In order to utilize not only the Rayleigh wave but also the Love wave for microtremor array survey, we verify the effectiveness of frequency wavenumber analysis (FK) method extended to horizontal components of microtremors by Saito (2007). An FK analysis seems to be practically effective because the arrangement of observation positions can be an arbitrary shape. In addition, this FK analysis has an advantage that it is not necessary to rotate radial / transverse coordinate system from EW / NS coordinate system. In this study, we used the FK analysis by Saito (2007) and a spatial autocorrelation (SPAC) analysis for microtremor records observed with a triplet triangle array consisting of 10 sensors whose radius is changed every 3 times in order to try to estimate phase velocities of Love wave. We could obtain phase velocities of transverse and radial components by the two methods. The phase velocities obtained from the FK method were approximately equal to the phase velocities obtained from the SPAC method. In the case of the radial FK analysis, phase velocities of higher mode Rayleigh wave which were faster than ones of the fundamental mode one were detected in the high frequency band. The Love wave power fractions calculated by SPAC analysis showed to be 40 % - 60 % in the ground of Faculty of Science and Engineering, Iwate University. Since Love wave and Rayleigh wave were present at almost the same ratio at frequencies of 3 Hz to 13 Hz, it was thought that an estimation of both Love wave and Rayleigh wave succeed.
The monitoring of the groundwater is basically carried out based on water pressure data and various other information observed in boreholes. However, it is difficult to evaluate the spatial distribution of the ground water by obtained water pressure data. A long-term groundwater decrease exceeding 80 m has been observed in the Tono region for over ten years, while we have been observing absolute gravity data in three stations. We first examined the correction of the gravity data, which was affected by some artificial phenomena around the gravity stations. We then simulated the temporal gravity change by assuming groundwater flow in the known aquifer with referring water pressure data in the research area. The agreement between the simulated and the observed gravity values was within about 3 micro-Gals. We conclude that the long-term groundwater level decrease in the Tono region was well explained with gravity data, and the continuous gravity observation is effective for the detection of the subsurface water movement.
In this study, in order to examine the applicability of phase velocity estimation method using Green's functions with a common virtual source by seismic interferometry, we conducted continuous seismic observation at 16 points in Wakasa Bay region, and estimated the phase velocity by using both the positive side of the cross correlation function having large amplitude and the negative side of the cross correlation function having small amplitude, and compared the both results. In a case where the signal of the cross-correlation function on the side corresponding to the propagation direction from the common virtual source to the observation points is clear, it is possible to estimate the phase velocity stably from the Green's functions with a common virtual source, and the result is almost same as the phase velocity obtained by the natural earthquake records. On the other hand, in a case where the signal of the cross-correlation function on the side corresponding to the propagation direction from the common virtual source to the observation points is unclear, the phase velocity can not be estimated. However, it was found that the phase velocity can be estimated satisfactorily by using the cross correlation function whose sign of delay time is opposite as the Green’s functions with the common virtual source. Furthermore, when the Green’s functions become asymmetric shape as in this study, it was found that we could estimate the phase velocity by placing a seismic observation point in the direction where the signal becomes strong.
The experts perform interpretation of GPR (ground penetrating radar) data by extracting characteristic reflection patterns from GPR time sections due to the contrast of the relative permittivity of the underground structure. An amount of acquired GPR data has been massive and keeps increasing due to an improvement of the data acquisition system and aging of the infrastructure to be maintained, therefore, automation and labor saving of interpretation works are demanded. Object recognition ability by deep learning which is one of machine learning methods has greatly improved, and many architectures of deep learning model have been advocated and studied. In particular, AlexNet is a sort of fundamental deep learning model which opened the leap-improving accuracy by using the method such as CNN (Convolution Neural Network) imitating the real visual cortex and it has been applied to a variety of object recognition applications. Today machine learning for interpretation of GPR data is often applied to grayscale images. However, like experts using GPR color images in visual inspection, it is considered that the information is less lost and more useful interpretation result is obtained in color images than gray images. We did the experiments of AlexNet for GPR data interpretation to design a learning model suitable for GPR data interpretation. The experiment result shows the colorized images makes a better performance, 0.9819 in the F value and 0.9875 in the accuracy than gray-scale images input. We also analyzed the internal output of the several steps in the AlexNet to discuss the appropriate learning model design for GPR data.
P-wave velocity (Vp) is an important parameter to construct a seismic model of the subsurface by using microtremors and earthquake ground motions, though Vp is not so strongly constrained as S-wave velocity (Vs). In order to reflect the ground survey information in Japan to the Vp structure, we investigated the relationship between Vs, Vp, and depth by using PS-logging data at the all sites of K-NET and KiK-net. Vp values are concentrated at around 500 m/s and 1500 m/s when Vs is lower than 1000 m/s, where these concentrated areas show the characteristics of the unsaturated soil and the saturated soil, respectively. The most of the Vp values in the layer shallower than 4 m are about 500 m/s, which suggests the dominance of the unsaturated soil, while the most of Vp values in the layer deeper than 4 m are larger than about 1500 m/s, which suggests the dominance of the saturated soil there. We also investigated those relationships for different types of the soil at K-NET sites. Although each soil type has their own depth range, the all soil types show similar relationships between Vs, Vp, and depth. Then considering the depth profile of Vp, we divided the dataset into two by the depth, which is shallower or deeper than 4 m, and calculated the geometric mean of Vp and the geometric standard deviation in every Vs bins of 200 m/s. Finally we obtained the regression curves for the average and standard deviation of Vp estimated from Vs to get the Vp conversion functions from Vs, which can be applied to the wide Vs range. We also obtained the regression curves for two datasets with Vp lower and higher than 1200 m/s. These regression curves can be applied in case that the groundwater level is known.
In this paper, I suggest an inversion technique to estimate the source structure of a vertical deformation without assuming the model shape. This method estimates a source structure that explains the vertical deformations observed at the surface by minimizing the area having an arbitrary target value established; it is capable of clearly showing the source structure. I applied this technique to the vertical deformation observed in the Jigokudani valley (Midagahara volcano, central Japan), and found that the estimated source depth was compatible with the upper depth of the cap rock structure obtained by the AMT(Audio-frequency Magneto Telluric) survey and with the source depth estimated from past crustal movements.
We carried out microtremor observations with a miniature array in the coastal area of Kuji city, Iwate prefecture and we tried to estimate shallow S wave velocity structures using microtremor H/V and phase velocity dispersion curves. The site amplification factors from the engineering bedrock based on an average S-wave velocity were evaluated. Then, the SH wave amplification ratios as a function of frequency were also calculated by using the S-wave velocity structure model. The results are shown as follows: Shallow S-wave velocities in the coastal area of Kuji City were very slow and were about 100m/s. The estimated structure in this area indicated the bowl-shape. The amplification factors obtained from microtremor observations were larger than the site amplification factor of J-SHIS. The amplification characteristics of the SH wave showed that the maximum of the amplification factor was six times at a frequency of 1 Hz.
We applied seismic interferometry analysis to linear array records of microtremors observed with eleven three-component seismometers in order to aim to develop a new method for exploring shallow two dimensional S wave velocity structures using short period microtremors. As a result, in the stacked cross correlation analysis, it was possible to confirm characteristic phases propagating in all directions of NS, EW and UD as a result of arranging seismometers at equal intervals in the linear array. In estimating the group velocities, the signal to noise ratios (SNRs) were particularly low in the NS direction, and it was difficult to estimate correct group velocities. On the other hand, in the EW and UD directions, it was shown that the frequency range where the SNR is higher than 10 is wide. Moreover, since it was close to the value calculated from the existing model, it is considered that a reasonable group velocity could be estimated. As a result of displaying the group velocity obtained in all combinations in two dimensions, it became clear that the group velocity slightly changed in the horizontal direction at high frequencies.
Microtremor observations were carried out for the purpose of exploring the S-wave velocity structures in order to clarify seismic risks in Rikuzen-Takata City, Iwate Prefecture, where severe damage occurred in the 2011 off the Pacific coast of Tohoku Earthquake and the great tsunami. Microtremor observations using a single seismometer were carried out at 78 points to examine the depth distribution of the bedrock was examined from the peak period of H/V spectral ratios. And, microtremor observations using a miniature array were also carried out at 14 sites to estimate the shallow S-wave velocity structures. As a result, the H/V peak period increases from the north area to the south area in the coastal plain, the H/V peak period increases from the east area to the west area, and it abruptly became shorter across the Kesen River. Therefore, the depth of bedrock gradually becomes deeper from the north area to the south area, from the west area to the east area, it is thought that the bedrock structure indicates a bowl-shaped one with a deep base near the center and with a shallow end at both ends. This result was similar to the geological cross section of Chida et al. (1984). In the place showing the basement depth of about 40 m, our S-wave velocity structure estimated using large size arrays and a miniature array were consistent with the geological cross section of Chida et al. (1984). The two dimensional shallow S-wave velocity structure obtained from the microtremor exploration using a single point and a miniature array was also in good agreement with the geological cross section by Chida et al. (1984).