3:30 PM - 3:45 PM
[SSS09-11] Microtremor Array Survey for Estimating Shallow S-wave Velocity Structure in the Kathmandu Valley
Keywords:Kathmandu Valley, Nepal, shallow S-wave velocity structure, microtremor array observation, microtremor H/V spectral ratio
In the Kathmandu Valley, twelve strong-motion stations are currently in operation. We had estimated the S-wave velocity structure for the depth equivalent to the seismic basement based on the strong-motion records [Bijukchhen et al., 2017] and for the surface layer with surface wave surveys [Shigefuji et al., 2020]. This study deals with microtremor array observations to clarify the shallow velocity structure in more detail.
Microtremor array observations were conducted at 18 points in and around the Kathmandu Valley for six days from September 16 to 21, 2022. The observation points were in the vicinity of the strong-motion stations and several points in Bhaktapur city, where the damage was concentrated during the 2015 Gorkha Nepal earthquake (Mw 7.8, Depth 8.2 km). A total of four seismometers were arranged in each vertex of the equilateral triangle and the center of it, and small to medium-sized arrays were carried out for each observation point. McSEIS-AT 3ch sensors and data loggers manufactured by OYO Corporation were employed with a sampling frequency of 250 Hz and a gain of 16 dB.
The H/V microtremor spectral ratio was obtained by dividing the vector sum of the two horizontal components’ spectra by the vertical component’s spectrum. The peak frequencies of the H/V spectral ratios were distinctly different depending on the surface geological classification, in a frequency range of 2.5 to 4.0 Hz for the bedrock site, 0.3 to 0.8 Hz for the lake sedimentary site, and 0.3 to 1.0 Hz for the fluvial sedimentary site.
The Rayleigh wave dispersion curves were obtained using the SPAC method [Okada, 2003]. Phase velocities at the high frequency range are from 100 to about 300 m/s at the bedrock site, 100 to 250 m/s at the fluvial sedimentary site, and 100 to 200 m/s at the lake sedimentary site. The phase velocity of Changu Narayan (CGN) site is low despite being on a higher elevation and classified as bedrock. This CGN site is on the landslide topography, and the surface layer is soft with semi-solid clay, sand, and gravel on the surface [Inagaki, 2015].
The S-wave velocity structure was estimated by inversion of Rayleigh wave phase velocities using a genetic algorithm (GA) [Yamanaka and Ishida, 1996]. The S-wave velocities and layer thickness were obtained using the S-wave velocity structure estimated by strong ground motions and surface wave surveys as a reference. About other sites, these parameters were decided through trial and error and by satisfying the shape of the phase velocity dispersion curve. The final solution is selected as the lowest misfit value, defined as the mean square of the difference between the observed and theoretical phase velocities.
The estimated S-wave velocity structure corresponds to each surface geologic formation. The average S-wave velocities to a depth of 30 m (AVS30) were calculated and compared with elevation and soil classification. A positive correlation between elevation and AVS30 was confirmed. The AVS30s at bedrock sites are high, with around 350 to 400 m/s. At the fluvial sedimentary sites, AVS30s are around 200 to 350 m/s. And the AVS30s at the lake sedimentary sites are low, with around 100 to 250 m/s. The AVS30 at CGN was about 200 m/s, despite being on a hill-top.
The accuracy of the shallow ground structure model of the Kathmandu Basin will be improved by increasing the number of observation points.
Microtremor array observations were conducted at 18 points in and around the Kathmandu Valley for six days from September 16 to 21, 2022. The observation points were in the vicinity of the strong-motion stations and several points in Bhaktapur city, where the damage was concentrated during the 2015 Gorkha Nepal earthquake (Mw 7.8, Depth 8.2 km). A total of four seismometers were arranged in each vertex of the equilateral triangle and the center of it, and small to medium-sized arrays were carried out for each observation point. McSEIS-AT 3ch sensors and data loggers manufactured by OYO Corporation were employed with a sampling frequency of 250 Hz and a gain of 16 dB.
The H/V microtremor spectral ratio was obtained by dividing the vector sum of the two horizontal components’ spectra by the vertical component’s spectrum. The peak frequencies of the H/V spectral ratios were distinctly different depending on the surface geological classification, in a frequency range of 2.5 to 4.0 Hz for the bedrock site, 0.3 to 0.8 Hz for the lake sedimentary site, and 0.3 to 1.0 Hz for the fluvial sedimentary site.
The Rayleigh wave dispersion curves were obtained using the SPAC method [Okada, 2003]. Phase velocities at the high frequency range are from 100 to about 300 m/s at the bedrock site, 100 to 250 m/s at the fluvial sedimentary site, and 100 to 200 m/s at the lake sedimentary site. The phase velocity of Changu Narayan (CGN) site is low despite being on a higher elevation and classified as bedrock. This CGN site is on the landslide topography, and the surface layer is soft with semi-solid clay, sand, and gravel on the surface [Inagaki, 2015].
The S-wave velocity structure was estimated by inversion of Rayleigh wave phase velocities using a genetic algorithm (GA) [Yamanaka and Ishida, 1996]. The S-wave velocities and layer thickness were obtained using the S-wave velocity structure estimated by strong ground motions and surface wave surveys as a reference. About other sites, these parameters were decided through trial and error and by satisfying the shape of the phase velocity dispersion curve. The final solution is selected as the lowest misfit value, defined as the mean square of the difference between the observed and theoretical phase velocities.
The estimated S-wave velocity structure corresponds to each surface geologic formation. The average S-wave velocities to a depth of 30 m (AVS30) were calculated and compared with elevation and soil classification. A positive correlation between elevation and AVS30 was confirmed. The AVS30s at bedrock sites are high, with around 350 to 400 m/s. At the fluvial sedimentary sites, AVS30s are around 200 to 350 m/s. And the AVS30s at the lake sedimentary sites are low, with around 100 to 250 m/s. The AVS30 at CGN was about 200 m/s, despite being on a hill-top.
The accuracy of the shallow ground structure model of the Kathmandu Basin will be improved by increasing the number of observation points.