Introduction

Earthquake, which results from the movements of the Earth's crust formed, contains many parameters related to the location. Therefore, it is a complicated event to understand and resolve. Monitoring the changes in these parameters during an earthquake process and their association with the earthquake is an important research topic for the whole world. One of the crucial data used in estimating the time and location of earthquakes is the observation of radon activity changes in or near the exact location. Radon (222Rn) is the decay product of radium (226Ra) and is a heavily radioactive noble gas. It originates from the alpha particle emission of 226Ra with a rebounding energy of 86 keV. This element with atomic number 86 has three natural isotopes, and the half-life of 222Rn is 3.82 days. [1] investigated the correlation between radon and radium levels to understand whether the radon concentration they measured in well water originates directly from the radium in the surrounding rocks or from transport from a more distant source. In their study, they calculated the linear correlation coefficient between radon and radium concentration as 0.99. Researchers have conducted many studies on radon-earthquake measurement in different environments such as soil, air, and water [2,3,4]. The rebound length of radon is short (3 × 10–10 m); it passes from rocks and soil to water, and this causes the presence of radon gas in groundwater [5]. It was observed in the studies that the efficient transfer of 222Rn in groundwater from solid to solution occurs on the grain surfaces, and when the surface area of aquifer rock increases, more radon will dissolve. Okabe conducted the first studies on radon gas and seismic activities in 1956 [6]. He determined a positive relationship between atmospheric radon concentration changes near the earth's surface and regional seismic activity in Japan. Thus, he pointed to radon as a new earthquake monitor. In the study conducted by Ulomov and Mavashev [7], significant changes were observed in the concentration of radon in well waters before the Tashkent earthquake on 26 April 1966. Thus, it was revealed for the first time that radon anomalies in deep well waters could be a possible tracer for earthquakes.

Radon is found everywhere in all types of rocks, earth's crust, soil, and water. It goes into water continuously from rocks of the terrestrial crust. Due to the high solubility of radon in water, it can be transported long distances through soil by groundwater [8]. The radon transport is controlled by geophysical and geochemical parameters [9]. Many studies highlight that there may be important relationships between changes in radon concentrations in soil, groundwater, and the atmosphere and earthquake activities [10,11,12,13,14,15,16,17,18,19,20,21] However, understanding the relationship between radon gas and seismic events and radon as a preliminary tracer is quite complex. In order to overcome this complexity, various models and approaches for the propagation of radon in fault zones have been proposed by researchers through statistical and mathematical studies [13, 22,23,24,25,26,27,28,29,30,31,32,33,34]. An Autoregressive Integrated Moving Average (ARIMA) simulation model and Monte Carlo simulation, which is a relatively new method to detect radon gas anomalies [28]; ARIMA simulation model [25]; chaos methodology [24]; simultaneous investigation of the change of ionospheric total electron content (TEC) according to earthquake and radon gas concentration [32] can be given as an example of these studies.

Studies emphasize that radon gas changes in fault zones and fractures [35, 36] and that radon measurement is a good tool to detect active faults and define the actual size of fractures [17, 37]. Because radon gas concentrations in active fault regions are expected to be higher than in passive faults [38,39,40,41]. [42] stated in their studies that there may be undiscovered fault lines in measurement areas with high radon gas levels. Some studies interpret that there may be significant increases in radon gas concentrations on fault lines and the regions they pass through [43]. Changes observed in radon gas are not only due to seismic activity. However, they are also affected by radon emission from surrounding rocks and meteorological parameters such as rainfall, temperature, barometric pressure, and relative humidity [2, 21, 44]. Numerous studies have reported that radon changes in groundwater can be a predictor of earthquakes, and radon anomalies can be a useful tracer for seismic events [20, 21, 45]. The radon level in groundwater varies depending on the path the water takes underground, that is, what type of rock and soil it comes into contact with on this road. Because radon gas penetrates the water from these rocks and soils and changes the level of radon in these waters [5], for all these reasons, understanding the time and place of earthquake formation with an observed radon anomaly is still one of the important problems [46]. Studies have indicated that more than 80 percent of the anomalies in radon gas observed before earthquakes in groundwater are due to increased radon concentration [47]. In some studies, a decrease in radon concentrations was observed [10, 20, 48, 49].

Several models explain the changes in radon gas before earthquakes and these mechanisms. One of them is the compression model. According to this model, compression occurs in the crust before an earthquake starts, so radon enriched in deep soil gas leaks up [50]. Another model, the dilatancy-diffusion model, takes into account the important changes that rocks go through before breaking [51]. According to this model, when tectonic stress increases in the Earth's crust, fractures occur with rock expansion. Then radon gas concentrated in the depths of the Earth's crust is pushed to the surface due to the compression that occurred in the crust before the earthquake. There is also the stress corrosion model [52] and the asperity model [14, 52]. Studies investigating the variations in radon concentrations observed in groundwater before earthquakes and their mechanisms are still in progress. [53,54,55,56] explained the anomalies before the earthquake’s radon concentrations using a volatilization model. [3] also modified this volation model for their work. They also explained the radon concentration behavior before the 1978 Izu-Oshima-Kinkai and 2011 Tohoku earthquakes with this modified volatilization model. [56] also calculated a linear correlation between radon minima and earthquake magnitude.

In the historical period, many destructive earthquakes occurred on the Akşehir-Simav Fault System (ASFS) in the study area. These earthquakes have caused loss of human life and damage. For example, two earthquakes of magnitude 5.9 Mw and 6.2 Mw occurred in ASFS in December 2000 and February 2002. ASFS can still produce destructive earthquakes due to its tectonic features. Due to the importance of ASFS, the deepest water wells in the districts and villages of Afyonkarahisar province, which are closest to this fault, were selected as the study area. It aimed to determine the radon concentrations in 6 selected deepwater wells and examine the relationship with earthquakes during the study period. The correlation between the magnitude of the earthquakes occurring in 24 months and the average radon concentrations measured in the well waters was investigated. In addition, this study will provide data on studies to determine possible radon gas concentrations that may be performed in ASFS or the region in the future.

Materials and methods

Geology of the study area

Turkey is under the influence of the Eurasian, African, and Arabian plates. The presence of many small plates between the large plates in its region causes a large part of Turkey to be located in an important earthquake zone [57]. It is located on one of the active tectonic sections of the Earth's crust, where frequent and large earthquakes occur in the Alpine-Himalayan Earthquake Belt (around the Mediterranean). According to recent studies in Turkey, 326 separate faults, fault zones, or combined systems are in active fault maps [58]. Four different neotectonic regions were suggested by [59, 60] across Turkey: (1) North Anatolian, (2) Eastern Anatolian contractional, (3) Central Anatolian planar, and (4) Western Anatolian extensional.

The study area is on the Akşehir Simav Fault System and includes the districts and villages of Afyonkarahisar province. Afyonkarahisar province is located in central-western Anatolia and seismically is in the first–second degree earthquake belt of Turkey. Afyonkarahisar and its surroundings are located in the middle part of the Akşehir-Simav Fault System (ASFS) [61], which is the most important tectonic element of the expanded Neotectonic zone of southwestern Turkey (Fig. 1). The ASFS is located in the area between Karaman in the southeast and Sındırgı district of Balıkesir province in the Northwest [62]. It is a normal fault system with an average width of 10–30 km, a length of 500 km, and an NW–SE-oriented, discontinuous oblique slip [62]. Since the ASFS is an active tectonic line, it has produced a series of historical and current earthquakes originating from the fault segments that form the system's Akşehir and Simav fault zones. Since 1921, 9 earthquakes over 4.5 Mw have occurred on the ASFS, whose magnitudes vary between 4.7 and 7.2 Mw. On 03.02.2002, an earthquake of Mw = 6.2 magnitude occurred in the Çay District of Afyonkarahisar province and created a surface fracture; this earthquake caused loss of life and property damage. Due to the tectonic characteristics of ASFS, there is a possibility of earthquakes up to 7 magnitude in the region.

Fig. 1
figure 1

a The study area and significant active fault systems of Turkey compiled from [61] b Geological map of Afyonkarahisar and its surroundings compiled from [63,64,65,66]

Full size image

Locations of wells are located within the Akarçay Basin. Akarçay Basin, which has approximately 7300 km2, is located within Afyonkarahisar and Konya provinces, mainly in Afyonkarahisar province. Rock communities are formed in different geological periods from the paleozoic to quaternary in Akarçay Basin. There are belts with different lithological and tectonic features in the region, including Afyonkarahisar and its surroundings [67] and [68] geologically divided Afyonkarahisar and its surroundings into three main units, including the study area. These are the Taurus Belt (Sultandağları segment), the inner Taurus Belt (north part of Afyon), and the northern part of the Sandıklı district. Rocks belonging to the Taurus and Inner Taurus Belt are observed in the study area. Afyonkarahisar and its surroundings have two basic bedrock communities separated tectonically from each other. One of these is the Taurus Paleozoic and Mesozoic units, which form the continuation of the Sultandağları belonging to the Taurus Belt. The other is the Afyon metamorphics, located in the inner Taurus Belt and covering large areas under the Neogene cover in the northeast, north, Northwest, and west of Afyonkarahisar. Paleozoic age metamorphic rocks are located in the Northwest and southeast of the basin. Metamorphic rocks that surface north of the Afyonkarahisar Plain in the Northwest appear at depths as they descend to the South. Metamorphic rocks in the southwest of the basin have a high topography in Sultandağları.

Sampling and measurements

Deepwater wells were selected from the villages and districts of Afyonkarahisar province. While choosing six water wells, attention was paid to ensuring that the wells were as close to ASFS as possible and were deepwater wells. Information and details of each well are shown in Table 1, and the locations of the wells in the ASFS are shown in (Fig. 2). Lines in Fig. 2 show the portion of the ASFS passing through the province of Afyonkarahisar. Water samples were collected in glass bottles with a leak-tight cap of 250 ml, with a special apparatus on the lid that prevents gas outflow. Before sampling each well, the water was flowed for about 5 min until clean water was obtained. The bottles were filled to the brim without any gaps, and the caps of the bottles were tightly closed under the flow of water to prevent radon gas escape. Water samples were taken at monthly periods for 24 months between 24 January 2014 and 22 December 2015. The lowest average radon concentration was observed in well 2 with 1.87 Bq/L, and the highest radon concentration was observed in well 6 with 11.93 Bq/L.

Table 1 Various measurement results of 6 wells
Full size table
Fig. 2
figure 2

Earthquakes (E1-E15) occurred during the study period and locations of 6 wells (1–6)

Full size image

Radon concentrations in well waters were measured with the AlphaGUARD PQ2000PRO detector [69,70,71]. This radon detector system can measure the radon concentration (Bq/m3) in air, water, soil, and building materials, as well as simultaneously three different meteorological parameters such as temperature (°C), atmospheric pressure (mbar), and humidity (% RH). The detector with a cylindrical pulse ionization chamber has a volume of 0.56 L and is an ionization chamber measuring alpha spectrometric technique. Special AquaKIT equipment is required for the AlphaGUARD PQ2000PRO detector to measure radon concentration in water. The detector is connected to this kit, which includes glass containers in the experimental setup via an air pump, as shown in Fig. 3. The manufacturer's "quick measurement procedure" was applied to measure radon concentrations in well water with the detector [70]. These glass bottles are the security vessel and the degassing vessel where bubbling occurs. The two glass bottles are connected to the AlphaGUARD PQ2000PRO detector and the AlphaPUMP, as shown in Fig. 3, together with the connecting tubes. AlphaPUMP is an electronic pump with three different pump speed settings: 0.03 L/min, 0.05 L/min, and 1 L/min. After all connections, 100 ml of water sample was slowly injected from the upper opening of the degassing vessel with a plastic syringe of the detector. The flow rate of the AlphaPUMP was set to 0.03 L/min to expel radon from the water for 10 min. AlphaPUMP was turned off after 10 min of pumping and kept off mode for 20 min without disconnecting any connections. Radon is expelled from the sample at this stage by AlphaPUMP in a closed circulation. After 20 min, AlphaGUARD's counting system procedure was started. The measurement data were transferred from AlphaGUARD to the PC with the help of the “DataEXPERT” system software. After the measurement of each water sample, the cleaning process was performed with the “active coal filter” to reduce the background level of the system to zero.

Fig. 3
figure 3

a Schematic of the experimental set-up b Schematic drawing of the ionization chamber of the Alphaguard detector

Full size image

DataEXPERT software was used to display, transfer and save the experimental data during the measurements with the detector. In addition to the radon concentration measurements in the well waters, simultaneously temperature (°C), atmospheric pressure (mbar), and humidity (% RH) parameters were also measured during the study period. Radon concentrations in the water wells at each well's measurement time were calculated using Eq. (1) [69, 70].

$${C}_{w}=\frac{{C}_{air}x\left(\frac{{V}_{system}-{V}_{sample}}{{V}_{sample}}+k\right)-{C}_{0}}{1000}$$
(1)

where Cw: Radon concentration in the water sample, Bq/L; Cair: Radon concentration in the measuring set-up after expelling the radon, Bq/m3; C0: Radon concentration in the measuring set-up before sampling, Bq/m3; Vsystem: The interior volume of the measurement set-up [1102] mL; Vsample: The volume of the water sample [100 mL]; and k: Radon partition coefficient, [k = 0,26].

Results and discussion

After radon measurements of 6 wells were completed with the detector, all radon measurements and simultaneously measured temperature (°C), atmospheric pressure (mbar), and humidity (% RH) values were displayed using DataEXPERT software [69, 70]. The water sample for each well was measured three times, and the results were averaged. All these measurements were made monthly over 24 months, and then these measurements were averaged. With these values and the necessary calculations, the measurement results given in Table 1 were obtained.

Radon concentrations in surface waters are generally lower than in groundwater because groundwater is in contact with emitted radon, and radon is exposed due to natural processes [72, 73]. The radon concentration values of the well waters in this study are not relatively high, as seen in Table 1. The results may have been most affected by the geological structure of Afyonkarahisar and its surroundings. Because the study area is in a geological structure with quaternary graben filling (Fig. 1). Quaternary layers as a source of radon emissions have not been adequately studied [74]. The highest radon concentration was measured as 113 Bq/L in the measurements carried out by [75] with the AlphaGUARD PQ2000PRO detector in the water wells of the quaternary deposits in Hashimiya, Iraq. In another study by [76] on 72 groundwater samples from the Norwegian quaternary, radon concentrations were < 500 Bq/L. Radon concentrations are generally lower in quaternary geological structures [77]. It is thought that the reason for the measurement of relatively low radon concentrations in this study may be since all of the wells are located in the quaternary geological structure.

According to earthquake data provided by Boğaziçi University Kandilli Observatory and Earthquake Research Institute Seismology Laboratory, earthquakes between 2.5 Mw and 3.4 Mw were recorded in the towns and villages of Afyonkarahisar during the study period. Earthquakes with Mw ≥ 2.5 were considered in the study [78]. The earthquakes that occurred during the monitoring period and their details are given in Table 2.

Table 2 The earthquakes that occurred during the study period
Full size table

Radon values with reasonable errors and mean radon concentrations of 6 well waters earthquakes that occurred during the monitoring period are shown in (Fig. 4). When Fig. 4 is examined, changes in the background level of radon concentrations can be seen. The value of ± 2σ was calculated and accepted as the threshold value/boundary condition and used to estimate radon anomalies (shown in striped spheres) [79]. Here, σ is the standard deviation. These changes can be briefly explained as “radon concentrations decrease from the background before earthquakes and then return to the background level and above,” as discussed in the studies conducted by [56] and [55]. For example, the 8.72 Bq/L radon concentration value of well no. 1 on 25.09.2014 decreased to 0.65 Bq/L on 23.10.2014. Afterward, the radon concentration returned to the background level of 6.15 Bq/L on 26.11.2014. In this measurement period, three earthquakes occurred in total: 2.9 Mw on 18.11.2014 and two 2.7 Mw earthquakes on 23.11.2014 on the same day. Similarly, well no. 1 decreased from the background level to a minimum value of 0.70 Bq/L on 04.08.2015 and then increased again. In this measurement period, two earthquakes, 2.6 Mw and 3.2 Mw, occurred on 22.09.2015, and 2.6 Mw earthquakes occurred on 28.10.2015. Similar anomalies in other wells (in striped spheres) are in Fig. 4. The radon concentration value of well no. 3 decreased from the background level and was measured on 29.12.2014 with a minimum value of 8.5 Bq/L. On 03.01.2015, an earthquake of 2.8 Mw occurred very close to well no. 3. Again, on 04.08.2015, from the background level of well no. 3, it decreased to a minimum value of 8.13 Bq/L. After this anomaly, two earthquakes, 2.6 Mw and 3.2 Mw, occurred on 22.09.2015. The radon concentration value measured at 10.00 Bq/L in well no. 4 on 25.09.2014 decreased to a minimum value of 0.94 Bq/L on 23.10.2014. After this anomaly, three earthquakes occurred, two with a magnitude of 2.9 Mw on 18.11.2014 and 2.7 Mw (same magnitude on the same day) on 23.11.2014. The radon concentration value of well no. 6 decreased from the background level to the minimum value of 9.73 Bq/L on 24.04.2014. After this anomaly, an earthquake of 2.7 Mw occurred on 03.05.2014. The radon concentration of well no. 6 decreased from the background level to its minimum value of 6.36 Bq/L on 31.03.2015. After this anomaly, a 2.8 Mw earthquake occurred on 26.04.2015, and a 2.6 Mw earthquake on 27.04.2015. As seen in Fig. 4, the radon concentration values of the wells first decreased in the range of + 2σ and -2σ and then rose again to the background level or above.

Fig. 4
figure 4

222Rn values with reasonable errors and mean radon concentrations of 6 well waters, earthquakes that occurred during the monitoring period [σ is the standard deviation, and striped spheres represent anomalies (predicted)]

Full size image

When the radon anomalies (predicted) shown with the spheres drawn in Fig. 4 and the earthquakes that occurred were examined, this relationship could not be observed in some wells. The radon concentration value of well no. 2 decreased from the background level to the minimum value of 0.43 Bq/L on 04.08.2015. Similarly, the radon concentration of well no. 4 decreased from the background level to the minimum value of 5.49 Bq/L on 25.06.2014. The radon concentration of well no. 5 decreased to a minimum with a value of 0.67 Bq/L on 01.07.2015 from the background level. No earthquakes occurred after the anomalies observed in these three wells on the mentioned dates. It is thought that anomalies observed in radon concentrations may be caused by changes in the strain area and stress in the fault line before earthquakes. Gas and fluid movement under the ground may have also been affected by the quaternary geological structure of the study area. In this study, radon anomalies (predicted) were observed in the pre-earthquake radon concentrations in the form of a decrease before the background level and an increase again to the background level or above (according to ± 2σ threshold value/boundary condition). However, these anomalies were not observed in every water well before every earthquake. This may be because of the lack of temporal data advantage when measuring radon with the monthly measurement period. The primary advantage of continuous radon measurements is their real-time radon concentration data [80]. [81] suggested in their study that periodic components in the radon time series should be taken into account before evaluating radon changes due to earthquakes, and the contributions of the periodic components of the time series should be eliminated. Other factors may have affected the concentration of radon in well waters. [82] explained that radon concentration in groundwater is not affected by the chemical composition of the water but rather by lithology, and rock-water interaction. Pre-earthquake seismic changes, geodynamics of the region [83] geology, and geological structure [56], can be affected by many complex factors, such as the depth difference of well waters in the same geological formation [84]. The expectation of a radon anomaly shortly before the earthquake and closer to the epicenter of the future earthquake is a temporally and spatially smooth pattern and is probably very simplistic. Because the relationship between radon anomalies and earthquakes seems much more complex in the natural process. However, studies on radon-earthquake may enable us to understand this relationship better.

[54,55,56] studied the anomalies in the radon concentration using a volatilization model. [56] observed decreases in radon concentrations in groundwater before the earthquakes in Taiwan in 2003, 2006, and 2008 and explained that recurrent radon anomalies are in a v-shaped progression. They explained radon anomalies occurrence in three stages before the earthquakes in research in Taiwan (2010). In phase 1, the radon concentration in groundwater is relatively constant. In the next phase, it switches to the gas phase, starts to decrease, and reaches a minimum. In phase 3, the radon concentration in groundwater increases and returns to the background level before the earthquake [56]. In the literature, radon concentrations decreased in about 20% of radon anomalies in groundwater before earthquakes [3]. [56] explain the situations in which the radon concentration decreases from the normal level using a volatilization model in their studies. The variation of the radon concentration measurement results measured in the well waters in Fig. 4 with the earthquakes that occurred for two years shows some similarities with the studies conducted by [56] and [55]. The details of these earthquakes can be seen better in Table 2. This way, radon concentrations decrease from the background level to the minimum value and then increase again. Figure 4 shows the changes and the earthquakes associated with these changes when the radon concentrations decrease from the background level to the minimum value and increase again. Such anomalies of radon concentrations in Fig. 4 (striped spheres represent anomalies) are slightly similar to the v-shaped progression observed by [56] and [55] but not as large as they observed.

[56] also examined the effect of meteorological factors such as atmospheric temperature, barometric pressure, and precipitation on the changes in radon gas. However they concluded that the environmental factors mentioned would be insufficient to explain such a significant radon decline because they stated that these meteorological parameters are periodic in the seasons. In their study on the Izu Peninsula, they explained both the positive anomalies in radon concentration observed in groundwater before the 2011 Tohoku earthquake and the negative anomalies before the 1977 Izu-Oshima-Kinkai earthquake with a modified volation model [3].

During this study period, the results of the parameters measured simultaneously with the radon concentration data in the well waters were compared one by one. No significant relationship was observed between radon concentration and air temperature, atmospheric pressure, and humidity. The relationships between radon gas and meteorological parameters have been investigated in many studies. Some studies have explained that radon gas in soil [85] and groundwater [20, 43] is affected by meteorological parameters. However, in some studies, it was emphasized that no correlation (or a low correlation) was observed [4, 86,87,88,89]. [90] calculated a weak correlation between radon gas in the soil, and meteorological parameters, and [88] calculated a weak correlation between radon gas and temperature and humidity. Similarly, observed a weak correlation between radon gas and air temperature and air pressure; [86] observed a weak correlation between radon gas and air temperature and precipitation. [87], in their study in the Klong Marui fault zone, explained that meteorological parameters have a non-significant effect on radon gas. According to the study results conducted by [91] in the groundwaters of the Yangsan Fault System, they explained that the radon anomaly observed in the groundwaters was not caused by a seasonal effect but by pre-earthquake effects. [89] in the groundwaters of the Yangsan Fault System, they explained that the radon anomaly observed in the groundwaters was not caused by a seasonal effect but by pre-earthquake effects. [89] used mathematical methods to determine radon anomalies in groundwater before earthquakes in the Yunnan region of Southwest China. In the results of their study, [89] stated that there was no correlation or weak correlation between radon concentration and meteorological factors. This result of the study conducted by [89] is similar to this study conducted at ASFS. In this study, it is thought that the reason why no relationship could be observed between radon concentration and air temperature, atmospheric pressure, and humidity may be related to the fact that water wells are deep wells. [50] stated that the radon gas in the soil measured at a depth of 100 cm (or deeper than 100 cm) was not affected by meteorological parameters (especially air temperature and pressure). The wells measured in this study are deep water wells, the details of which are shown in Table 1.

Radon minima- EQ Magnitude (Mw) graphs in the well waters in the study area were obtained, as shown in Fig. 5. To see the correlation better, the radon minima precursory to the earthquakes were plotted as a function of earthquake magnitude. The best-fitting straight lines are obtained in Eq. (2):

$${C}_{w}=a+b{M}_{w}$$
(2)

where Cw is the radon minimum in groundwater observed in relating well during an anomalies decline, Bq/L; Mw is the earthquake magnitude in the Richter scale; b is the slope of the line, and a is the intercept value of the well. As seen in (Fig. 5), high R2 values for six wells indicate a good correlation between radon minima and earthquake magnitude.

Fig. 5
figure 5

Relationship between radon minima and EQ Magnitude (Mw)

Full size image

[56] calculated the correlation between radon minima and earthquake magnitude as 0.946 (regression coefficient: R2) in their study in eastern Taiwan. The correlation graphs obtained between radon minima and earthquake magnitude in this study are shown in Fig. 5. The correlation coefficients (R2) values of the wells from 1 to 6 were obtained as 0.93, 0.83, 0.82, 0.97, 0.87, and 0.85, respectively. These high squared regression coefficient (i.e., R2 = 0.97) values for each well shows that the two parameters are well correlated. These values and the equations in (Fig. 5) may be a relationship that can help estimate the earthquake magnitude near the ASFS of the radon minima observed in all wells. This study showed similar behaviors to the results obtained by [56]. Anomalous decreases in radon concentrations were observed before earthquakes at all measurement points on ASFS. Similar results were obtained with the research conducted by [56] in terms of correlations between radon minima and earthquake magnitude. However, the v-shaped pattern stated by [56] could not be observed clearly in this study. As can be seen, when Fig. 4 is examined, there is a slight resemblance to a v-shaped pattern. Before an earthquake, radon concentrations decreased from the background level and then increased to either the background level or above. Anomalies that can be associated with earthquakes have been observed. Still, it is thought that a much more extended study period and data are needed to examine the v-shaped progression better. For example, it is believed that if larger earthquakes had occurred during the study period, perhaps this pattern would have been observed much better at that time. In addition, although the study area was selected in ASFS and a seismically active region, the quaternary geological structure of the study area is thought to have highly affected radon concentrations.

In this study, similar results were obtained with the research conducted by [49] at ASFS in Afyonkarahisar. [49] got a good correlation between radon minima and the earthquake magnitude in 4 wells before earthquakes ranging from 2.6 to 3.9 Mw. However, in the measurements performed by [49] on the same fault line, a v-shaped pattern in well water radon anomalies could not be observed clearly as in this study. It is thought that the reason why the v-shaped model could not be observed clearly in this study may be because large earthquakes did not occur during the study period. As shown in Table 2, which shows the earthquakes that occurred during the period in which the measurements were made, the largest earthquake was 3.4 Mw. [49] explained in their study that they could not clearly observe the v-shaped model, just like in this study. In the study of [49], the largest earthquake that occurred during the measurement period was 3.9 Mw. Whereas [56] compared radon concentrations with earthquakes of 5.4, 6.1, and 6.8 Mw in their study in Taiwan. Similarly, as can be seen from the studies of [53,54,55], they observed the v-shaped pattern in radon anomalies with earthquakes of more than 6 Mw in general. In this study, no earthquake larger than 3.4 Mw occurred 24 months. Although v-shaped progression in radon anomalies was not clearly observed in ASFS, both in this study and in the study conducted by [49], correlations between radon minima and earthquake magnitude were obtained very well in both studies. In both studies, similar results were obtained regarding the relationship between radon minima and EQ Magnitude (Mw) as in the research by [56]. In other words, the precursory minimum observed in radon concentration in groundwater in both studies decreased as the magnitude of the earthquake increased.

Conclusions

The ASFS, where the study area is located, is an important fault zone since it has produced devastating earthquakes since the 1970s. Devastating earthquakes with magnitudes between 6.0 and 7.1 Mw occurred in the region. This study was carried out monthly in 6 different deep water wells for 24 months between 24 January 2014 and 22 December 2015. Although especially deep water wells were selected in the study area in the ASFS, the radon concentration results were determined in the range of 0.11–19.84 Bq/L. Findings supporting studies of relatively non-high radon concentrations obtained in quaternary geological areas were obtained. A clear relationship could not be observed between radon concentrations in well waters and simultaneously measured air temperature, atmospheric pressure, and humidity. [53, 54, 56] studied the mechanisms and decreases in radon concentrations in groundwater before the earthquakes and conducted their study using a volatilization model. The v-shaped progression of recurrent radon anomalies observed in their studies could not be clearly observed in this study. However, radon concentrations in all wells decreased from the background level to the minimum value before the earthquakes and then increased again to the background level or above. Although the shape of the v-shaped progression cannot be observed clearly, anomalies that can be associated with earthquakes are observed in radon concentrations by accepting the ± 2σ value as the threshold value/boundary condition. However, these anomalies were not observed in every water well before every earthquake. Before a few earthquakes, no radon anomalies occurred in any of the wells. After the radon anomalies were observed in water wells 2 and 5, no earthquakes occurred. In this study, radon anomalies of ± 2σ value according to threshold value/boundary condition in deep water wells 1, 3,4, and 6 before earthquakes with codes E4, E6, E7, E8, E9, E11, E12, E13, E14, whose details are shown in Table 2 observed. Investigating the relationship of anomalies in radon concentrations with earthquakes with continuous measurements and time series observation may provide a better understanding of seismic-induced radon anomalies.

The precursor minimum observed in the radon concentration decreased as the earthquake’s magnitude increased. The R2 (regression coefficient) values of the wells from 1 to 6 were obtained as 0.93, 0.83, 0.82, 0.97, 0.87 and 0.85, respectively. Although the magnitude of the earthquakes in the 24-month study period was from 2.5 to 3.4 Mw, good correlations (the high R2 values) were obtained between radon minima and earthquake magnitude in 6 wells. The best correlation between the precursory radon minima and earthquake magnitude was obtained in well 4 (R2 = 0.97). In addition, it is recommended to measure and monitor well 4 in a possible future study. Equations between radon minima and earthquake magnitudes (Mw) may be a relationship that can help estimate the earthquake magnitude near the ASFS of the radon minima observed in all wells. It is thought that long-term and continuous radon concentration measurements in the study area may be beneficial due to the high R2 values (between radon minima and earthquake magnitude) of the six wells and the tectonic characteristics of the ASFS. There is a possibility that ASFS can generate earthquakes of up to 7 Mw in the region. If larger earthquakes (greater than 5 or 6 Mw) occur in the ASFS, then the observed relationship between radon minima and earthquake magnitudes (Mw) in this study can be further improved in the future. The results obtained from this study will contribute data to future radon gas studies that can be done at ASFS or in the region.