Effects of seismic activity on the fluorescence signal of groundwater

Forecasting earthquakes has attracted considerable attention recently but thus far there is no suitable explanation as to why non-seismic pre-earthquake manifestations exist. Friedemann Freund proposed the theory that the existence of peroxy defects and positive holes in rocks explained these phenomena and so attempted to lay the scientific basis for the cause of many earthquake precursor signals (Freund, 2006). Freund’s theory was later expanded to show how the redox conversion of OH ⁻ pairs into peroxy anions and molecular H₂ works. Freund showed that by maintaining thermodynamic rules (Freund and Freund, 2015).

Pre-seismic events such as thermal infrared emissions, surface temperature anomalies, radon irregularities and changes in physical and chemical properties of groundwater prior to major earthquakes have been reported (Tramutoli et al., 2005; Piroddi et al., 2014; Guangmeng, 2008, Ouzounov et al., 2006; Singh et al., 2016; Biagi et al., 2000; Fidani et al., 2017). Anomalies in pH, conductivity or variations in ion content of groundwater can be explained through the interaction between positive holes and water (Grant et al., 2011). Within stressed rock peroxy links can break. When this occurs an electron next to O²⁻ anion is transferred onto the broken peroxy link. The electron donor changes its valence from 2- to 1- (Balk et al., 2009). An O⁻ surrounded by O²⁻ (termed as “positive hole”) is an oxygen anion defect by 1 electron in the O²⁻ anion sublattice and acts as a charge carrier h^. (Griscom, 2011; Freund and Freund, 2015). These charge carriers flow out of stressed rock regions and into less stressed rock regions where they can interact with groundwater.

The charge carriers are chemically seen highly oxidizing O⁻ radicals. These radicals can oxidize water to hydrogen peroxide H₂O₂ on the rock–water boundary and partially oxidize organic compounds dissolved in the groundwater. The partial oxidation of organic compounds can cause a shift in fluorescent intensity and in the fluorescent spectrum (Grant et al., 2011). An example is the oxidation of terephthalates where an O⁻ radical is added to the aromatic ring which leads to a detectable change in the fluorescent spectrum (Saran and Summer, 1999). However, exactly which dissolved organic substances are present in the oxidation reaction involving the charge carriers and their change in the fluorescent spectrum is not yet known.

The reaction between the charge carriers and the groundwater showing an increase in the fluorescent intensity signal could provide evidence of increasing tectonic stress rates in the subsurface.

A limited number of studies have monitored fluorescent intensities using long interval measurements with a flow-through fluorometer. Fluorescent intensity changes were first observed in 1999 using synchronous scans of fluorescent spectra in water samples taken by chance prior to and after a strong earthquake (Balderer and Leuenberger, 2007, Grant, R.A., T. Halliday, W.P. Balderer, F. Leuenberger, M. Newcomer, G. Cyr, F.T. Freund. 2011, Fidani et al., 2017. Our study is one of the first to monitor fluorescent intensities over a time of 11 weeks in two different springs in the northern part of Switzerland. Changes in fluorescent intensities are correlated in our study with the seismic data of nearby stations to deduce potential coincidences. The method of fluorescent monitoring used in our study and the geological and tectonic setting of our chosen physical area in Baden and Rheinfelden Switzerland are explained. Subsequently, the results are presented, discussed and conclusions about the method are drawn.

A commercially available fluorometer (GUGN-FL30) developed by the University of Neuchatel to detect events of artificial fluorescence tracer tests, was used for the continuous monitoring of fluorescence at both spring sites observed.

An optical system is located inside the flow-through fluorometer. Water flows through a quartz tube within a steel case in this device. It is crucial to maintain continuous water flow in the measurement tool. To achieve this end, the fluorometer was installed at a lower height than the water flow.

Occasional removal of calcareous deposits inside the tube is essential to ensure that measurements were not adversely affected by these deposits. In our study at both spring sites, the calcareous deposits were not significantly large enough due to the limited measuring period. Therefore, the influence of the purgation on the fluorescence measurement was assumed to be insignificant and a correction analysis for the purgation was neglected.

The optical system consists of four different LEDs. Each light source has a specific wavelength. In this case 370 nm, 470 nm, 525 nm and 660 nm were used to excite the fluorescent components in the water. The emission optics are arranged perpendicular to the excitation plains. The emissions of the fluorophores are filtered to determine the wavelength range detected by photodiodes (Schnegg, 2002). The monitored measurement interval of the fluorescence intensity in mV was 300 s. The schematic concept of the fluorometer is shown in Fig. 1.

A high turbidity (mainly caused by degassing) can adversely influence the intensity of the fluorescent measurements. In the case of the Rheinfelden spring where a high CO₂ content exists, a degassing device was installed to reduce turbidity. Besides degassing and turbidity, a possible mixing of the thermal spring water with a neighbouring aquifer can influence the fluorescence signal. The order of magnitude by which the fluorescence intensity is affected by such a mixing depends on the constituents of the inflowing aquifer. Anthropogenic influxes in the form of chemicals such as cleaning agents from nearby industry or thermal baths resorts could also adversely influence measurements taken.

The measurements presented here originate from an unused thermal deep well at Rheinfelden (latitude 47.5505°, longitude 7.8058°) and from a hot spring in Baden (latitude 47.4808°, longitude 8.3130°). The geological overview of Switzerland is presented in Fig. 2 whereas the tectonic setting and the seismicity of the investigated area are presented in Fig. 3.

The Rheinfelden thermal well temperature is 12 °C with a pH of 6.6 and a conductivity of 5420 μS/cm. This well contains Na-Ca-Cl-HCO₃-SO₄ thermal water with a high CO₂ content and originates at a depth of 550 m. Geologically it is in late Palaeozoic sediments (Perm, Rotliegendes) below dense evaporitic layers (Muschelkalk and Keuper) from the early Mesozoic era (Burger, 2009). Rheinfelden is situated in close vicinity to the Rhine Graben-structure, a 300–350 km long rift where the geology and tectonics have been studied in detail (Sissingh, 1998; Becker, 2000; Ustaszewski and Schmid, 2007). The detailed well profile of the Rheinfelden spring is presented by Ryf (1984).

The thermal springs of Baden lie in the “Lägeren” – structure where the Limmat eroded into the Muschelkalk layers in the eastern part of the Faltenjura. The Baden hot spring contains fluids originating from the base rock ascending through the evaporites of the middle Muschelkalk layers, along a tectonic fault, which is characterized as the Jura-Overthrust (Burger, 2009; Löw, 1987; Nagra, 2014, Fig. 3.3–2). The aquifer has a piezometric head of 359 m a.s.l. and is therefore artesian against the local river Limmat (350 m a.s.l.) and the surrounding groundwater streams (348–357 m a.s.l.) (Löw, 1987). The Baden spring water with a total mineralization of around 4.5 g/l and a conductivity of 5970 μS/cm and is hydro-chemically characterized by Na-Ca-(Mg)-Cl- SO₄ type and contains higher amounts of characteristic dissolved gasses, namely H₂S (3.0 mg/l), CH₄ (< 0.3 mg/l) and CO₂ (292 mg/l) (Rick, 2007; Högl, 1980). The water has a temperature around 46 °C and a pH of 6.6.

The following figures (Figs. 4 and 5) show fluorescent intensities over the measured time and the location and magnitude of the earthquakes that occurred. Only earthquakes which had a magnitude 1 or higher and were experienced within a radius of 100 km from the spring site measurement location were considered. Out of the four possible wavelength sensors only the most significant results are presented. These were using the 470 nm and 525 nm signals. The strain (ε) and strain radius of the nearby registered earthquakes are calculated using the Dobrovolsky relation (Eqs. 1 and 2) (Dobrovolsky et al., 1979).

$$ {10}^{0.43\ast \mathrm{M}}=\mathrm{strain}\ \mathrm{radius} $$

(1)

$$ {10}^{1.3\ast \mathrm{M}-8.19}=\upvarepsilon $$

(2)

The strain radius is the estimated zone around the epicentre of an effective manifestation of precursor deformation and is dependent on the magnitude M of the earthquake (Dobrovolsky et al. 1979). This relation is a very useful approach to obtain a semi-quantitative number for the precursor manifestation zone. This approach assumes that non-mechanical precursors are caused by deformation of the surrounding medium.

The theory states that a possible manifestation of a precursor occurs within the strain radius and can be detected when the monitoring site lies inside this estimated zone. The calculation of the strain combines the impact of the magnitude and the distance to the monitoring site. This result helps predict if a precursor is expected at the measurement site or not. According to Dobrovolsky et al. (1979) lies the monitoring side inside the deformation zone if the determined strain is above 10^− 8. However, whether the monitoring site lies inside the strain radius is not that significant when the expected changes in fluorescence intensities are assumed to be caused through the stress activated charge carriers.

It has been demonstrated in lab experiments that these charge carriers are highly mobile and it is assumed that they can propagate far away from the stressed rock region (Freund and Freund, 2015; Scoville et al., 2015). Nevertheless, the strain and the strain radius were calculated to follow the initial scaling approach of earthquake precursors and to get a rough estimation of the extent of the deformation zone. The positive holes are generated in the deformation zone and therefore it is expected that stronger earthquakes with larger deformation zones tend to activate more charge carriers than weaker ones. Therefore, only earthquakes with magnitude higher than 2.0 were considered for the strain respectively strain radius calculations and as possible detection targets. The four registered earthquakes with a magnitude higher than 2.0 are listed in Table 1 and mapped in Fig. 6. Data and information about registered earthquakes are provided from the SED (Swiss Seismological Service).

Earthquake	Lat / Lon [°]	M	Depth [km]	Distance to B [km]	Distance to R [km]	Strain radius [km]	Strain ε at: B	Strain ε at: R
E1	46.90 / 7.42	2.6	5.7	93.3	77.9	13.1	1.91*10−11	3.28*10−11
E2	47.15 / 7.15	2.4	9.8	95.0	66.5	10.8	9.93*10−12	2.89*10−11
E3	48.24 / 9.05	2.5	9.6	100.0	>  100	11.9	1.15*10−11	<  1.15*10− 11
E4	47.66 / 8.74	2.2	7.4	37.7	71.0	8.8	8.73*10−11	1.31*10− 11

The fluorescence intensities vary at moderate magnitudes at both spring sites over the whole monitoring period. Intervals with stable intensity values are scarce. The main trend of the observed fluorescence monitoring is dominated by several increases which are followed by sudden decreases in fluorescent intensities.

Intensified tectonic stresses in the subsurface leads to an increase of stress generated charge carriers which flow out of the stressed rock region into the less stressed rock regions. When more charge carriers are generated, the oxidation on the rock–water boundary is increased and the fluorescence intensity rises. The expected shape and the duration of the increasing fluorescence intensity curve prior to an earthquake is not known yet but it is assumed that it is most likely variable.

The theory assumes that the higher the magnitude of an earthquake is, the higher the impact of the fluorescent signal in the measured groundwater will be. Because of the lack of strong seismicity in close vicinity to the monitoring stations during the measuring period it is difficult to determine correlations between changes in fluorescence and earthquakes.

The calculated strains of the earthquakes E1 – E4 for both monitoring sites are much smaller than the postulated 10^− 8. This means that these four earthquakes should not manifest themselves at the monitoring site using the Dobrovolsky approach. However, all the four earthquakes (E1 – E4) happened at very shallow depths and it should be considered that the stress generated charge carriers can flow much further than the strain radius estimates.

At the Rheinfelden spring site the fluorescence signal steadily increased until the earthquake (E1) with a magnitude of 2.6 close to Bern happened. Subsequently the fluorescence intensity rapidly dropped after the registered earthquake. However, this phenomenon cannot be clearly identified as a coherence between the fluorescence signal and the earthquake because of the low magnitude and the large distance to the measuring station (R). Additionally, on April 11th, 2015 a similar rapid drop in fluorescence intensity occurred but in absence of an earthquake. The registered earthquakes close to Biel (E2) and Albstadt (E3) do not show any possible changes to the fluorescence signals at both spring sites. The earthquake E3 is not reflected in Fig. 5 because of the distance, which exceeded the maximum distance of 100 km.

Prior to the earthquake (E4) with a magnitude of 2.2 close to Andelfingen, an increase in fluorescence intensity was measured at both spring sites. Moreover, the fluorescence signals suddenly dropped immediately after the earthquake (E4) occurred. The increase and sudden decrease of the fluorescence intensity was more pronounced at the Baden spring (B) than in Rheinfelden (R). The reason behind this could be found in the shorter distance from the epicentre to the measuring station in Baden compared to the distance to Rheinfelden. This recorded specific change in fluorescence intensity is quite remarkable because of its occurrence at both spring sites prior to and after the earthquake. Such an event in fluorescence change would be expected and would be consistent with the theory of the release of stress generated charge carriers which cause an oxidation of the groundwater. However, that this change in the fluorescence signal can be identified because of increased and diminished tectonic stresses in the subsurface is beyond the current understanding of the phenomena. Nevertheless, this method could be a useful tool in future earthquake forecast because of its sensitivity to a wide catchment area which depends on the extent of the aquifer and its flowrate. Furthermore, the modest methodology and the straightforward installation and maintenance of the fluorometer is another point in favour of this monitoring method. The limitation of the method lies in the strong influence of outgassing and turbidity on the fluorescence signal of the groundwater. External factors which influence the fluorescence intensity besides aquifer mixing and anthropogenic influxes are not yet known and further research is needed.

The results presented clearly show the variability of fluorescence intensity of ground water at the distinct wavelengths. Furthermore, the results support a possible link between the stress generated charge carriers caused by tectonic processes leading to earthquakes and their interaction with the investigated thermal water. However, these preliminary results should be confirmed by further monitoring of fluorescence of groundwater at alternate groundwater sites in tectonically more active areas.

The selection of appropriate springs or wells is crucial. Most suited for in-situ fluorescence monitoring are sites with artesian groundwater outflow such as the investigated Baden spring and the Rheinfelden borehole. Groundwater of pumped wells are more affected by disturbances due to pumping intervals and non-steady state situations in respect to considering hydraulic head and degassing processes.

Furthermore, the monitoring period should be extended to enhance the probability of earthquake occurrences and to allow long term analysis. Fluorometers with a wider wavelength spectrum and an advanced sensitivity would be a further asset. This would allow the wavelength spectrum to be tuned to the most significant wavelengths in terms of changes in fluorescent intensities as an earthquake precursor. Furthermore, a wider wavelength spectrum could provide further information about possible shifts in fluorescence spectra. Beyond that, further investigations into the oxidation process of dissolved organic compounds in the ground water should be done in order to improve the understanding of the phenomena.