 2016) and dispersion curve of Rayleigh wave phase velocity can explain the H/V peak reasonably well. In the SIP2017 model the depth to the Vs = 2700 m/s layer (1417 m) is much deeper than our estimation but the depth of the high-Vs layer (Vs = 2100 m/s) above the bedrock layer (597 m), which was not included in our model, is very close to our estimated bedrock depth, indicating that the difference in starting structure models for inversions produced different solutions for the bedrock depth. The exact depth of the bedrock for this area is still controversial, but at least both observations suggest that the depth of Vs > 2000 m/s layer is much deeper than that of the KiK-net logging data (234 m) and is expected to be about 600 m. Our aftershock data and velocity model help improve knowledge of the deep sedimentary structure model in this area.
We conducted aftershock and microtremor observations at eight sites in and around the damaged areas in Mashiki Town, Kumamoto Prefecture, 2 months after the 2016 Kumamoto earthquakes. The aftershock data indicate that site amplifications at approximately 1 Hz are significant during the small events in the less damaged zone where a thick low-velocity surface layer exists, whereas the dominant frequencies are higher than 3 Hz in the damaged zones. These trends are similar to the results from microtremor surveys in the area, indicating that the characteristics of ground motions for the mainshock and small earthquakes show opposite trends in Mashiki. Our observations support the idea that the dominant frequencies of the heavily damaged area were shifted to 1–2 Hz which has a great effect on wooden buildings during the largest foreshock and mainshock, while in the less damaged zone, the 1–2 Hz peaks would be shifted to much lower frequency range, which had less effect on the buildings. The estimated phase velocities of Rayleigh wave using microtremor data as well as the aftershock data (EQ18) are smaller than those of the P–S logging model at the KiK-net station KMMH16. The derived microtremor H/V and earthquake R/V ratios from our data show common dominant peaks around 0.4 Hz, which is possibly related to the response of deeper structure beneath the observation area. We performed a GA inversion for the deep sedimentary structure model with the derived phase velocities and the dominant frequency peaks to estimate an optimal structure model near station KMMH16. The estimated velocity structure model indicates that the Vs values at shallower depths in the target area are substantially smaller compared to existing models. Also, the depth to the bedrock layer might be much deeper (about 600 m), in comparison with the existing model from logging data (234 m). The earthquake R/V spectral ratios at station KMMH16 also show the common peak at 0.4 Hz, suggesting that the similarity of deeper structures between the KiK-net station and our observation area in the severely damaged zone. Therefore, mainshock data recorded in the borehole at station KMMH16 could be directly used as input motions for seismic response analysis in the damaged zone.
TH, MY1, JM, MY2 and KH planned the survey, and TH, MY1, MY2, KH, JM, YF, HS, SF, EN, TO and AF participated in the observations. TH analyzed the observed data and MY1 performed velocity structure inversions. TH, MY1, MY2, KH, JM and YF contributed to interpretations of the results. TH wrote the initial draft of the manuscript and MY1 and JM edited it. All authors read and approved the final manuscript.
We appreciate two reviewers and Dr. Takuji Yamada for their helpful comments to improve the manuscript. We used the ground motion data from KiK-net and Hi-net Mashiki stations and referred a survey report of P–S logging at the site provided by NIED. We also used the unified earthquake catalogue and seismic intensity database published by JMA. Dr. Shin Koyama provided four sets of observation systems (SMAR-6A3P and LS-8800) for this survey. We would also like to show our gratitude to Dr. Hiroyuki Goto for showing us the modified S wave velocity model at station KMMH16 and Dr. Shigeki Senna for providing us a newly constructed 3D deep sedimentary structure model of Kumamoto plain, including Mashiki. We express our deepest gratitude to the local people in Mashiki for their understanding and cooperation with our observations.
The authors declare that they have no competing interests.
Availability of data and materials
Aftershock and microtremor data measured in this study (except site S4) are available to all interested researchers upon request, after the approval by all authors.
Ethics approval and consent to participate
This work was partly supported by a research grant from the Kinki Kensetsu Association in FY2016.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Fukumoto S, Yamanaka H, Midorikawa S, Irie K (2004) Array measurements of microtremors and seismic motions for evaluating long-period ground motions-estimation of S-wave velocity structure and characteristics of surface-wave propagation in the Keiyo coastal area of the Kanto plain, Japan. J Jpn Assoc Earthq Eng 4(4):87–106. https://doi.org/10.5610/jaee.4.4_87 (in Japanese with English abstract) Google Scholar
- Gardner GHF, Gardner LW, Gregory AR (1974) Formation velocity and density—the diagnostic basics for stratigraphic traps. Geophysics 39(6):770–780. https://doi.org/10.1190/1.1440465 View ArticleGoogle Scholar
- Goto H, Hata Y, Yoshimi M, Yoshida N (2016) Non-linear site response at KiK-net Mashiki site. In: Proceedings of the 36th JSCE earthquake engineering symposium, Kanazawa Theatre, Kanazawa, 17–18 October 2016 (in Japanese with English abstract) Google Scholar
- Hayashi K, Martin A, Hatayama K, Kobayashi T (2013) Estimating deep S-wave velocity structure in the Los Angeles Basin using a passive surface-wave method. Lead Edge 32(6):620–626. https://doi.org/10.1190/tle32060620.1 View ArticleGoogle Scholar
- Kagawa T, Yoshida S, Ueno H (2017) Strong motion characteristics in the vicinity of surface fault rupture at suburbs of Mashiki Town due to the 2016 Kumamoto earthquake. J Jpn Soc Civil Eng 73(4):I_841–I_846. https://doi.org/10.2208/jscejseee.73.i_841 (in Japanese with English abstract) Google Scholar
- Kawase H, Matsushima S, Nagashima F, Baoyintu Nakano K (2017) The cause of heavy damage concentration in downtown Mashiki inferred from observed data and field survey of the 2016 Kumamoto earthquake. Earth Planets Space 69:3. https://doi.org/10.1186/s40623-016-0591-1 View ArticleGoogle Scholar
- Kitagawa Y, Hiraishi H (2004) Overview of the 1995 Hyogo-ken Nanbu earthquake and proposals for earthquake mitigation measures. J Jpn Assoc Earthq Eng 4(3):1–29. https://doi.org/10.5610/jaee.4.3_1 Google Scholar
- Mori T, Matsushita K, Kawasaki A (2017) Buildings and residential lands damage survey in Mashiki-machi, Kumamoto Prefecture on the 2016 Kumamoto earthquake. Jpn Geotech J 12(4):439–455. https://doi.org/10.3208/jgs.12.439 (in Japanese with English abstract) View ArticleGoogle Scholar
- Morikawa H, Sawada S, Akamatsu J (2004) A method to estimate phase velocities of Rayleigh waves using microseisms simultaneously observed at two sites. Bull Seismol Soc Am 94(3):961–976. https://doi.org/10.1785/0120030020 View ArticleGoogle Scholar
- Nagao T, Lohani TN, Fukushima Y, Ito Y, Hokugo A, Oshige J (2017) A study on the correlation between ground vibration characteristic and damage level of structures at Mashiki Town by the 2016 Kumamoto earthquake. J Jpn Soc Civil Eng 73(4):I_294–I_309. https://doi.org/10.2208/jscejseee.73.i_294 (in Japanese with English abstract) Google Scholar
- Nakagawa H, Kashiwa H, Arai H (2017) Dynamic deformation properties and ground motion amplification characteristics in Mashiki Town. Paper presented at the 13th annual meeting of JAEE, the University of Tokyo, Tokyo, 13–14 November 2017 (in Japanese) Google Scholar
- National Institute for Land and Infrastructure Management and Building Research Institute (2016) Quick report of the field survey on the building damage by the 2016 Kumamoto earthquake. Technical note of NILIM 929/building research data 173, Tsukuba (in Japanese with English abstract) Google Scholar
- Ogawa N, Suzuki H, Senna S, Jin K, Wakai A, Matsuyama H, Sakurai K, Yatagai A, Komazawa M (2017) 3D subsurface structural model using microtremor array measurements for strong motion simulation around Kumamoto plain. In: Abstracts of the 136th SEGJ conference, Waseda University, Tokyo, 5–7 June 2017 (in Japanese) Google Scholar
- Okada S, Takai N (1999) Classifications of structural types and damage patterns of buildings for earthquake field investigation. J Struct Constr Eng Trans AIJ 64(524):65–72. https://doi.org/10.3130/aijs.64.65_5 (in Japanese with English abstract) View ArticleGoogle Scholar
- Saito M (1988) DISPER80: a subroutine package for the calculation of seismic normal mode solutions. In: Doorn-bos DJ (ed) Seismological Algorithms. Academic Press, Newyork, pp 293–319Google Scholar
- Sakai Y, Nojiri S, Kumamoto T, Tanaka Y (2008) Damage investigation of surroundings of the seismic stations in the 2007 Noto-hanto earthquake and correspondence of damage to buildings with strong ground motions. J Jpn Assoc Earthq Eng 8(3):79–106. https://doi.org/10.5610/jaee.8.3_79 (in Japanese with English abstract) Google Scholar
- Senna S, Wakai A, Jin K, Cho I, Matsuyama H, Yatagai A, Fujiwara H (2017a) Modeling of the subsurface structure from the seismic bedrock to the ground surface for a broadband strong motion evaluation in Kumamoto plain. In: Abstracts of the 136th SEGJ conference, Waseda University, Tokyo, 5–7 June 2017 (in Japanese) Google Scholar
- Senna S, Wakai A, Jin K, Matsuyama H, Fujiwara H (2017b) Modeling of the subsurface structure from the seismic bedrock to the ground surface for a broadband strong motion evaluation in Kumamoto plain. In: Abstracts of the 2017 AGU fall meeting, Ernest N. Morial Convention Center, New Orleans, 11–15 December 2017Google Scholar
- Shingaki Y, Yoshimi Y, Goto H, Kurita T, Sato K, Hosoya Y, Arai Y, Morita S (2017) Physical and dynamic properties of the volcanic ash soil in the heavily damaged site of the 2016 Kumamoto earthquake, Mashiki Town. J Jpn Soc Civil Eng 73(3):552–559. https://doi.org/10.2208/jscejseee.73.552 (in Japanese with English abstract) Google Scholar
- Shirahama Y, Yoshimi M, Awata Y, Maruyama T, Azuma T, Miyashita Y, Mori H, Imanishi K, Takeda N, Ochi T, Otsubo M, Asahina D, Miyakawa A (2016) Characteristics of the surface ruptures associated with the 2016 Kumamoto earthquake sequence, central Kyushu, Japan. Earth Planets Space 68:191. https://doi.org/10.1186/s40623-016-0559-1 View ArticleGoogle Scholar
- Sugino M, Ohmura S, Tokuoka S, Hayashi Y (2016a) Maximum response evaluation of traditional wooden houses based on microtremor measurements. J Struct Constr Eng Trans AIJ 81(729):1869–1879. https://doi.org/10.3130/aijs.81.1869 (in Japanese with English abstract) View ArticleGoogle Scholar
- Sugino M, Yamanuro R, Kobayashi S, Murase S, Ohmura S, Hayashi Y (2016b) Analyses of building damages in Mashiki Town in the 2016 Kumamoto earthquake. J Jpn Assoc Earthq Eng 16(10):10_69–10_83. https://doi.org/10.5610/jaee.16.10_69 (in Japanese with English abstract) Google Scholar
- Sugito N, Goto H, Kumahara Y, Tsutsumi H, Nakata T, Kagohara K, Matsuta N, Yoshida H (2016) Surface fault ruptures associated with the 14 April foreshock (Mj 6.5) of the 2016 Kumamoto earthquake sequence, southwest Japan. Earth Planets Space 68:170. https://doi.org/10.1186/s40623-016-0547-5 View ArticleGoogle Scholar
- Suzuki Y, Watanabe M, Nakata T (2016) Problems and recommendations for active fault and disaster mitigation learned from the 2016 Kumamoto earthquake. Kagaku 86(8):839–847 (in Japansese) Google Scholar
- Toda S, Kaneda H, Okada S, Ishimura D, Mildon ZK (2016) Slip-partitioned surface ruptures for the Mw 7.0 16 April 2016 Kumamoto, Japan, earthquake. Earth Planets Space 68:188. https://doi.org/10.1186/s40623-016-0560-8 View ArticleGoogle Scholar
- Tomozawa Y, Motoki K, Kato K, Hikita T, Ishiki K (2017) Investigation of tombstones fall-down rates and damages of wooden houses due to the 2016 Kumamoto earthquake—comparative examination of near-fault region and Mashiki-machi Miyazono. J Jpn Assoc Earthq Eng 17(4):4_62–4_80. https://doi.org/10.5610/jaee.17.4_62 (in Japanese with English abstract) Google Scholar
- Yamada M (2017) Damage islands in Mashiki Town from the 2016 Kumamoto earthquakes. J Jpn Assoc Earthq Eng 17(5):5_38–5_47. https://doi.org/10.5610/jaee.17.5_38 (in Japanese with English abstract) Google Scholar
- Yamada M, Mori J, Yamada M, Hada K, Fujino Y, Hayashida T (2017a) Unconventional damage patterns in Mashiki town during the 2016 Kumamoto earthquakes. In: Abstracts of the 2017 AGU fall meeting, Ernest N. Morial Convention Center, New Orleans, 11–15 December 2017Google Scholar
- Yamada M, Omura J, Goto H (2017b) Wooden building damage analysis in Mashiki Town for the 2016 Kumamoto earthquakes on April 14 and 16. Earthq Spectra 33(4):1555–1572. https://doi.org/10.1193/090816EQS144M View ArticleGoogle Scholar
- Yamada M, Yamada M, Mori J, Sakaue H, Hayashida T, Hada K, Fujino Y, Fukatsu S, Nishihara E, Ouchi T, Fujii A (2017c) Investigation of building damage in Mashiki-town for the 2016 Kumamoto earthquake—effect of local soil conditions. J Jpn Soc Civil Eng 73(4):I_216–I_224. https://doi.org/10.2208/jscejseee.73.i_216 (in Japanese with English abstract) Google Scholar
- Yamanaka H, Ishida H (1996) Application of genetic algorithms to an inversion of surface-wave dispersion data. Bull Seismol Soc Am 86(2):436–444Google Scholar
- Yamanaka H, Chimoto K, Miyake H, Tsuno S, Yamada N (2016) Observation of earthquake ground motion due to aftershocks of the 2016 Kumamoto earthquake in damaged areas. Earth Planets Space 68:197. https://doi.org/10.1186/s40623-016-0574-2 View ArticleGoogle Scholar
- Yoshimi M, Goto H, Haya Y, Yoshida N (2017) Nonlinear site response at the worst-hit area of the 2016 Kumamoto earthquakes in the Mashiki Town, Kumamoto, Japan. In: Abstracts of the 2017 disaster prevention research institute annual meeting, Kyoto University, Uji, 21–22 February 2017 (in Japanese with English abstract) Google Scholar