Continuous Monitoring of Heavy Metals in Groundwater As a Prediction of Earthquakes
Continuous monitoring of heavy metals in ground water as a tool for the detection and verification of earthquake activity
Øyvind Mikkelsen1*, Silje M. Skogvold1, Tonje B. Østebrød1, Kristina Strasunskiene1, Lillemor Claesson2, Alasdair Skelton3.
1Norwegian University of Science and Technology, Department of Chemistry, N-7491 Trondheim, Norway.
2 Swedish Nuclear Fuel and Waste Management, S-10450 Stockholm, Sweden
3Stockholm University, Department of Geology and Geochemistry, S-10691 Stockholm, Sweden
*e-mail:
Abstract
Development of an automatic trace metal monitoring station has provided an opportunity to monitor chemical signals in ground water related to earthquake activity. The system was installed at a ground water source in Iceland, and continuous determination of metal concentrations clearly indicated changes in the chemical composition of ground water prior to an earthquake. The results revealed increased electrolabile concentrations of zinc, iron and copper 7-8 days before an earthquake, which lasted until the occurrence of the earthquake. These changes detected thus have the potential to be interpreted as earthquake precursors. The automatic monitoring station determines the metal concentrations through the use of a solid silver amalgam working electrode.
Keywords: voltammetry, solid silver amalgam electrode, continuous monitoring, earthquake activity, ground water
1. Introduction
Earthquake is a phenomenon that affects people all over the world. In areas with high earthquake activity, earthquakes can occur at any time – and they may result in severe material damage and loss of human lives. Prediction of earthquakes is therefore a prioritized and wide research field, including investigation of behavior of ants [1] to seismological factors [2] and hydrogeology. Studies of hydrogeological changes prior to earthquakes began in the late sixties and early seventies, and some of the many hydrological changes investigated as possible indicators of earthquake activity are; pressure, flow rate, color, taste and chemical composition of surface and ground water, oil and gas [3, 4]. Increased concentrations of radon in ground water were detected one year before an earthquake in Tashken, Usbekistan in April 1966 [4]. In Russia in 1969, changes in the conductivity in ground water was observed six months before an earthquake [5]. Changes in the frequency of seismic oscillations have also been detected before several earthquakes in Russia (1972) and North America (1973) [5, 6]. Before an earthquake in Kobe (Japan) in 1995, significant changes in the ground water flow were reported, in addition to increased water temperature[7]. Analyses of ground water in Kobe also showed increased chloride concentrations with maximum values detected four days before the earthquake Concentrations of other ions (Na+, Ca2+, HCO3- and SO42-) in ground water have also been used as earthquake precursors [8]. The many different hydrological changes observed is said to have a common physical explanation and because of this Scholtz claimed that it should be possible to predict most, if not all, earthquakes [4]. Expansions of bedrock and diffusion of water is suggested as a reason for most observed earthquake precursors, and the most successful precursor has so far been changes in ion concentration in water [9]. Reviews of earthquake precursor are given in [3, 9].
In 2002 an earthquake measured to MW 5.8, occurred near the village of Húsavík at the Grimsey Lineament at Iceland [10]. Iceland straddles the Mid-Atlantic spreading ridge. Seismic activity occurs at shallow depths and is mainly concentrated to the two transform zones; the South Iceland Seismic Zone and the Tjörnes Fracture Zone [10, 11]. Húsavík is situated close to the Húsavík-Flatey Fault, which forms the offshore part of the Tjörnes Fracture Zone in northern Iceland together with the Grimsey Lineament. Ground water in Húsavík is affected by changes in ground water fluxes at these faults. The type of rock present determines the composition of the water. The main leakage of water in Húsavík is going through an area where basalt is the dominating rock [12]. Basalt mainly consists of the minerals plagioclase and pyroxene at approximately equal amounts. Investigations of water trapped inside warm basalt along the Mid-Atlantic spreading ridge has revealed water with lower pH and high amounts of Ca, K, Fe, Mn, Zn, Cu, SiO and H2S [13].
Analyses of ground water sampled in the area of Húsavík during 2002 showed a 12-19 % increase in concentrations of a number of trace elements in samples taken 2-9 days before the earthquake occurred. This provides strong evidence for a seismic-hydrogeochemical coupling [10]. The largest anomaly was observed for zinc two weeks before the earthquake occurred, and also copper, manganese and iron was found to increase significant. The increase in metal concentrations reflects a result of increased water-rock interaction between the warm water and the basalt, since these metals are leached preferentially from basalt containing rocks. The anomalies were therefore explained by the fact that warmer water containing higher concentration of heavy metals could leak into the ground water source due to changes in flow patterns. Another research project in the Húsavík-Flatey fault has indicated propagation through cracks containing fluids at high pore-fluid pressures as a reason for the earthquake. High pore-fluid pressures play a key role in earthquake source mechanisms, and are necessary in seismically-active fault zone to release friction stress and allowing breakage and earthquakes [14]. It has therefore been argued that the observed anomalies could be possible earthquake precursors, relating to pre-seismic modification of crustal permeability [10]. A summary of several research projects confirms that changes in ion concentration in ground water may be used for predicting earthquakes [9]. The reported changes in [10] still failed to meet the validation criteria of the International Association of Seismology and Physics of the Earth’s Interior sub-commission on earthquake prediction [15, 16] demanding detection at more than one site or by using more than one instrument. The argument that these anomalies are earthquake precursors is further weakened because the anomalies were not detected until after the earthquake had occurred, since all samples were manually collected and shipped for detection in the laboratory.
The increased metal concentrations, reported by Claesson et al. 2004 are however still of interest for further evaluation as possible earthquake precursors, and in this study we show how the limitations mentioned above can be overcome using on-line automatic monitoring of metal concentrations. Since the concentrations of the actual metal ions often are very low in natural water samples, sensitive and sophisticated instruments are required, rarely being suitable for remote operations in the field. There are only a few methods available for automatic on-line field monitoring, and among these are dynamic electroanalytical techniques. Voltammetry has proved to be a suitable method for continuous automatic monitoring, combining speed, sensitivity and low costs as well as good sensitivity for several metals [17, 18]. Voltammetry in combination with conventional methods like ICP-MS and AAS might also yield valuable information about speciation in addition to serve as quality assurance.
Another advantage of voltammetric analyses is that it reports the electrolabile content, which includes free aqua ions, weak inorganic complexes and readily dissociable organic complexes. This makes it possible to determine chemical speciation of metals directly in a sample. In connection to earthquake activity, this might be of great importance since changes in pH or temperature may cause changes in the composition of ground water, which again may have significant influence on the balance between various metal species including the electrolabile fraction [19, 20].
There are several papers dealing with automated voltammetric systems, but the majority of these papers report systems using liquid mercury or mercury film electrodes [21-23]. The choice of electrode material is very important, and several attempts have been made to find solid electrodes with properties desired for field equipment. The introductions of solid amalgams and alloy electrodes have created new possibilities for constructing useful voltammetric apparatus for use in field with long-term measuring stability [18, 24-28]. It is now possible to construct apparatus with sufficient long-time stability for continuous remote surveillance of several important heavy metals. The present paper reports how an automatic trace metal monitoring system in combination with solid silver amalgam working electrode have been used to monitor chemical changes related to earthquake activity, and thus how it indirectly can be used as a method for predicting earthquakes.
2. Experimental
An automated voltammetric system (ATMS 500) from (SensAqua AS) was installed at a low-temperature geothermal ground water source in Iceland. This system is specifically designed for early warning detection and continuously monitoring of rivers, lakes and water resources. The system consists of a two cabinets, where the first contains an industry computer with a data acquisition and control unit, including internet communication with the possibility for remote system control. The second cabinet contains the voltammetric cell, an automatic sample pump and a drain valve system (Figure 1).
Fig. 1. The automatic trace metal monitoring system, consisting of an industry computer with internet connection, an automatic sampling system, and a voltammetric cell system with a solid silver amalgam electrode.
The ground water was free flowing from a borehole due to overpressure, but the efficiency (liters per minute) was enhanced by installing a down-hole pump in the borehole. The borehole penetrated the basalt-hosted aquifer at a depth of 1220m with a down hole temperature of 94-110°C. Samples were then injected as a partial flow to the voltammetric cell using an automated peristaltic pump. A coarse filter (Schott-Duran, Germany) was used to avoid larger particles to assemble in the system.
Water reaching the surface had a temperature of 75°C, and was further cooled down to a final temperature of 45°C when reaching the voltammetric cell. The samples were then added supporting electrolyte (NH4Cl) and immediately analyzed by differential pulse anodic stripping voltammetry (DPASV). The supporting electrolyte was added by a micro volume pump (ProMinet Beta, AxFlow) to a final concentration of 0.02 M. Potentials were measured vs. an Ag/AgCl/KCl (3M) reference electrode, and the counter electrode was a platinum wire. A solid silver amalgam electrode (d = 3mm) was used as working electrode. The ground water samples were analyzed automatically every third hour for a period of four weeks. Due to problems with the internet connection, the results where manually transferred from the station once a week.
Calibration of the system was performed by standard addition. Before calibration the water sample was filtered through 0.45 mm cellulose acetate filters from VWR. The system was validated for determination of zinc by quantification of Standard Reference Material SPS-SW2 Batch 115, a common reference material for measurement of elements in surface waters. A sample from the standard reference material was diluted 1:3 with distilled water, and added NH3 to pH 8. The concentration of zinc was then determined by standard addition.
3. Results and discussion
The highest change in concentration prior to the earthquake in 2002 was measured for zinc, along with copper, iron and manganese. Because of the measured concentration level of zinc in 2002, and the fact that the solid silver amalgam electrode has shown to be suitable for especially zinc analysis, zinc was chosen as an indicator metal for further experimental studies.
3.1 Calibration and validation of the system
In order to achieve reliable results, proper calibration and validation of the system is important. Direct calibration of real samples is often difficult or impossible due to the presence of colloids and particles, which will interfere through metal adsorption [29]. To avoid such adsorption problems under the calibration routine, particles should be removed by filtration or decomposition by e.g. ultra clave before calibration. In the present work, filtration of the sample through a 0.45 cellulose acetate filter was found to be satisfactory to achieve a successful calibration. A typical calibration of a filtered sample added 10 to 50 mg/L zinc, and the corresponding linearity graph (r2 = 0.971), is shown in Figure 2.
Fig. 2. Standard addition and calibration plot for zinc in a ground water sample from Húsavik. DPASV, 300 s deposition time at -1300 mV, scan rate 20 mV/s, modulation pulse 75 mV, equilibrate time 10 s, added NH4Cl (0.02 M). The baseline has been subtracted.
Validation of the measuring system and the reproducibility was carried out by quantification of zinc in a Standard Reference Material SPS-SW2 (Batch 115) as described above. Determination of zinc in Standard Reference Material is shown in Figure 3. The recovery value for zinc was found to be 90 %, of a total concentration of 33 mg/L.
Fig. 3. Quantification of Zn in certificated reference material SPS-SW2 Batch 115. DPASV, 300 s deposition time at -1300 mV, scan rate 20 mV/s, modulation pulse 75 mV, equilibrate time 10 s. The standard deviation was 2.4%.
3.2 Surveillance of ground water at Húsavík, Island
On the November 1 2006 at 1:55 pm, a Mw 4.5 earthquake occurred on the Húsavík – Flatey – fault. 150 minor aftershocks were detected the same day followed by 80 aftershocks the next day. Significant electrolabile metal concentrations were not detected during the first three weeks of the monitoring period, since concentrations lower than three times standard deviation were reported by the monitoring system as zero. However, continuous measurements showed significant changes in the voltammogram from eight days before the earthquake occurred, as seen in Figure 4.
Fig. 4. Scans from 02.10.06 (solid line) and 24.10.06 (dotted line) show changes in the voltammograms. The solid line represents scans from periods with no earthquake activity and the dotted line represents scans 8 days before the earthquake. Dep. pot. -1250 mV, start pot. -1250 mV, stop pot. -93 mV, scan rate 10 mV/s, diff. puls 75, current range + 75 mV, dep. time 600s, eq. time 5 s.
When examining the voltammetric scans, the first peak, corresponding to zinc at –1100 mV, occurred eight days prior to the earthquake as seen in Figure 4. In the same period an increase in the current was also observed at the potential around – 500 mV (Figure 4), most likely to be iron based on standard addition identification on manually collected samples (not shown). For this peak the same pattern was observed as for zinc, with the appearance of electrolabile iron 8 days before the earthquake and the highest concentration measured 7 days prior to the earthquake. A change in current was also observed at -250 mV, corresponding to copper which had a maximum electrolabile concentration also 7 days before the earthquake (Figure 4). Graphs showing daily average current values measured for zinc and copper from October 1st 2006 to November 2nd 2006 is shown in Figure 5. The samples were analyzed, and thus the changes detected, a few minutes after the sampling, but due to problems with the internet connection the results where not transferred from the system, and thus not observed, until a few days later.