- Email: [email protected]
Time evolution of hydraulic and electrokinetic parameters around the Nojima fault, Japan, estimated by an electrokinetic method
Tectonophysics 443 (2007) 200 – 208 www.elsevier.com/locate/tecto
Time evolution of hydraulic and electrokinetic parameters around the Nojima fault, Japan, estimated by an electrokinetic method Hideki Murakami a,⁎, Naoto Oshiman b , Satoru Yamaguchi c , Takeshi Hashimoto d , Ryokei Yoshimura b a
Department of Natural Environmental Science, Faculty of Science, National University Corporation Kochi University, Kochi 780-8520, Japan b Disaster Prevention Research Institute, National University Corporation Kyoto University, Uji 616-0011, Japan c Department of Earth and Planetary Sciences, Faculty of Science, National University Corporation Kobe University, 657-8501, Japan d Institute of Seismology and Volcanology, National University Corporation Hokkaido University, Sapporo 060-0810, Japan Received 14 May 2004; accepted 29 January 2007 Available online 27 March 2007
Abstract The Nojima Fault Zone Probe was designed to study the properties and healing processes of the Nojima fault, which is the surface fault rupture of the Hyogo-ken Nanbu earthquake (M7.2) of 1995 (1995 Kobe earthquake). In this project, water injection experiments were conducted in a borehole of 1800 m depth at the Nojima fault. We set up electrodes around the borehole and observed self-potential variations to investigate the magnitude of electrokinetic and hydraulic parameters around the Nojima fault zone. In the 1997 experiment, self-potential variations were in the range of a few to about 20 mV across 320–450 m electrode dipoles with hydraulic pressure variations from 3.5 to 4 MPa. In the 2000 experiment, self-potential variations were in the range of a few to about 85 mV across 160–260 m electrode dipoles with the hydraulic pressure variations from 3 to 4.5 MPa. In the 2003 experiment, self-potential variations were in the range of a few to about 30 mV across 20–80 m electrode dipoles with hydraulic pressure of 4 MPa. These observed self-potential variations were explained well with an electrokinetic effect due to the underground flow of the injected water. From the observed results, we estimated that the ratio of hydraulic parameters (permeability, porosity, and tortuosity) to electrokinetic parameters (zeta potential and dielectric constant) decreased approximately 40% during eight years after the earthquake. This result suggests that the healing process around the fault zone progress. © 2007 Elsevier B.V. All rights reserved. Keywords: Nojima Fault Zone Probe; Water injection experiment; Self-potential; Electrokinetic phenomena; Healing process; Active fault
1. Introduction It is generally accepted that crustal earthquakes are caused by sudden displacement on faults and large earthquakes repeatedly occur along a large-scale fault. However, fault healing and re-strengthening, which occurred just after the previous earthquake, are not clear. ⁎ Corresponding author. Fax: +81 88 844 8356. E-mail address: [email protected] (H. Murakami). 0040-1951/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2007.01.019
The Nojima Fault Zone Probe was designed to study the properties and healing processes of the Nojima fault, which moved during the 1995 Hyogo-ken Nanbu earthquake (Kobe earthquake;MJMA7.2) (Ando, 2001). Shimazaki et al. (1998) have shown a concept of the repeated water injection experiments that the healing process of a fault zone can be detected by monitoring its permeability derived from water injections into the fault zone. In order to carry out the water injection experiments, three boreholes with depths of 500 m,
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
201
800 m, and 1800 m were drilled into or near the fault zone by the Disaster Prevention Research Institute (DPRI), Kyoto University. Water injection experiments using the DPRI 1800 m borehole were repeated three times at 1997, 2000, and 2003 to investigate the properties and healing processes of the fault zone. Many kinds of geological and geophysical observations were done (e.g. Fujimori et al., 2001; Kitagawa et al., 2002; Mukai and Fujimori, 2003; papers in this issue). We have carried out monitoring of surface electric potential variations around the DPRI 1800 m borehole during water injection experiments to investigate the electrokinetic phenomena associated with the ground water flow from the borehole to the fault zone. It has been known that fluid flow through a porous medium generates an electric potential variation known as a streaming potential, which is one of the electrokinetic phenomena (Ishido and Mizutani, 1981). The selfpotential method has been applied to study geophysical phenomena related to fluid flow (e.g. Ishido et al., 1983; Marquis et al., 2002). We already reported self-potential variations associated with the 1997 water injection experiment in a previous paper (Murakami et al., 2001). In this paper we report the results of 2000 and 2003 experiments and show the evolution of hydraulic and electrokinetic parameters around the Nojima fault. 2. Field observation method Fig. 1 shows locations of the DPRI 1800 m borehole, the Nojima earthquake fault, and the Nojima branch
Fig. 2. Location of the DPRI 1800 m borehole and Cu–CuSO4 electrodes used for measuring self-potentials. (a) (■) Cu–CuSO4 electrodes used during the 1997 water injection experiment and (○) DPRI 1800 m borehole. (b) (■) Cu–CuSO4 electrodes used during the 2000 water injection experiment and (○) DPRI 1800 m borehole. (c) (■) Cu–CuSO4 electrodes used during the 2003 water injection experiment and (○) DPRI 1800 m borehole.
Fig. 1. Location of the Nojima fault, its branch fault, and (○) Disaster Prevention Research Institute (DPRI) 1800 m borehole (water injection well) in Awaji Island, Japan.
fault. The water injection borehole locates near the southwestern end of the Nojima fault. This borehole is drilled vertically for the first 1200 m and then bent
202
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Table 1 Summary of the water injection experiments of the Nojima Fault Zone Probe Year
Period
Pressure (MPa)
Injection water flow rate (L/min)
1997
Feb. 9–11 Feb. 12–13 Mar.16–25 Jan. 22–26 Jan. 31–Feb. 5 Mar. 3–11 Mar. 13–23 May 8–13
2.8–3.7 3.4–4.2 4.4 3.0 4.0 4.6 3.0–4.5 4.0
8–10 12–20 18 10–12 16–18 20–26 15–23 15–25
2000
2003
gradually with maximum deviation of 20° from the vertical and reaches the fault zone at 1800 m depth (Ando, 2001). The Nojima fault is a surface rupture of the 1995 Hyogo-ken Nanbu earthquake (Kobe earthquake) and is activated as a right-lateral strike–slip fault with a reverse sense (Murata et al., 2001). Along the Nojima branch fault only small surface ruptures occur during the earthquake (Awata et al., 1995).
Fig. 2(a), (b), and (c) show the location of electrodes of 1997, 2000, and 2003 experiments. We used Cu–CuSO4 electrodes to measure the self-potential and berried electrodes 20 cm depth into the ground. As shown in Fig. 2, sixteen electrodes were distributed around the 1800 m borehole in the 1997 experiment, twenty-two electrodes in the 2000 experiment, and twenty-two electrodes in the 2003 experiment. Electrical potential difference between each observation point and a common point (indicated by ‘common ground’ in Fig. 2) was monitored on digital recorders with a 10 s sampling period. The water injection conditions are summarized in Table 1. The 1997 water injection experiment was made over three periods: Feb. 9–11, Feb. 12–13, and Mar. 16–25, 1997. And, we observed the self-potential variations from Feb. 9 through Feb. 13, 1997. The 2000 water injection experiment was made over three periods: Jan. 22–26, Jan. 31–Feb. 5, and Mar. 3–7, 2000. And, we analyzed the data observed during three periods. The 2003 experiment was made over two periods: Mar. 13–23 and May 8–13, 2003. And, we analyzed only the data
Fig. 3. Noise reduction data of self-potentials to the electrode located at the site D2 in Fig. 2(a) during the water injection experiment of Feb. 9–13, 1997. Pressure and flow rate of the injection water are shown in the bottom.
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
observed during Mar. 13–23 in this paper because used electrodes were unstable during the second period. 3. Observation results Fig. 3 shows filtered self-potential variations, flow rate of injection, and hydraulic pressure at the wellhead of the 1800 m borehole during the 1997 water injection experiment. Self-potentials with respect to the electrode located at the site D2 in Fig. 2(a) are shown here in order to compare the previous result (Murakami et al., 2001). In order to reduce large artificial noise and diurnal variations, we applied independent component analysis (ICA) that is a statistical and computational technique for finding underlying factors or components from multivariate (multidimensional) statistical data. This method is similar to the principal component method, but the sources in ICA are assumed both nongaussian and statistically independent. Recently, ICA has been applied to geophysical problems (e.g. Filipe et al., 2002; Ciaramella et al., 2004). We analyzed the observed selfpotential data using ICALAB toolboxes for MATLAB (Cichocki and Amari, 2002). The obtained results were
203
consistent with the previous results (Murakami et al., 2001). The self-potential variations of ∼ 20 mV in response to water injection operations were extracted and its magnitude decreased with increasing distance from the injection well. Fig. 4 shows a typical example of filtered selfpotential variations during the 2000 water injection experiment. This figure shows self-potential variations during Mar. 3–7, 2000. Self-potentials with respect to the common ground electrode located 256 m southwest of the 1800 m borehole are shown. The self-potential variations of ∼ 100 mV were observed and were synchronized with turning on and off the water injection. However, the amplitudes of self-potential variations during the experiment were not constant when the pressure and fluid flow rate were approximately constant. These variations may be reflected in the complex flow of subsurface ground water. Fig. 5 shows filtered self-potential variations during Mar. 13–23, 2003. Self-potentials at each electrode with respect to the electrode located at the site B7 in Fig. 2(c) are shown here because the common electrode was unstable. The data at sites C1–D4 were also omitted
Fig. 4. Noise reduction data of self-potentials to the common ground electrode in Fig. 2(b) during the water injection experiment of Mar. 3–11, 2000. Pressure and flow rate of the injection water are shown in the bottom.
204
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Fig. 5. Noise reduction data of self-potentials to the electrode located at the site B7 in Fig. 2(c) during the water injection experiment of Mar. 13–23, 2003. Pressure and flow rate of the injection water are shown in the bottom.
because many electrodes were unstable. The selfpotential variations of ∼ 30 mV synchronized with the water injection operations were observed. As described in the previous paper (Murakami et al., 2001), the following three characters of self-potential variations were observed in common with the three water injection experiments: 1) self-potential variations appeared to correspond to the operation of water injections, 2) these variations were observed at all the observation sites around the water injection borehole, 3) the negative voltage appeared around the water injection borehole. 4. Discussion and conclusion As already described in the previous paper (Murakami et al., 2001), the characters of self-potential changes during the water injection experiments suggest the possibility that all of the observed self-potential changes represent an electrokinetic effect due to the underground flow of the injected water. Thus we consider the model based on electrokinetic effects shown in Fig. 6. Ishido and Mizutani (1981)
have described the phenomenological equations of electrokinetic effect in rock–water systems are: i ¼ /t 2 rjV þ /t 2 ðef=lÞjP
ð1Þ
j ¼ /t 2 ðef=lÞjV ðk=lÞjP
ð2Þ
in which i is the drag electric current density (A/m 2 ), j is the fluid flow flux (m/s); V is the electric potential (V), P is the pore fluid pressure (N/m); ϕ, t, and k are the porosity, tortuosity, and permeability (m 2 ) of the aquifer, respectively; ε and μ are the dielectric constant (F/m) and viscosity (Ns/m 2 ) of water, respectively; ζ is the zeta potential (V); σ = σf + m − 1 σs (σf, pore fluid electric conductivity; σs, surface electric conductivity; and m, hydraulic radius). In the case of neglecting the quantity ϕt − 2 ε 2 ζ 2 / σμk (Murakami et al., 2001), the relation between the total conductive electric current Itotal (A) and the amount of injection water Jtotal (m3 /s) is given by the following equation: Itotal ¼ idrag d2krH ¼ ½ef=ðk=/t 2 ÞJtotal
ð3Þ
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
205
Fig. 6. Line source model for self-potential variations. Electric field can be produced by the subsurface fluid flow through electrokinetic coupling. The change in voltage in the aquifer is conducted to the whole part of the injection well through the highly conductive iron casing pipe. In this model, we may estimate the total conductive current from the semilogarithmic plot of distance vs. self-potential changes.
in which H is the thickness of the aquifer and r is the pore radius. In the above model the injection well will be electrically negative if the zeta potential in aquifer is negative. In the water injection experiments, negative voltage appeared on the ground surface around the water injection borehole because the change in voltage in the aquifer was conducted to the whole part of the well through the iron borehole pipe that is not insulated electrically. In the experiments the iron borehole pipe acted as a line source of electric current. In this model the potential difference between two points on the surface is expressed by the following equation: ! dz0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi b2 þ z20 0 0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! Itotal b L þ L2 þ a2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi log d ¼ a L þ L2 þ b 2 2krc L
Itotal DV ¼ 2krc L
Z
L
dz0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi a2 þ z20
Z
L
ð4Þ
in which ΔV is the potential difference between two points on the surface, Itotal is the total conductive current flowing out from the borehole, L is the length of line current source
(iron borehole pipe), σc is the averaged electrical conductivity of the ground, and a is the distance of each electrode from the borehole, b is the distance of the reference electrode from the borehole. In the case of a, b ≪L, we can use the following approximate expression (Matsushima et al., 1997): DV ¼ ðItotal =2krc LÞlogðb=aÞ ¼ ðItotal =2krc LÞðlogb logaÞ;
½a; b≪L:
ð5Þ
In this study we estimated the total conductive electric current and the standard error of it using Eq. (5) and the following parameters: L = 1800 m, σc = 0.00125 S/m (Makino and Oshiman, 1998), and b (= 104 m for the 1977 experiment, 256 m for the 2000 experiment, and 74 m for the 2003 experiment). Fig. 7 shows the magnitude of self-potential change as a function of the distance from the borehole. Solid symbols indicate the observed self-potential changes and solid lines are linear regression lines determined using the method of least squares. The value of the total conductive current derived from the slope of the regression line and standard error are shown in Fig. 7.
206
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Fig. 7. Semilogarithmic plot of distance vs. self-potential changes. Closed symbols are observed values of self-potential change and solid lines are linear regression lines. (a) The 1997 water injection experiment. (b-1) and (b-2) the 2000 water injection experiment. (c) the 2003 water injection experiment.
The total conductive currents of − 0.18 to − 0.31 A during the 1997 experiment from Fig. 5(a), the total conductive currents of − 0.20 to − 0.61 A during the 2000 experiment from Fig. 5(b-1) and (b-2), and the
total conductive current of − 0.41 A during the 2003 experiment were estimated. Fig. 8 shows the evolution of the ratio of the amount of injected water to the total conductive electric current.
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
The absolute value of the ratio of Jtotal/Itotal decreases with time. The absolute value of the ratio in the 2000 experiment decreases by nearly 80% of that in the 1997 experiment. The absolute value of the ratio in the 2003 experiments is equal to or 10% smaller than that of the 2000 experiment. This suggests the possibility that the hydraulic parameter k/(ϕt− 2) decreases with time if electric parameters (ε and ζ) are not changed. The same result was also obtained from the analysis of permeability using the water level changes (Kitagawa et al., 2002, 2007-this issue) and using strain changes (Mukai and Fujimori, 2003) in the DPRI 800 m borehole, which is located about 40 m south from the DPRI 1800 m borehole. These results suggest that the hydraulic parameters at about 550 m depth in the hanging side of the Nojima fault decrease with time. Yamano and Goto (2005) has estimated from the temperature changes in the 1800 m borehole that the injected water leaked out of the borehole at about 550 m depth. We measured the chemical and physical elements of the injected water because the zeta potential is a function of the electrolyte resistivity, temperature, pH, and ionic composition. The electrolyte conductivity, temperature, pH, and chemical compositions of the injected water of the 1997 and 2003 experiments are summarized in Table 2. The large changes of electrolyte conductivity and chemical compositions were not observed. Since the concentration of aluminum ion was equal to or less than 0.1 mg/L that is the detection limit of aluminum ion, we could not conclude that there was the change of the concentration of aluminum ion. The absolute magnitude
Fig. 8. Time evolution of the ratio of hydraulic parameters (k: permeability, ϕ: porosity, and t: tortuosity) to electrokinetic parameters (ε: dielectric constant and ζ: zeta potential). The solid line is the linear regression line and curved broken lines represent 95% confidence belt around regression.
207
Table 2 The chemical compositions of the injection water to DPRI 1800 m borehole, pH, the electrical conductivity, and temperature Date of water sampling Feb. 13, 1997 Mar.14, 2003 May 14, 2003 +
Na (mg/L) 21.4 K+ (mg/L) 2.88 Ca+ (mg/L) 34.0 5.72 Mg2+ (mg/L) Al3+ (mg/L) 0.1 Cl− (mg/L) 18.0 SO2− 21.1 4 (mg/L) 115 HCO3− (mg/L) pH 9.2 Conductivity (mS/m) 28.8 Temperature 7.8
18.3 2.91 25.9 6.22 b0.1 19.2 30.7 83.4 7.3 28.0 15.0
16.9 3.20 23.7 4.12 b0.1 17.4 25.1 71.2 8.3 24.4 21.7
of the zeta potential increases with decreasing the concentration of aluminum ion and increasing the temperature and pH (Ishido and Mizutani, 1981; Lorne et al., 1999). The absolute value of the zeta potential of quartz increases by 4.7 mV using the coefficient of −0.65 mV/°C at pH 6.1 (Ishido and Mizutani, 1981) when the temperature increases from 7.8 °C to 15.0 °C. The absolute value of the zeta potential increases by 19% according to the experimental equation for sandstone of Lorne et al. (1999) when pH decreases from 9.2 to 7.3. These variations of the zeta potential with pH and temperature, which are rough estimates since physical and chemical conditions are not equal to the real conditions, cannot be accounted for the observed decrease of 40% in the absolute value of the ratio of Jtotal/Itotal. From our observation result, we conclude that the hydraulic parameters of the hanging side of the Nojima fault decrease approximately 40% during eight years after the 1995 Hyougo-ken Nanbu earthquake. Tokunaga (1999) reported that the hydraulic permeability at just after the earthquake increases at least 5 times over the pre-seismic value from the analysis of observed hydrologic changes related to the earthquake in Awaji Island. Our result suggests that the hydraulic parameters (especially, permeability) changed by the earthquake are rapidly decreasing due to the healing processes (e.g. the compaction or the cementation process) around the Nojima fault. However, further studies are required to confirm it. Acknowledgements It is a pleasure to acknowledge the hospitality and encouragement of the members of the Nojima Fault Zone Probe. We also thank Setsuro Nakao of the
208
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Disaster Prevention Research Institute, Kyoto University for his assistance with the sampling of injection water. We also thank two anonymous reviewers for their valuable comments for improvement of the original manuscript. References Ando, M., 2001. Geological and geophysical studies of the Nojima Fault from drilling: an outline of the Nojima Fault Zone Probe. The Island Arc 10, 206–214. Awata, Y., Mizuno, K., Sugiyama, Y., Shimokawa, K., Imura, R., Kimura, K., 1995. Surface faults associated with the Hyogoken– Nanbu earthquake of 1995. Chishitsu News 486, 16–20. Ciaramella, A., De Martino, S., Di Lieto, B., Falanga, M., Ruggiero, L., Scarpa, R., Tagliaferri, R., 2004. Blind separation of seismological signals by using independent component analysis in time and frequency domain. Geophysical Research Abstracts 6, 2018. Cichocki, A., Amari, S., 2002. Adaptive Blind Signal and Image Processing. John Wiley & Sons, Ltd., England. Filipe, A., Rossow, W.B., Chedin, A., 2002. Rotation of EOFs by the independent component analysis: toward a solution of the mixing problem in the decomposition of geophysical time series. Journal of the Atmospheric Sciences 59, 111–123. Fujimori, K., Ishii, H., Mukai, A., Nakao, S., Matsumoto, S., Hirata, Y., 2001. Strain and tilt changes measured during a water injection experiment at the Nojima fault zone. The Island Arc 10, 228–234. Ishido, T., Mizutani, H., 1981. Experimental and theoretical basis of electrokinetic phenomena in rock–water systems and its applications to geophysics. Journal of Geophysical Research 86, 1763–1775. Ishido, T., Mizutani, H., Baba, K., 1983. Streaming potential observations, using geothermal wells and in situ electrokinetic coupling coefficients under high temperature. Tectonophysics 91, 89–104. Kitagawa, Y., Fujimori, K., Koizumi, N., 2002. Temporal change in permeability of the rock estimated from repeated water injection experiments near the Nojima fault in Awaji Island, Japan. Geophysical Research Letter 29. doi:10.1029/2001GL014030.
Kitagawa, Y., Fujimori, K., Koizumi, N., 2007. Temporal change in permeability of the Nojima fault zone by repeated water injection experiments. Tectonophysics 443, 183–192 (this issue). doi:10.1016/j.tecto.2007.01.012. Lorne, B., Perrier, F., Avouac, J.P., 1999. Streaming potential measurements 1.Properties of the electrical double layer from crushed rock samples. Journal of Geophysical Research 104, 857–877. Makino, Y., Oshiman, N., 1998. The three-dimensional resistivity structure of the Nojima fault determined by the hole-to-surface dipole method. Earth Monthly-Extra 21, 76–81. Marquis, G., Darnet, M., Sailhac, P., Singh, A.K., Gerard, A., 2002. Surface electric variations induced by deep hydraulic stimulation: an example from the Soultz HDR site. Geophysical Research Letters 29. doi:10.1029/2002GL015046. Matsushima, N., Tosha, T., Ishido, T., Deleman, I., Kiryukin, A., 1997. Self-potential monitoring around a well in Mutnovsky geothermal field, Kamchatka. Journal of the Geothermal Research Society of Japan 19, 157–164. Mukai, A., Fujimori, K., 2003. Permeability in fracture zone nearby the Nojima fault estimated using strain changes due to water injection experiments, Zisin. Journal of the Seismological Society of Japan 56, 171–179. Murakami, H., Hashimoto, T., Oshiman, N., Yamaguchi, S., Honkura, Y., Sumitomo, N., 2001. Electrokinetic phenomena associated with a water injection experiment at Nojima Fault on Awaji Island, Japan. The Island Arc 10, 244–251. Murata, A., Takemura, K., Miyata, T., Lin, A., 2001. Quaternary vertical offset and average slip rate of the Nojima Fault on Awaji Island, Japan. The Island Arc 10, 360–367. Shimazaki, K., Ando, M., Nishigami, K., Oshiman, N., 1998. Water injection experiments at Ogura along the Nojima fault, Japan. Earth Monthly-Extra 21, 33–37. Tokunaga, T., 1999. Estimating permeability change and specific yield from earthquake-induced hydrological changes: an example from Kobe earthquake. Journal of the Japan Society of Engineering Geology 40, 99–106. Yamano, M., Goto, S., 2005. Long-term monitoring of temperature profiles in a deep borehole: temperature variations associated with water injection experiments and natural groundwater discharge. Physics of the Earth and Planetary Interiors 152, 326–334.
Time evolution of hydraulic and electrokinetic parameters around the Nojima fault, Japan, estimated by an electrokinetic method Hideki Murakami a,⁎, Naoto Oshiman b , Satoru Yamaguchi c , Takeshi Hashimoto d , Ryokei Yoshimura b a
Department of Natural Environmental Science, Faculty of Science, National University Corporation Kochi University, Kochi 780-8520, Japan b Disaster Prevention Research Institute, National University Corporation Kyoto University, Uji 616-0011, Japan c Department of Earth and Planetary Sciences, Faculty of Science, National University Corporation Kobe University, 657-8501, Japan d Institute of Seismology and Volcanology, National University Corporation Hokkaido University, Sapporo 060-0810, Japan Received 14 May 2004; accepted 29 January 2007 Available online 27 March 2007
Abstract The Nojima Fault Zone Probe was designed to study the properties and healing processes of the Nojima fault, which is the surface fault rupture of the Hyogo-ken Nanbu earthquake (M7.2) of 1995 (1995 Kobe earthquake). In this project, water injection experiments were conducted in a borehole of 1800 m depth at the Nojima fault. We set up electrodes around the borehole and observed self-potential variations to investigate the magnitude of electrokinetic and hydraulic parameters around the Nojima fault zone. In the 1997 experiment, self-potential variations were in the range of a few to about 20 mV across 320–450 m electrode dipoles with hydraulic pressure variations from 3.5 to 4 MPa. In the 2000 experiment, self-potential variations were in the range of a few to about 85 mV across 160–260 m electrode dipoles with the hydraulic pressure variations from 3 to 4.5 MPa. In the 2003 experiment, self-potential variations were in the range of a few to about 30 mV across 20–80 m electrode dipoles with hydraulic pressure of 4 MPa. These observed self-potential variations were explained well with an electrokinetic effect due to the underground flow of the injected water. From the observed results, we estimated that the ratio of hydraulic parameters (permeability, porosity, and tortuosity) to electrokinetic parameters (zeta potential and dielectric constant) decreased approximately 40% during eight years after the earthquake. This result suggests that the healing process around the fault zone progress. © 2007 Elsevier B.V. All rights reserved. Keywords: Nojima Fault Zone Probe; Water injection experiment; Self-potential; Electrokinetic phenomena; Healing process; Active fault
1. Introduction It is generally accepted that crustal earthquakes are caused by sudden displacement on faults and large earthquakes repeatedly occur along a large-scale fault. However, fault healing and re-strengthening, which occurred just after the previous earthquake, are not clear. ⁎ Corresponding author. Fax: +81 88 844 8356. E-mail address: [email protected] (H. Murakami). 0040-1951/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2007.01.019
The Nojima Fault Zone Probe was designed to study the properties and healing processes of the Nojima fault, which moved during the 1995 Hyogo-ken Nanbu earthquake (Kobe earthquake;MJMA7.2) (Ando, 2001). Shimazaki et al. (1998) have shown a concept of the repeated water injection experiments that the healing process of a fault zone can be detected by monitoring its permeability derived from water injections into the fault zone. In order to carry out the water injection experiments, three boreholes with depths of 500 m,
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
201
800 m, and 1800 m were drilled into or near the fault zone by the Disaster Prevention Research Institute (DPRI), Kyoto University. Water injection experiments using the DPRI 1800 m borehole were repeated three times at 1997, 2000, and 2003 to investigate the properties and healing processes of the fault zone. Many kinds of geological and geophysical observations were done (e.g. Fujimori et al., 2001; Kitagawa et al., 2002; Mukai and Fujimori, 2003; papers in this issue). We have carried out monitoring of surface electric potential variations around the DPRI 1800 m borehole during water injection experiments to investigate the electrokinetic phenomena associated with the ground water flow from the borehole to the fault zone. It has been known that fluid flow through a porous medium generates an electric potential variation known as a streaming potential, which is one of the electrokinetic phenomena (Ishido and Mizutani, 1981). The selfpotential method has been applied to study geophysical phenomena related to fluid flow (e.g. Ishido et al., 1983; Marquis et al., 2002). We already reported self-potential variations associated with the 1997 water injection experiment in a previous paper (Murakami et al., 2001). In this paper we report the results of 2000 and 2003 experiments and show the evolution of hydraulic and electrokinetic parameters around the Nojima fault. 2. Field observation method Fig. 1 shows locations of the DPRI 1800 m borehole, the Nojima earthquake fault, and the Nojima branch
Fig. 2. Location of the DPRI 1800 m borehole and Cu–CuSO4 electrodes used for measuring self-potentials. (a) (■) Cu–CuSO4 electrodes used during the 1997 water injection experiment and (○) DPRI 1800 m borehole. (b) (■) Cu–CuSO4 electrodes used during the 2000 water injection experiment and (○) DPRI 1800 m borehole. (c) (■) Cu–CuSO4 electrodes used during the 2003 water injection experiment and (○) DPRI 1800 m borehole.
Fig. 1. Location of the Nojima fault, its branch fault, and (○) Disaster Prevention Research Institute (DPRI) 1800 m borehole (water injection well) in Awaji Island, Japan.
fault. The water injection borehole locates near the southwestern end of the Nojima fault. This borehole is drilled vertically for the first 1200 m and then bent
202
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Table 1 Summary of the water injection experiments of the Nojima Fault Zone Probe Year
Period
Pressure (MPa)
Injection water flow rate (L/min)
1997
Feb. 9–11 Feb. 12–13 Mar.16–25 Jan. 22–26 Jan. 31–Feb. 5 Mar. 3–11 Mar. 13–23 May 8–13
2.8–3.7 3.4–4.2 4.4 3.0 4.0 4.6 3.0–4.5 4.0
8–10 12–20 18 10–12 16–18 20–26 15–23 15–25
2000
2003
gradually with maximum deviation of 20° from the vertical and reaches the fault zone at 1800 m depth (Ando, 2001). The Nojima fault is a surface rupture of the 1995 Hyogo-ken Nanbu earthquake (Kobe earthquake) and is activated as a right-lateral strike–slip fault with a reverse sense (Murata et al., 2001). Along the Nojima branch fault only small surface ruptures occur during the earthquake (Awata et al., 1995).
Fig. 2(a), (b), and (c) show the location of electrodes of 1997, 2000, and 2003 experiments. We used Cu–CuSO4 electrodes to measure the self-potential and berried electrodes 20 cm depth into the ground. As shown in Fig. 2, sixteen electrodes were distributed around the 1800 m borehole in the 1997 experiment, twenty-two electrodes in the 2000 experiment, and twenty-two electrodes in the 2003 experiment. Electrical potential difference between each observation point and a common point (indicated by ‘common ground’ in Fig. 2) was monitored on digital recorders with a 10 s sampling period. The water injection conditions are summarized in Table 1. The 1997 water injection experiment was made over three periods: Feb. 9–11, Feb. 12–13, and Mar. 16–25, 1997. And, we observed the self-potential variations from Feb. 9 through Feb. 13, 1997. The 2000 water injection experiment was made over three periods: Jan. 22–26, Jan. 31–Feb. 5, and Mar. 3–7, 2000. And, we analyzed the data observed during three periods. The 2003 experiment was made over two periods: Mar. 13–23 and May 8–13, 2003. And, we analyzed only the data
Fig. 3. Noise reduction data of self-potentials to the electrode located at the site D2 in Fig. 2(a) during the water injection experiment of Feb. 9–13, 1997. Pressure and flow rate of the injection water are shown in the bottom.
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
observed during Mar. 13–23 in this paper because used electrodes were unstable during the second period. 3. Observation results Fig. 3 shows filtered self-potential variations, flow rate of injection, and hydraulic pressure at the wellhead of the 1800 m borehole during the 1997 water injection experiment. Self-potentials with respect to the electrode located at the site D2 in Fig. 2(a) are shown here in order to compare the previous result (Murakami et al., 2001). In order to reduce large artificial noise and diurnal variations, we applied independent component analysis (ICA) that is a statistical and computational technique for finding underlying factors or components from multivariate (multidimensional) statistical data. This method is similar to the principal component method, but the sources in ICA are assumed both nongaussian and statistically independent. Recently, ICA has been applied to geophysical problems (e.g. Filipe et al., 2002; Ciaramella et al., 2004). We analyzed the observed selfpotential data using ICALAB toolboxes for MATLAB (Cichocki and Amari, 2002). The obtained results were
203
consistent with the previous results (Murakami et al., 2001). The self-potential variations of ∼ 20 mV in response to water injection operations were extracted and its magnitude decreased with increasing distance from the injection well. Fig. 4 shows a typical example of filtered selfpotential variations during the 2000 water injection experiment. This figure shows self-potential variations during Mar. 3–7, 2000. Self-potentials with respect to the common ground electrode located 256 m southwest of the 1800 m borehole are shown. The self-potential variations of ∼ 100 mV were observed and were synchronized with turning on and off the water injection. However, the amplitudes of self-potential variations during the experiment were not constant when the pressure and fluid flow rate were approximately constant. These variations may be reflected in the complex flow of subsurface ground water. Fig. 5 shows filtered self-potential variations during Mar. 13–23, 2003. Self-potentials at each electrode with respect to the electrode located at the site B7 in Fig. 2(c) are shown here because the common electrode was unstable. The data at sites C1–D4 were also omitted
Fig. 4. Noise reduction data of self-potentials to the common ground electrode in Fig. 2(b) during the water injection experiment of Mar. 3–11, 2000. Pressure and flow rate of the injection water are shown in the bottom.
204
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Fig. 5. Noise reduction data of self-potentials to the electrode located at the site B7 in Fig. 2(c) during the water injection experiment of Mar. 13–23, 2003. Pressure and flow rate of the injection water are shown in the bottom.
because many electrodes were unstable. The selfpotential variations of ∼ 30 mV synchronized with the water injection operations were observed. As described in the previous paper (Murakami et al., 2001), the following three characters of self-potential variations were observed in common with the three water injection experiments: 1) self-potential variations appeared to correspond to the operation of water injections, 2) these variations were observed at all the observation sites around the water injection borehole, 3) the negative voltage appeared around the water injection borehole. 4. Discussion and conclusion As already described in the previous paper (Murakami et al., 2001), the characters of self-potential changes during the water injection experiments suggest the possibility that all of the observed self-potential changes represent an electrokinetic effect due to the underground flow of the injected water. Thus we consider the model based on electrokinetic effects shown in Fig. 6. Ishido and Mizutani (1981)
have described the phenomenological equations of electrokinetic effect in rock–water systems are: i ¼ /t 2 rjV þ /t 2 ðef=lÞjP
ð1Þ
j ¼ /t 2 ðef=lÞjV ðk=lÞjP
ð2Þ
in which i is the drag electric current density (A/m 2 ), j is the fluid flow flux (m/s); V is the electric potential (V), P is the pore fluid pressure (N/m); ϕ, t, and k are the porosity, tortuosity, and permeability (m 2 ) of the aquifer, respectively; ε and μ are the dielectric constant (F/m) and viscosity (Ns/m 2 ) of water, respectively; ζ is the zeta potential (V); σ = σf + m − 1 σs (σf, pore fluid electric conductivity; σs, surface electric conductivity; and m, hydraulic radius). In the case of neglecting the quantity ϕt − 2 ε 2 ζ 2 / σμk (Murakami et al., 2001), the relation between the total conductive electric current Itotal (A) and the amount of injection water Jtotal (m3 /s) is given by the following equation: Itotal ¼ idrag d2krH ¼ ½ef=ðk=/t 2 ÞJtotal
ð3Þ
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
205
Fig. 6. Line source model for self-potential variations. Electric field can be produced by the subsurface fluid flow through electrokinetic coupling. The change in voltage in the aquifer is conducted to the whole part of the injection well through the highly conductive iron casing pipe. In this model, we may estimate the total conductive current from the semilogarithmic plot of distance vs. self-potential changes.
in which H is the thickness of the aquifer and r is the pore radius. In the above model the injection well will be electrically negative if the zeta potential in aquifer is negative. In the water injection experiments, negative voltage appeared on the ground surface around the water injection borehole because the change in voltage in the aquifer was conducted to the whole part of the well through the iron borehole pipe that is not insulated electrically. In the experiments the iron borehole pipe acted as a line source of electric current. In this model the potential difference between two points on the surface is expressed by the following equation: ! dz0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi b2 þ z20 0 0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! Itotal b L þ L2 þ a2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi log d ¼ a L þ L2 þ b 2 2krc L
Itotal DV ¼ 2krc L
Z
L
dz0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi a2 þ z20
Z
L
ð4Þ
in which ΔV is the potential difference between two points on the surface, Itotal is the total conductive current flowing out from the borehole, L is the length of line current source
(iron borehole pipe), σc is the averaged electrical conductivity of the ground, and a is the distance of each electrode from the borehole, b is the distance of the reference electrode from the borehole. In the case of a, b ≪L, we can use the following approximate expression (Matsushima et al., 1997): DV ¼ ðItotal =2krc LÞlogðb=aÞ ¼ ðItotal =2krc LÞðlogb logaÞ;
½a; b≪L:
ð5Þ
In this study we estimated the total conductive electric current and the standard error of it using Eq. (5) and the following parameters: L = 1800 m, σc = 0.00125 S/m (Makino and Oshiman, 1998), and b (= 104 m for the 1977 experiment, 256 m for the 2000 experiment, and 74 m for the 2003 experiment). Fig. 7 shows the magnitude of self-potential change as a function of the distance from the borehole. Solid symbols indicate the observed self-potential changes and solid lines are linear regression lines determined using the method of least squares. The value of the total conductive current derived from the slope of the regression line and standard error are shown in Fig. 7.
206
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Fig. 7. Semilogarithmic plot of distance vs. self-potential changes. Closed symbols are observed values of self-potential change and solid lines are linear regression lines. (a) The 1997 water injection experiment. (b-1) and (b-2) the 2000 water injection experiment. (c) the 2003 water injection experiment.
The total conductive currents of − 0.18 to − 0.31 A during the 1997 experiment from Fig. 5(a), the total conductive currents of − 0.20 to − 0.61 A during the 2000 experiment from Fig. 5(b-1) and (b-2), and the
total conductive current of − 0.41 A during the 2003 experiment were estimated. Fig. 8 shows the evolution of the ratio of the amount of injected water to the total conductive electric current.
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
The absolute value of the ratio of Jtotal/Itotal decreases with time. The absolute value of the ratio in the 2000 experiment decreases by nearly 80% of that in the 1997 experiment. The absolute value of the ratio in the 2003 experiments is equal to or 10% smaller than that of the 2000 experiment. This suggests the possibility that the hydraulic parameter k/(ϕt− 2) decreases with time if electric parameters (ε and ζ) are not changed. The same result was also obtained from the analysis of permeability using the water level changes (Kitagawa et al., 2002, 2007-this issue) and using strain changes (Mukai and Fujimori, 2003) in the DPRI 800 m borehole, which is located about 40 m south from the DPRI 1800 m borehole. These results suggest that the hydraulic parameters at about 550 m depth in the hanging side of the Nojima fault decrease with time. Yamano and Goto (2005) has estimated from the temperature changes in the 1800 m borehole that the injected water leaked out of the borehole at about 550 m depth. We measured the chemical and physical elements of the injected water because the zeta potential is a function of the electrolyte resistivity, temperature, pH, and ionic composition. The electrolyte conductivity, temperature, pH, and chemical compositions of the injected water of the 1997 and 2003 experiments are summarized in Table 2. The large changes of electrolyte conductivity and chemical compositions were not observed. Since the concentration of aluminum ion was equal to or less than 0.1 mg/L that is the detection limit of aluminum ion, we could not conclude that there was the change of the concentration of aluminum ion. The absolute magnitude
Fig. 8. Time evolution of the ratio of hydraulic parameters (k: permeability, ϕ: porosity, and t: tortuosity) to electrokinetic parameters (ε: dielectric constant and ζ: zeta potential). The solid line is the linear regression line and curved broken lines represent 95% confidence belt around regression.
207
Table 2 The chemical compositions of the injection water to DPRI 1800 m borehole, pH, the electrical conductivity, and temperature Date of water sampling Feb. 13, 1997 Mar.14, 2003 May 14, 2003 +
Na (mg/L) 21.4 K+ (mg/L) 2.88 Ca+ (mg/L) 34.0 5.72 Mg2+ (mg/L) Al3+ (mg/L) 0.1 Cl− (mg/L) 18.0 SO2− 21.1 4 (mg/L) 115 HCO3− (mg/L) pH 9.2 Conductivity (mS/m) 28.8 Temperature 7.8
18.3 2.91 25.9 6.22 b0.1 19.2 30.7 83.4 7.3 28.0 15.0
16.9 3.20 23.7 4.12 b0.1 17.4 25.1 71.2 8.3 24.4 21.7
of the zeta potential increases with decreasing the concentration of aluminum ion and increasing the temperature and pH (Ishido and Mizutani, 1981; Lorne et al., 1999). The absolute value of the zeta potential of quartz increases by 4.7 mV using the coefficient of −0.65 mV/°C at pH 6.1 (Ishido and Mizutani, 1981) when the temperature increases from 7.8 °C to 15.0 °C. The absolute value of the zeta potential increases by 19% according to the experimental equation for sandstone of Lorne et al. (1999) when pH decreases from 9.2 to 7.3. These variations of the zeta potential with pH and temperature, which are rough estimates since physical and chemical conditions are not equal to the real conditions, cannot be accounted for the observed decrease of 40% in the absolute value of the ratio of Jtotal/Itotal. From our observation result, we conclude that the hydraulic parameters of the hanging side of the Nojima fault decrease approximately 40% during eight years after the 1995 Hyougo-ken Nanbu earthquake. Tokunaga (1999) reported that the hydraulic permeability at just after the earthquake increases at least 5 times over the pre-seismic value from the analysis of observed hydrologic changes related to the earthquake in Awaji Island. Our result suggests that the hydraulic parameters (especially, permeability) changed by the earthquake are rapidly decreasing due to the healing processes (e.g. the compaction or the cementation process) around the Nojima fault. However, further studies are required to confirm it. Acknowledgements It is a pleasure to acknowledge the hospitality and encouragement of the members of the Nojima Fault Zone Probe. We also thank Setsuro Nakao of the
208
H. Murakami et al. / Tectonophysics 443 (2007) 200–208
Disaster Prevention Research Institute, Kyoto University for his assistance with the sampling of injection water. We also thank two anonymous reviewers for their valuable comments for improvement of the original manuscript. References Ando, M., 2001. Geological and geophysical studies of the Nojima Fault from drilling: an outline of the Nojima Fault Zone Probe. The Island Arc 10, 206–214. Awata, Y., Mizuno, K., Sugiyama, Y., Shimokawa, K., Imura, R., Kimura, K., 1995. Surface faults associated with the Hyogoken– Nanbu earthquake of 1995. Chishitsu News 486, 16–20. Ciaramella, A., De Martino, S., Di Lieto, B., Falanga, M., Ruggiero, L., Scarpa, R., Tagliaferri, R., 2004. Blind separation of seismological signals by using independent component analysis in time and frequency domain. Geophysical Research Abstracts 6, 2018. Cichocki, A., Amari, S., 2002. Adaptive Blind Signal and Image Processing. John Wiley & Sons, Ltd., England. Filipe, A., Rossow, W.B., Chedin, A., 2002. Rotation of EOFs by the independent component analysis: toward a solution of the mixing problem in the decomposition of geophysical time series. Journal of the Atmospheric Sciences 59, 111–123. Fujimori, K., Ishii, H., Mukai, A., Nakao, S., Matsumoto, S., Hirata, Y., 2001. Strain and tilt changes measured during a water injection experiment at the Nojima fault zone. The Island Arc 10, 228–234. Ishido, T., Mizutani, H., 1981. Experimental and theoretical basis of electrokinetic phenomena in rock–water systems and its applications to geophysics. Journal of Geophysical Research 86, 1763–1775. Ishido, T., Mizutani, H., Baba, K., 1983. Streaming potential observations, using geothermal wells and in situ electrokinetic coupling coefficients under high temperature. Tectonophysics 91, 89–104. Kitagawa, Y., Fujimori, K., Koizumi, N., 2002. Temporal change in permeability of the rock estimated from repeated water injection experiments near the Nojima fault in Awaji Island, Japan. Geophysical Research Letter 29. doi:10.1029/2001GL014030.
Kitagawa, Y., Fujimori, K., Koizumi, N., 2007. Temporal change in permeability of the Nojima fault zone by repeated water injection experiments. Tectonophysics 443, 183–192 (this issue). doi:10.1016/j.tecto.2007.01.012. Lorne, B., Perrier, F., Avouac, J.P., 1999. Streaming potential measurements 1.Properties of the electrical double layer from crushed rock samples. Journal of Geophysical Research 104, 857–877. Makino, Y., Oshiman, N., 1998. The three-dimensional resistivity structure of the Nojima fault determined by the hole-to-surface dipole method. Earth Monthly-Extra 21, 76–81. Marquis, G., Darnet, M., Sailhac, P., Singh, A.K., Gerard, A., 2002. Surface electric variations induced by deep hydraulic stimulation: an example from the Soultz HDR site. Geophysical Research Letters 29. doi:10.1029/2002GL015046. Matsushima, N., Tosha, T., Ishido, T., Deleman, I., Kiryukin, A., 1997. Self-potential monitoring around a well in Mutnovsky geothermal field, Kamchatka. Journal of the Geothermal Research Society of Japan 19, 157–164. Mukai, A., Fujimori, K., 2003. Permeability in fracture zone nearby the Nojima fault estimated using strain changes due to water injection experiments, Zisin. Journal of the Seismological Society of Japan 56, 171–179. Murakami, H., Hashimoto, T., Oshiman, N., Yamaguchi, S., Honkura, Y., Sumitomo, N., 2001. Electrokinetic phenomena associated with a water injection experiment at Nojima Fault on Awaji Island, Japan. The Island Arc 10, 244–251. Murata, A., Takemura, K., Miyata, T., Lin, A., 2001. Quaternary vertical offset and average slip rate of the Nojima Fault on Awaji Island, Japan. The Island Arc 10, 360–367. Shimazaki, K., Ando, M., Nishigami, K., Oshiman, N., 1998. Water injection experiments at Ogura along the Nojima fault, Japan. Earth Monthly-Extra 21, 33–37. Tokunaga, T., 1999. Estimating permeability change and specific yield from earthquake-induced hydrological changes: an example from Kobe earthquake. Journal of the Japan Society of Engineering Geology 40, 99–106. Yamano, M., Goto, S., 2005. Long-term monitoring of temperature profiles in a deep borehole: temperature variations associated with water injection experiments and natural groundwater discharge. Physics of the Earth and Planetary Interiors 152, 326–334.