17:15 〜 18:45
[SSS10-P19] A technique for determining phase velocities in the high frequency range with large arrays
キーワード：微動探査、レイリー波、空間自己相関法、アレイ観測
Array observation of microtremors aiming to obtain the dispersion curve of the surface wave is one of the most effective ways to estimate the velocity structure of the ground. The spatial autocorrelation (SPAC), one of the most popular methods, estimates the phase velocity by calculating the spatial autocorrelation (SPAC) coefficients from the observation records and utilising the fact that they follow a zero-order Bessel function of the first kind (hereafter referred to as J0 function) whose argument consists of the phase velocity, the distance between observation points and frequency. However, the phase velocity cannot be uniquely determined from the J0 function because the phase velocity depends on the frequency. Therefore, it is common practice to estimate the phase velocity only in the interval where the frequency and phase velocity correspond one-to-one in the SPAC coefficients obtained from observations, i.e., the frequency range between the first peak and the first trough of the J0 function. Consequently, the range of frequencies of phase velocity that is obtainable from array observations with a particular array size is limited. It is required to arrange multiple arrays with different distances between observation points to overcome this limitation.
The direct-fitting method (DFM) by Asten et al. (2004) is a method that uses the acquired SPAC coefficients directly to estimate the velocity structure of the ground without explicitly calculating the phase velocity. DFM was developed to avoid the non-linear and multi-variable problem of obtaining the phase velocity from the SPAC coefficients. However, it is not so difficult to solve such a problem owing to recent improvements in computer performance. In addition, visualizing the shape of the dispersion curve obtained from the observations is useful to create an initial model for inversely estimating the velocity structure of the ground. Therefore, this study proposes a method for estimating phase velocities up to the high-frequency range, even when large array sizes are employed for observations.
We applied the proposed method to synthetic data for verification. In generating the synthetic data, we modelled the Rayleigh wavefield as an ergodic stationary stochastic process. The synthetic data were generated as time series and analysed by treating them as observed waveforms so that the influence of the finiteness of the obtainable time series is taken into account. We also investigated the robustness of the method by adding noise to the generated time series and analysing them.
The proposed method was applied to actual field data to validate the effectiveness of the method. Observations were carried out in the vicinity of the Kashima industrial complex in Kamisu City, Ibaraki Prefecture, Japan. We deployed arrays with a maximum inter-station distance of about 450 m and a minimum of about 30 m. Observations were made using dynamic coil velocimeters with a sampling rate of 100 sps and a data depth of 24 bits. As a result, phase velocities up to 3 Hz were obtained from the 450 m array using our proposed method, whereas 0.7 Hz was the upper limit frequency of the conventional method.
Acknowledgments
We are grateful to The High Pressure Gas Safety Institute of Japan, Ibaraki Prefectural Government, Government of Kamisu City, Ohsaki Research Institute, Inc., Dr Atsushi Nozu of the Port and Airport Research Institute (PARI) and Dr Masayuki Yoshimi of the National Institute of Advanced Industrial Science and Technology (AIST) for their supports in the microtremor observations in the Kashima Complex area.
The direct-fitting method (DFM) by Asten et al. (2004) is a method that uses the acquired SPAC coefficients directly to estimate the velocity structure of the ground without explicitly calculating the phase velocity. DFM was developed to avoid the non-linear and multi-variable problem of obtaining the phase velocity from the SPAC coefficients. However, it is not so difficult to solve such a problem owing to recent improvements in computer performance. In addition, visualizing the shape of the dispersion curve obtained from the observations is useful to create an initial model for inversely estimating the velocity structure of the ground. Therefore, this study proposes a method for estimating phase velocities up to the high-frequency range, even when large array sizes are employed for observations.
We applied the proposed method to synthetic data for verification. In generating the synthetic data, we modelled the Rayleigh wavefield as an ergodic stationary stochastic process. The synthetic data were generated as time series and analysed by treating them as observed waveforms so that the influence of the finiteness of the obtainable time series is taken into account. We also investigated the robustness of the method by adding noise to the generated time series and analysing them.
The proposed method was applied to actual field data to validate the effectiveness of the method. Observations were carried out in the vicinity of the Kashima industrial complex in Kamisu City, Ibaraki Prefecture, Japan. We deployed arrays with a maximum inter-station distance of about 450 m and a minimum of about 30 m. Observations were made using dynamic coil velocimeters with a sampling rate of 100 sps and a data depth of 24 bits. As a result, phase velocities up to 3 Hz were obtained from the 450 m array using our proposed method, whereas 0.7 Hz was the upper limit frequency of the conventional method.
Acknowledgments
We are grateful to The High Pressure Gas Safety Institute of Japan, Ibaraki Prefectural Government, Government of Kamisu City, Ohsaki Research Institute, Inc., Dr Atsushi Nozu of the Port and Airport Research Institute (PARI) and Dr Masayuki Yoshimi of the National Institute of Advanced Industrial Science and Technology (AIST) for their supports in the microtremor observations in the Kashima Complex area.