5:15 PM - 6:45 PM
[SSS07-P16] Evaluation of the structural characteristics around Wasabizawa geothermal area using S-wave splitting
Keywords:S-wave splitting, geothermal
We applied S-wave splitting technique to seismic waves passing through a geothermal reservoir to determine the temporal changes in the crack characteristics. S-wave splitting is the phenomenon in which the S-waves are separated into fast and slow directions when seismic S-waves pass through an anisotropic medium containing fractures and cracks. The operation of a geothermal power plant repeats the production and reinjection of hot water, and it could affect in the crack direction and closure in the geothermal reservoir. In previous studies, S-wave splitting had been used to confirm heterogeneity changes in both large and narrow region associated with large earthquakes. For example, Tadokoro and Ando(2002) observed the closure of cracks around the large earthquake source fault. There had not been many studies utilizing this technique for geothermal region. Therefore, we attempted to detect cracks around the geothermal reservoir and to monitor their dominant direction and density. These characteristics of the cracks were quantified by using Leading Shear-wave Polarization Direction(LSPD) and the difference in arrival time of two separated S-waves(δt ).
We selected the Wasabizawa geothermal power plant in Akita prefecture as a study area because it was the 4th largest geothermal production in Japan. We used a total of 32 Hi-net velocity waveforms provided by NIED, occurred from January 2006 to September 2022. Selected events met the following two requirements among JMA unified catalog. Firstly, the incident wave at the Ogachi station (N.OGCH) passed through the power plant area and the incident angle is less than 60 degrees. Secondly, the magnitude was distributed between 1 and 5, and the source depth was between 3 km and 30 km. The waveforms were recorded at the station with a sampling frequency of 100 Hz. Butterworth bandpass filter from 1 to 15 Hz was applied to the all waveforms. Then, we used the cross-correlation method (Shih and Meyer, 1990) to detect the S-wave splitting: using the NS and EW components, and obtaining the cross-correlation values(cc) of the two waveforms every 0.01 second at each rotation angle(φ ) while rotating the coordinate axis from 0 to 90 degrees in 5 degrees intervals. The time window was 0.50 seconds starting from 0.05 seconds before the S-wave arrival time. We displayed φ , δt , and cc as contour images. We recorded φ and δt at the highest cc for each seismic waveform. We also made rose diagrams to evaluate the crack direction by using LSPD obtained from φ and δt .
The direction of LSPD obtained from the rose diagram was generally considered to be a parallel to the long axis of crack and the direction of the maximum principal stress. δt was considered as an index of crack closure.
Before the power plant started the operation, LSPD was N70E-N110W, average δt was 0.08 seconds, and maximum δt was 0.17 seconds. After the power plant started the operation, LSPD was N85W-N95E, average δt was 0.046 seconds, and maximum δt was 0.09 seconds. Thus, the changes in maximum δt was 0.06 seconds. This changes in the δt can be interpreted as the closure of the cracks. Therefore, we considered that the overall amount of hydrothermal fluids in the subsurface had decreased during production and reinjection process through the operation of the geothermal power plant.
The largest change in LSPD was observed before and after the 2011 off the Pacific coasts of Tohoku Earthquake. Before the earthquake, LSPD was predominant in two directions, N55E-N125W and N30W-N150E, while it was predominant in N70W-N110W after the earthquake. Tadokoro et al. (1999) and Hiramatsu et al. (2010) also suggested that the changes in the dominant direction of cracks had been accompanied by large earthquakes. Large earthquakes probably cause a greater impact on S-wave splitting analysis than fluid and crack conditions in geothermal reservoir; thus, we need to carefully evaluate the heterogeneity change occurred in the geothermal reservoir.
We selected the Wasabizawa geothermal power plant in Akita prefecture as a study area because it was the 4th largest geothermal production in Japan. We used a total of 32 Hi-net velocity waveforms provided by NIED, occurred from January 2006 to September 2022. Selected events met the following two requirements among JMA unified catalog. Firstly, the incident wave at the Ogachi station (N.OGCH) passed through the power plant area and the incident angle is less than 60 degrees. Secondly, the magnitude was distributed between 1 and 5, and the source depth was between 3 km and 30 km. The waveforms were recorded at the station with a sampling frequency of 100 Hz. Butterworth bandpass filter from 1 to 15 Hz was applied to the all waveforms. Then, we used the cross-correlation method (Shih and Meyer, 1990) to detect the S-wave splitting: using the NS and EW components, and obtaining the cross-correlation values(cc) of the two waveforms every 0.01 second at each rotation angle(φ ) while rotating the coordinate axis from 0 to 90 degrees in 5 degrees intervals. The time window was 0.50 seconds starting from 0.05 seconds before the S-wave arrival time. We displayed φ , δt , and cc as contour images. We recorded φ and δt at the highest cc for each seismic waveform. We also made rose diagrams to evaluate the crack direction by using LSPD obtained from φ and δt .
The direction of LSPD obtained from the rose diagram was generally considered to be a parallel to the long axis of crack and the direction of the maximum principal stress. δt was considered as an index of crack closure.
Before the power plant started the operation, LSPD was N70E-N110W, average δt was 0.08 seconds, and maximum δt was 0.17 seconds. After the power plant started the operation, LSPD was N85W-N95E, average δt was 0.046 seconds, and maximum δt was 0.09 seconds. Thus, the changes in maximum δt was 0.06 seconds. This changes in the δt can be interpreted as the closure of the cracks. Therefore, we considered that the overall amount of hydrothermal fluids in the subsurface had decreased during production and reinjection process through the operation of the geothermal power plant.
The largest change in LSPD was observed before and after the 2011 off the Pacific coasts of Tohoku Earthquake. Before the earthquake, LSPD was predominant in two directions, N55E-N125W and N30W-N150E, while it was predominant in N70W-N110W after the earthquake. Tadokoro et al. (1999) and Hiramatsu et al. (2010) also suggested that the changes in the dominant direction of cracks had been accompanied by large earthquakes. Large earthquakes probably cause a greater impact on S-wave splitting analysis than fluid and crack conditions in geothermal reservoir; thus, we need to carefully evaluate the heterogeneity change occurred in the geothermal reservoir.