Natural radioactivity in European drinking water: A review

Abstract More than three quarters of the dose from ionizing radiation is caused by natural radiation sources. In most European countries, groundwater is used as a source of drinking water, what can be an additional source of radiation since naturally occurring radionuclides in the bedrock can dissolve into groundwater. Although the Directive 2013/51/EURATOM establishes several parametric and screening values for different radioactivity parameters, there is a lack of harmonized information and publications available that evaluate the overall condition of drinking water in EU. Thereby, it is challenging to uniformly assess the current situation in European countries in regards of effective doses caused by the consumption of drinking water. This paper aims to collect the available information on the main radionuclides of interest in drinking waters of different European countries and create an overview of the existing situation. Graphical abstract


Introduction
More than three quarters of the dose from ionizing radiation is caused by natural radiation sources (CSN, 2015). Naturally occurring radionuclides can be found in air, water, soil and living organisms. Some rocks and soil emit radiation because they contain uranium and its daughter nuclides, thorium and its daughter nuclides and natural potassium, the so-called primordial radionuclides (Su arez et al., 2000). But the radionuclide content is not the same for all soils, for example granitic and shale formations show the highest radioactivity. In most European countries groundwater is used as a source of drinking water. Groundwater is in contact with soil and rock and it is able to dissolve the radionuclides generated in the decay series of uranium and thorium from the surrounding media. As a result, radionuclides can be integrated into the chemical composition of groundwater in concentrations that may exceed the standards required by legislation. Different naturally occurring radionuclides can be present in drinking water sources. Studies concerning radioactivity in drinking water are mostly focused on the determination of gaseous Rn-222 and long-lived isotopes from Th-232 and U-238 series, probably because of the chemical and radiological toxicity of uranium and the radiological hazard that radium isotopes Ra-226 and Ra-228 may pose. In specific cases, long-lived Rn-222 daughter products Pb-210 and Po-210 may also be significant dose contributors (Ek et al., 2008;Vesterbacka, 2007) because of their high dose coefficients (International Commission on Radiological Protection (ICRP), 2012). Thorium isotopes are not usually a critical concern in natural water (Jia et al., 2009) as the solubility of thorium in water is very low, although Th-232 and Th-230 analyses in surface water and groundwater are often considered when uranium or thorium mining areas are investigated (Carvalho et al., 2009;Carvalho et al., 2016b).
Legislative frameworks consider the intake of radionuclides by drinking water as a consumption of a commodity in a planned exposure situation (International Commission on Radiological Protection (ICRP), 1999(ICRP), , 2007 rather than as a source of natural background radiation, regardless of the origin of radionuclides in drinking water supply. The International Commission on Radiological Protection (ICRP) and World Health Organization (WHO) suggest using a dose constraint of 0.10 mSv/yr to restrict individual doses in planned exposure situations (ICRP, 2007), what also applies for drinking water consumption (World Health Organization [WHO, 2017]), since a risk caused by a dose equal to 0.10 mSv/yr can be considered low enough not to give rise to any adverse health effects (World Health Organization [WHO, 2017]).
In general, 0.10 mSv/yr can be considered a modest dose compared to the global average committed effective dose of 3.0 mSv/yr (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2008). As a worldwide average, dose from ingestion (both drinking water and food) accounts for less than 10% of the overall effective dose À 0.29 mSv/yr. Still, typical values may range from 0.2 to 1 mSv/yr (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2008).
The European Commission (EC) has adopted the dose constraint of 0.10 mSv/yr as a parametric value for the indicative dose (ID) from drinking water consumption in its Directive 2013/51/ Euratom. ID is defined as the committed effective dose for one year of ingestion resulting from all the radionuclides (from both natural and artificial origin) whose presence has been detected in a supply of water intended for human consumption, but excluding tritium, potassium-40, radon and shortlived radon decay products (European Parliament, 2013). Exceeding the parametric value should not be taken as a sign that the water is unsafe to drink but as a trigger for further investigation. In case of noncompliance with the parametric value, each member state shall assess whether it poses a risk to human health which requires or not remediation actions (European Parliament, 2013).
Besides ID, gross alpha and beta particle measurements, which are respectively related to alpha and beta radiation, are a tool suggested by the WHO and EC for the initial screening of water samples. In most cases, a specific analysis of radionuclides concentration would be only needed if gross alpha or beta screening levels are exceeded; otherwise ID is believed to remain below the parametric value of 0.10 mSv/yr. WHO suggests a screening level of 0.5 Bq/l for gross alpha activity (GAA) and 1.0 Bq/l for gross beta activity (GBA) (World Health Organization [WHO, 2017]), whereas the EC recommends 0.1 Bq/l and 1.0 Bq/l, respectively (European Parliament, 2013).
In the European Union, a parametric value of 100 Bq/l is also set for radon activity concentration -(European Parliament, 2013). In cases when the Rn-222 concentration exceeds 1,000 Bq/l, remedial actions are considered to be justified on radiological protection grounds (European Commission [EC], 2001;European Parliament, 2013). These requirements are applicable for water supplies used for commercial or public activity. For an individual water supply, it is recommended to use the value 1,000 Bq/l as a consideration for remedial action.
Despite the existing legislation and the available information about natural radioactivity in drinking water scattered in governmental and scientific reports and papers, no review article on such matter has been published so far. Thereby it is difficult to uniformly assess the current situation in European countries in regards of effective doses caused by the consumption of drinking water and to estimate if radioactivity in drinking water is a widespread issue to be concerned about.
The current paper aims to collect the available information about natural radionuclides in drinking water. The article is fed by two different sources of information. Firstly, peer-reviewed scientific publications and national monitoring studies reporting the radiological parameters of drinking water in 23 European countries (almost all of them belonging to the EU, besides the United Kingdom). Altogether, more than 100 publications and reports were analyzed. Some data from untreated groundwater and mineral and bottled water have been also included to enrich the discussion. Secondly, information directly provided by different public and private bodies are responsible for water management or related to radioactivity control in each country. Namely, this information was obtained from about 20 public bodies and 250 private and public operators of drinking water treatment plants (DWTPs) from 12 countries. The countries tracked and the sources of information consulted in each country during this work are depicted in Figure 1. These 12 countries were selected according to their high radionuclide concentration in their surface soils (REM, 2018), a phenomenon that may be potentially related to radionuclides presence in their water. Further information about the specific data and the way they were compiled can be found elsewhere (https://www.lifealchemia.eu/en/home/). Thanks to this mapping campaign, did not only was more data obtained but also a critical awareness was acquired about the efforts that each country devotes to the radiological characterization of its water and the amount of information freely accessible to citizens about this issue in different places of the European continent.
2. State of radiological quality of drinking water in European countries 2.1. ID, GAA and GBA Different levels of natural radioactivity were found in drinking water sources from the 23 countries checked. All obtained data are compiled in Tables S1, S2 and S3 in the Supplementary Material. ID, GAA and GBA parameters as well as radon concentration provide the most relevant information about the radiological characteristics of drinking water. The following sections classify the checked countries according to ID, GAA and GBA values of their water, although Figure  2 shows an overview of the radiological quality of the water consumed in the 23 countries analysed. Radon concentration is discussed in a later section.

Countries where the parametric ID is exceeded
The compiled data revealed that the parametric value for ID was exceeded for a part of the population in Estonia, France, Portugal and Sweden.
In Estonia, it has been estimated that approximately 18% of inhabitants (230,000 people) consume drinking water with ID higher than 0.10 mSv/yr (Forte et al., 2010). This problem is mostly limited to the north part of the country where public water supply relies on the Cambrian-Vendian aquifer system (Fig. 3), a source of water with high radionuclide content as it lies on uranium and thorium rich crystalline basement rocks.
High ID in Cambrian-Vendian water is caused by radium isotopes Ra-226 and Ra-228, whereas the contribution of other radionuclides (U isotopes, Po-210, Pb-210, Rn-222) remains marginal (Suursoo, 2017). ID typically exceeds the parametric value 2-3 fold (Kiisk et al., 2011), and may even reach 1 mSv/yr in rare locations. Radium concentrations in other aquifer systems are much lower, although the parametric value for ID may sometimes be exceeded in Ordovician-Cambrian aquifer system. In recent years a government body has made public some reports (https://www.terviseamet.ee/et/keskkonnatervis/ettevotjale/joogivee-ja-loodusliku-mineraalvee-kaitlemine/joogivee-kvaliteedi-ulevaated) that support this conclusion. For example, a 2019 report collected data concerning 1,309 DWTPs located throughout the 15 Estonian counties. Harju county, where Tallinn is located, was the most thoroughly mapped with more than 230 samples analyzed. According to the results of this report, 90 DWTPs throughout the country exceeded the parametric value allowed for ID. This implied 6.9% of DWTPs producing water that did not meet the standards value for natural radioactivity in Europe. 75% out of the 90 DWTPs with radioactivity problems in their water were located in the north counties of L€ a€ ane-Viru and Ida-Viru, where groundwater mostly comes from Cambrian-Vendian aquifer. It is worthy to note that this information is made completely accessible for citizens by Estonian governmental bodies despite showing that ID parameter is above EC recommendations in several locations.
Since 2005, health checks of the radiological quality of water intended for human consumption are mandatory in France. The monitoring of natural radioactivity is overseen by the Regional Health Agencies, supported by the French Institute of Radioprotection and Nuclear Security (French acronym IRSN) and the Authority on Nuclear Security. The radiological quality of drinking water in any French location can be freely consulted (https://solidarites-sante.gouv.fr/sante-etenvironnement/eaux/eau). Based on 2011 national monitoring results (French Institute of Radioprotection and Nuclear Security [IRSN], 2014), 10 to 20% of analyzed water samples required specific radionuclide determination because they exceeded the GAA and/or GBA reference levels. The most commonly measured radionuclides were Ra-226, U-238 and U-234, and more rarely, Po-210, Pb-210 and Ra-228. Based on the report, less than 5% of analyzed samples exceeded a 0.1 mSv/yr ID. Results from a tap water radiological quality monitoring campaign conducted between 2008 and 2009 (Caamaño et al., 2011) revealed that 99.83% of population received water that complied with the ID limit of 0.1 mSv/yr at all times. Data collected within the framework of this work are consistent with these previous reports as they show that, among the approximately 16,500 French DWTPs, only 3% of them produced water with GAA exceeding 0.1 Bq/l and/or GBA exceeding 1.0 Bq/l at any time between 2016 and 2017, and only 0.2% DWTPs produced water exceeding the ID parametric value, most of them small facilities with a production capacity lower than 1,000,000 m 3 /yr.
In Portugal, a 2006 nationwide drinking water screening was done using the gross alpha and beta measurement approach (Lopes et al., 2006). GBA values were all below 1.0 Bq/l and varied in a range from 0.018 to 0.457 Bq/l. GAA was above 0.1 Bq/l in 18% of samples. These samples came from the center and north regions of the country where the bedrock, mainly granite, is enriched in natural radioactivity. High levels of naturally occurring radionuclides were also found in old uranium mining areas (Carvalho et al., 2009;. Concentrations of uranium isotopes, Ra-226, and Po-210 in surface water and wells were often high enough to invalidate that water as drinking supply (Carvalho et al., 2009). Yet, measurements from 2016 showed that the ID in consuming water from surface sources in an old mining area in Viseu district was below 0.10 mSv/yr (Carvalho et al., 2016a). Radionuclide specific measurements from other regions is not common, but the operators of DWTPs throughout the country periodically report information about GAA, GBA and ID values of the drinking water that they produce. All this information can be freely accessed by the population. By this way, data of radiological quality of the water produced by more than 260 DWTPs and consumed in more than 240 Portuguese cities and villages between 2016 and 2018 were collected in the framework of this work. Big DWTPs supplying large cities as Lisbon and Porto treated surface water from rivers as Tajo, Duero or C avado Rivers and their radiological levels were low. 24 DWTPs were found that produced water with GAA and/or GBA levels that exceeded the limit value, and 6 among them produced water that exceeded the allowed ID value. This number may be larger because in a few locations the ID was not analyzed although GAA and/or GBA exceeded the limit value. Thereby, at least 2.3% of the checked Portuguese DWTPs did not fulfill the radioactivity requirements. All those DWTPs were fed by groundwater and were located in areas with uranium and thorium enriched soils (REM, 2018).
The Swedish Radiation Safety Authority (SSM) is in charge of assuring a safe radiation environment for Swedish citizens. Between 2003 and 2005, this body published three reports (Falk et al., 2004;€ Ostergren et al., 2003;€ Ostergren et al., 2005) containing massive information about the radiological quality of Swedish drinking water. In general, the radiation dose from uranium and other radionuclides is low in tap water from Swedish public water supplies. About 50% of waterworks use surface water and 50% comes from groundwater produced by artificial infiltration in soil aquifers. According to the reports published by SSM in 2003SSM in , 2004SSM in and 2005 and the collected data from Swedish water suppliers, 18% of the DWTPs exceeded the GAA limit value and about 6% exceeded the GBA limit value, but water exceeding the ID limit value was detected only in two municipalities, Falk€ oping and Torsås (0.6% from the collected data of the DWTPs not fulfilling ID requirements). As opposed to treated tap water, ID values as high as 5 mSv/yr have been reported for drilled wells (Ek, et al., 2008). The main dose contributor was U-238: uranium concentrations were above the guideline value of 15 lg/l in 17% of wells (Ek et al., 2008) and 2% of the wells have uranium concentration above 100 lg/l. (Ek et al., 2008). High values of radium isotopes are rare, although a study of 328 drilled wells in € Osterg€ otland and Kalmar counties (southern Sweden) reported several cases where ID was higher than 0.10 mSv/yr due to the occurrence of Ra-226 and Ra-228 (Salih et al., 2002).

Countries where the parametric GAA and/or GBA is exceeded
Besides the confirmed high ID data reported for Estonia, France, Portugal and Sweden, eight additional countries were found during this screening with high radioactivity levels in their water: Finland, Germany, Spain, Italy, Poland, Denmark, Belgium and Bulgaria. In most of them, GAA and/or GBA exceeded the parametric levels in some locations and the existing data suggest that the parametric value for ID may be exceeded, but this statement has not been confirmed by any published data. Other countries like Finland and Denmark were also included in this group since high radioactivity levels were detected in individual private sources of drinking water such as wells and boreholes, hence a minor fraction of population may be affected by this issue as well.
The Radiation and Nuclear Safety Authority of Finland has studied radioactivity in drinking water produced from surface or groundwater sources since 1970. The reports published by this body together with scientific papers are the two main sources of information about this issue in Finland, since operators of DWTPs essentially report no data about the radiological quality of the water that they produce. High radium activity concentrations in drinking water are rare. In Finland, about 90% of population consumes drinking water from public water supply and the radionuclide activity concentrations for such water has mostly remained low; for example, reported mean values for Rn-222, Ra-226, and U-238 were 27, 0.003 and 0.015 Bq/l, respectively (M€ akel€ ainen et al., 2001). However, numerous studies , Vesterbacka et al., 2006, 2007Salonen, 1994;Juntunen, 1991) point that radionuclides concentration in drilled (in the bedrock) private wells can be significantly higher. For example, a study concluded that raw water from drilled wells may lead to IDs above the parametric value with a mean of 0.17 mSv/yr (Salonen, 1994). A more recent study (Vesterbacka et al., 2006) about drinking water from private wells stated that 95% of results were below 0.10 mSv/yr; that is to say, 5% of Finnish private wells produce water with ID above that value. The mean annual effective dose from radionuclides has been estimated to be 0.4 mSv for drilled well users, four times above the parametric value, and 0.05 mSv for users of wells dug in soil (Vesterbacka, 2005). At least, for the users of drilled wells over 90% of the dose is caused by Rn-222 and from the radiation protection point of view, the most important nuclides in terms of effective dose are  In Germany, 75% of public water supply originates from groundwater. The most comprehensive study about the issue treated herein was carried out between 2003 and 2007 (Beyermann et al., 2009;Beyermann et al., 2010). It consisted of a nationwide groundwater study from 564 locations; thereby the majority of the country was under the survey. It concluded an average 0.0046 mSv/yr ID, with a range of values from below 0.002 to 0.108 mSv/yr, which is slightly above the 0.1 mSv/yr limit. The average ID increased to 0.086 mSv/yr and the maximum doses up to 1.140 mSv/yr if radon and its decay products were included in the calculation. When the data were analyzed more in depth, it was revealed that 184 out of 1,045 samples had a GAA value above 0.1 Bq/l. Those numbers lead to a remarkable percentage of 17.6% of samples not meeting the GAA requirements, although it seemed that this high GAA values did not lead to exceed the ID parametric value most of times. The study noted that high concentrations of natural radionuclides were found in some regions of eastern and southern Germany, what agreed with the regions containing the highest uranium and radium content in German soils (REM, 2018). Additional data were reported by Birke et al., (2010) in a study that focused on the uranium content on mineral bottled water and tap water. Only one bottle exceeded the baseline value of 15 lg/l stated by the WHO for uranium among 908 bottled water samples and tap water from 163 municipal systems. It is noteworthy that the concentration of this specific radionuclide in tap water is usually freely reported by the operators of DWTPs of each German municipality.
In Spain, the governmental body Drinking Water National System of Information (Spanish acronym SINAC) is responsible for collecting data about drinking water quality from private or public bodies which manage the supply of water. According to its 2019 technical report on the sanitary quality of water for human consumption (Palau et al., 2020), ID, GAA and GBA were monitored in 11.8, 19.65 and 9.0% of all water supply areas in the country. 3,431 ID, 8,469 GAA and 4,053 GBA analyses were carried out during this monitoring campaign and the percentages of these parameters above the reference levels were 0.12%, 7.9%, and 0.01%, respectively. It would be interesting to know if the geographical distribution of the samples exceeding the limits agree with those areas in Spain with a high content of natural radionuclides in their soils (west and north-west parts of Spain above all, and in a minor extent other areas in the south-east and in the east) where, furthermore, small populations usually lacking of DWTPs are located ( Alvaro et al., 2007). However, this information is not reported by the Spanish governmental body. Private and public operators of DWTPs do not usually make the radiological characterization of drinking water freely accessible either. According to the results of a mapping campaign carried out in the framework of this work from November 2017 to July 2019, the radioactivity parameters mostly measured by Spanish operators of DWTP were GAA and GBA, and tritium and radon were the most measured radionuclides (https://www.lifealchemia.eu/en/home/). All DWTPs that provided data met the requirements established by Directive 2013/51/EURATOM. In addition, essentially no bibliographic reference addresses the radiological study of Spanish tap water. On the other hand, several nationwide studies about bottled drinking water have been reported. For example, Palomo et al. (2007) stated that 16% of the samples they analyzed were over the screening level for GAA (Palomo, et al., 2007), whereas P erez-Moreno et al. (2020) found that 20% among their group of analyzed samples exceeded that value and only one sample exceeded an ID of 0.1mSv/yr. In Italy, the Territorial Environmental Protection Agencies publicly reports GAA and GBA levels in water from about 700 locations. 7.1% of GAA and 9% of GBA analyses exceeded the screening levels among the data and reports checked herein (https://www.lifealchemia.eu/en/ home/). These results implied that radionuclides concentration should be measured to calculate ID but this parameter was not available for these samples. Thereby, it cannot be categorically affirmed that no tap water in Italy exceeds the ID limit. Most of the analyses exceeding the GAA screening level were found in Piemonte and Lombardia, although some problems with GAA and radon were also identified in the southern region of Calabria. Some scientific papers are focused on drinking water from underground sources in some areas like Lombardia area, Milano and Central Italy (Borio et al., 2005;Desideri et al., 2007;Forte et al., 2007;, and they state that the parametric value of ID was never exceeded in those areas. Mineral water was also analyzed elsewhere (Jia et al., 2009; and the parametric values were either exceeded. In Poland, the available data suggest that uranium isotopes are not of radiological concern either in underground and surface water, (Pietrzak-Flis et al., 2005;Sekudewicz & Gąsiorowski, 2019) nor in bottled mineral water (Skwarzec et al., 2003). However, groundwater in Sudetes and Carpathians Mountains in southern Poland contains high concentrations of Ra-226 (up to 1.77 and 0.62 Bq/L, respectively) (Przylibski et al., 2004;Przylibski et al., 2014). High Rn-222 and Ra-226 contents in this shallow circulation groundwater are caused by high Ra-226 concentration in reservoir rocks and favorable geological conditions resulting from the presence of weathered rocks, faults and deep fissures with high emanation coefficients (Przylibski et al., 2014). Measurements on Ra-228 activity carried out in Upper Silesia, the Poland region that the Sudetes and the Carpathians cross indicate detectable but moderate concentrations of this radionuclide, up to 0.065 Bq/l (Chmielewska et al., 2014). The ID calculated from the radionuclide measurements available in bibliography for Upper Silesia does not exceed 0.10 mSv/yr and either do the scarce ID data about water treated by DWTPs in the zone. However, as a result of the large presence of natural radionuclides in groundwater, accumulation of radium isotopes in water treatment plants has been detected (Chmielewska et al., 2014), so that drinking water produced from this source may contain high radioactivity levels if it is not suitably treated.
Most of Danish drinking water (about 99%) is obtained from groundwater. An early study (Ulbak & Klinder, 1984) indicated that, due to certain geological formations in the country, Ra-226 concentrations in raw or treated water could be high, up to 0.550 Bq/l. However, a national study (Nielsen, 2006) conducted by Danish Environmental Protection Agency involving 296 water suppliers throughout the whole country demonstrated that the values for U-238 remained below 0.22 Bq/l. The few high concentrations that exceeded the total alpha screening level were caused due to the presence of uranium in water from individual boreholes. Radiation doses from consumption of water at these uranium levels are estimated to be well below a 0.1 mSv/yr ID and thus the risk of finding drinking water in Denmark with unacceptable concentrations of radioactivity is very small. Namely, this report by the Environmental Protection Agency in 2006 (Nielsen, 2006) includes the water quality produced by 262 DWTPs from 5 regions and just one sample from the DWTP Egedal in Ebeltoft was the only one that exceeded the GBA screening level, whereas 5% of 262 DWTPs exceeded the screening level for GAA (0.1 Bq/l). Consequently, although most of Danish drinking water is obtained from groundwater, according to these results the risk for high concentrations of radioactivity in drinking water is low.
The high GAA values found in other countries may also lead to ID above 0.1 mSv/yr. For example, in Belgium, nationwide studies have reported some cases where GAA in drinking water was as high as 0.42 Bq/l (Sombr e & Bouchonville, 2008) or 0.23 Bq/l (Sombr e et al., 2018). The Federal Agency for Nuclear Control (FANC) is the national responsible body for monitoring radionuclides in Belgian drinking water. There are hundreds of sampling points in Belgium for monitoring its water and the control of radioactivity covers GAA, GBA, Ra-226 and K-40, as well as tritium H-3. Based on the monitoring results (e.g. Sombr e & Bouchonville, 2008; Sombr e et al., 2018) the ID never reaches the parametric value of 0.1 mSv/yr. In recent years, FANC also regularly obtains data on natural radioactivity in groundwater from different aquifers in Belgium to develop a detailed cartography of natural radioactivity in groundwater at long term. Researchrelated work done on natural radioactivity in Belgian drinking water is scarce. One of the few works is reported by Vasile et al. (2016), focused on tap and bottled water. Various radionuclides, such as radium and uranium isotopes as well as Pb-210 and Po-210 were included. The results indicated that the activity concentration of Ra-226 and Ra-228 remained below 0.15 Bq/l and 0.1 Bq/l, respectively. The work concludes that the water consumed in Belgium is safe from a radiological point of view.
Publications on measurements of natural radionuclides in drinking water in Bulgaria are limited, although there are a few long-term studies (up to 10 years), which provide a representative overview of the situation (Kamenova-Totzeva et al., 2003;. A drinking water study in south-western Bulgaria recorded 0.177 Bq/l as the mean value of GAA (Kamenova-Totzeva et al., 2015). Although this data almost doubles the screening level suggested by the EC, additional studies suggest that Bulgarian water does not imply a radiological hazard for consumers. Namely, the National Center of Radiobiology and Radiation Protection in Bulgaria, undertakes regular radiological monitoring of drinking water mostly on that same southwestern part of Bulgaria, where the main radionuclides of interest have been U-238 and Ra-226. The activity concentrations of these radionuclides have remained low and the effective doses received from drinking water consumption due to these isotopes remained below 0.1 mSv/yr. It has also been stated that Ra-226 constituted 90% of the ID, thus being the main radionuclide of interest.

Countries not exceeding ID, GAA and GBA
The information reported by scientific papers concerning other 11 European countries (Austria, Croatia, Cyprus, Czech Republic, Greece, Hungary, Ireland, Luxembourg, Romania, Slovenia, and United Kingdom) was also analysed and summarized in the Supplementary Material. The amount of available information varies depending on each country but, as a general rule, all data point that the screening levels suggested by the WHO and the EC are not exceeded even for GAA and GBA. Thereby it can be deduced that an ID of 0.1 mSv/yr will not be exceeded in any case for the countries commented in this section.
Consistent research on natural radionuclides in groundwater and drinking water in Austria is available since the end of 80's (Sch€ onhofer 1989). Initial efforts in the field were devoted to measure Rn-222 in water. Friedmann (Friedmann 2006(Friedmann , 2008 studied Rn-222 in spring and groundwater in Austria, looking back to the beginning of the 20th century as well. From the 1990s and 2000s more studies have also involved other radionuclides, mainly Ra-226 and Ra-228, but also Pb-210, Po-210, U-238 and anthropogenic radionuclides (Gruber et al., 2009;Kralik et al., 2003;Landstetter & Katzlberger, 2009;Wallner et al., 2002Wallner et al., , 2007Wallner et al., , 2009. In general, the published work indicates that Ra-226 and Ra-228 mainly account for the effective dose for adults through drinking water in Austria, although a possible high impact of Po-210 and Pb-210 has been also suggested (Gruber et al., 2009). The activity concentrations of Ra-226 and Ra-228 have consistently remained below 0.2 Bq/l and 0.1 Bq/l, respectively. In summary, the involved studies have mostly indicated that the ID received by adults from water consumption in Austria remains below 0.1 mSv/yr.
In Croatia, monitoring on radionuclides including Ra-226, has been included into the national radioactivity monitoring programme, what includes tap water collection in Croatia's largest cities. The earliest results of radionuclides in different drinking water samples can be tracked back to the 90's (Marovi c 1994;, where Ra-226 was measured in well, mineral and tap waters. Later studies in the 2000s (Bronzovic et al., 2005, Bronzovic et al., 2007Ro zmari c et al., 2012Ro zmari c et al., , 2014 have expanded the research, also involving Ra-228 and U-238. All studies show that Ra-226 and Ra-228 values in well and tap water have remained below 0.06 Bq/l and 0.01 Bq/l, respectively. Thus, the ID value for these waters can be considered as well below 0.1 mSv/yr. Cypriot drinking water usually comes from desalinated water and as a consequence of its maritime origin; its uranium concentration levels are generally low (Philippou & Pashalidis, 2017).
In Czech Republic, about 60% of drinking water is taken from surface sources and about 40% from groundwater sources (Neznal et al., 2014;EU NORM). In 2015, it was assessed that the content of all radionuclides present in drinking water is responsible for an average effective dose of 0.007 mSv/yr (National radiation protection institute [NRPI], 2015). Studies (Otahal et al., 2014;Cerny, 2017) conducted on private wells have indicated that 16% and 0.3% of the samples exceeded the reference limits set by the Council Directive No. 2013/51/Euratom for Rn-222 and U-238, respectively. In general, Rn-222 values in private wells are higher than in other EU countries. The available data has otherwise shown that Ra-226 and U-238 values remain dominantly below reference levels and recommendations by WHO (World Health Organization [WHO, 2017]).
In Greece, radioactivity in drinking water has been analyzed in the region of Attika where three lakes are used as water source. Namely, a sampling network has been established in order to monitor the radioactivity levels from the Attika network and from the results it is possible to conclude that the committed effective dose for one year of consumption is well below the value of 0.1 mSv/yr. (Kehagia et al., 2007). Radionuclides  were also measured in bottled water and the total effective dose is much lower than the dose reference level of 0.1 mSv/yr (Karamanis et al., 2007).
In Hungary, the highest uranium concentrations were usually found in spring waters, whereas the lowest were found in tap waters (Jobb agy et al., 2009). The values do not exceed the parametric values in Euratom Drinking Water Directive in any case. Therapeutic spa, mineral spring and bottled spring waters were also analyzed and none of them exceeded the parametric value (Jobb agy et al., 2010). On the other hand, Kov acs et al. (2004) stated that the upper levels of results in some bottled mineral water may cause an annual effective dose greater than 0.1 mSv/yr. They conclude that 12-17 years old youngsters is the most endangered age group because of the intensive bone growth and their drinking habit of soft drinks made from mineral water.
Radiological Protection Institute of Ireland (RPII) published reports on radioactivity in both bottled  and drinking waters (Dowdall et al., 2013). For bottled waters, all 21 bottled water samples complied with WHO recommendations and Euratom Drinking Water Directive. For drinking water, RPII provides the summary of a national survey of radioactivity carried out together with the Environmental Protection Agency for compliance with the parameters set out in the Euratom Drinking Water Directive. It is noted that approximately 18% of drinking water comes from groundwater and springs and the reminder from surface water. Results show that all groundwater sources were found to be suitable for human consumption and no action to reduce radioactivity is necessary.
Luxembourg approximately gets 2/3 of its drinking water from groundwater, mainly from the Liassic sandstone aquifer, called Luxembourg Sandstone aquifer. The rest is produced from surface water at Esch-sur-Sûre Lake. The occurrence of natural radioactivity in groundwater (springs) is thoroughly studied (Tosheva et al., 2010). 316 springs were sampled, most of them in the outcropping region of the Luxembourg Sandstone in the central part of the country. Activity concentrations of naturally occurring radionuclides are detectable, but they are low and do not pose any significant risks from the viewpoint of radiation protection (Tosheva et al., 2009).
In Romania, occurrence of U-238 and Th-232 series radionuclides has mostly been investigated in mineral water, where the concentrations of Ra-226 and U-238 may go up to 0.45 Bq/l and 0.21 Bq/l, respectively (Ion et al., 2019). However, as the daily intake of mineral water is lower than for regular drinking water, the ID estimations are well below 0.10 mSv/yr. GAA and GBA have been used for drinking water screening (Pintilie et al., 2016a(Pintilie et al., , 2016b. All the results remained below the screening values and the highest values were recorded in drilled wells. Dose assessment based on GAA and GBA measurements estimated an ID lower than 0.0576 mSv/yr for adults. In Slovenia, the limited available information on drinking water and mineral water analyses reveal low concentrations of Ra-226, Ra-228, and U-isotopes leading to an ID as low as 0.004-0.007 mSv/yr (Benedik & Jeran, 2012). High Ra-226 concentrations, up to 0.614 Bq/l, were measured in thermal and mineral springs in 1979 by Kobal et al. (1979). In 1990, Ra-226 activity concentration as high as 0.510 Bq/l was also measured in a groundwater from a borehole (Kobal et al., 1990).
Doses from drinking water are generally very low in the United Kingdom, as an average 0.008 mSv/yr is reported (RIFE-23, 2018). Private groundwater sources have higher risk of elevated uranium and radium concentrations. Uranium concentration in groundwater exceeds 15 mg/l in up to 2% of cases. By contrast, 73% of samples from a groundwater survey by Smedley et al. (2006) had concentrations <1 mg/l. The largest ranges and highest concentrations of uranium are found in groundwater from red-bed sandstone aquifers such as the Old Red Sandstone and PermoTriassic Sandstone; one relatively high uranium concentration (6.6 mg/l) was also found in a groundwater from the Scottish Torridonian Sandstone. Occasional high uranium concentrations were also found in groundwater from a number of other aquifers, though none exceeded 15 mg/l. Studies in Britain showed that measured groundwater uranium concentrations can vary by up to five orders of magnitude within a given aquifer (Smedley et al., 2006).
It can be also stated that high radionuclide concentrations are often found in mineral waters. For example, ID values estimated for Estonian mineral water bodies may go up to 2 mSv/yr (Estonian Health Board, 2014). In Poland, unbottled mineral water from Kłodzko valley showed a maximum ID of 0.298 mSv/yr (Walencik-Łata et al., 2016). A nationwide study on bottled mineral water in Hungary concluded that most of the results were below 0.10 mSv/yr, yet the maximum value was 0.61 mSv/yr (Kov acs, 2004).
Ra-226 is mostly included in ID calculations but it is interesting to note that scarce information is provided on Ra-228 activity concentrations, although its dose coefficient is more than two times higher than the one for Ra-226 (ICRP, 2012). In fact, Ra-228 is not often included in the measurements even if its mother nuclide Th-232 is analyzed (Carvalho et al., 2009;. Thorium has low solubility in water and is characterized by poor environmental mobility (Cook et al., 2003), while radium is soluble in anoxic acidic conditions. Therefore the absence of mother nuclide in water does not indicate the absence of the daughter nuclide. Concentrations of Ra-228 are more likely to be controlled by the occurrence of thorium in the aquifer rock. Thus, the exceedance of the ID could be more widespread if the analysis of Ra-228 would be also done together with Ra-226. This may not be revealed by the GBA screening method, because GBA is not the best screening tool to estimate the effective dose caused by the beta emitter Ra-228 due to its high dose coefficient. If Ra-228 would be the only beta emitting radionuclide in a water sample, ID of 0.1 mSv/year would be caused by Ra-228 activity concentration of only 0.2 Bq/l. This is five times lower than the GBA screening level suggested by EC and WHO (Suursoo, 2019).
Several authors have pointed out that even if ID for the adults is lower than 0.10 mSv/yr, doses for children may still exceed the parametric value (Kov acs, 2004). The infant age group (younger than 1 year) is especially at risk with doses up 5 times higher than adults (Salih et al., 2002).

Radon
In countries where drinking water is obtained from deep groundwater sources, such as wells and boreholes, the likelihood of high radon concentrations is normally higher compared to surface water from reservoirs, rivers or lakes.
In Portugal, Rn-222 concentrations as high as 3,856 Bq/l have been found in the groundwater of Nisa region (Pereira et al., 2015).
Radon concentration was measured in seven of 16,500 French DWTPs were ID was above 0.1 Bq/l. It exceeded 100 Bq/l in all cases and even exceeded the 1000 Bq/l limit value in two plants.
In Germany, according to a nationwide groundwater study (Beyermann et al., 2010), 74 out of 1,045 locations had a radon concentration above 100 Bq/l, which means that 7.1% of samples did not meet the radon concentration standards.
In Italy, the Territorial Environmental Protection Agencies publicly report radon concentration from about 400 locations and the percentage of samples that exceeded a radon concentration of 100 Bq/l was 1.1%.
An early study (Ulbak & Klinder, 1984) from Denmark indicated that Rn-222 concentration in raw and treated water could be high, up to 1,070 Bq/l for Rn-222. On the other hand, a national study (Nielsen, 2006) conducted by Danish Environmental Protection Agency demonstrated that the values for Rn-222 remained below 0.036 Bq/l. This report by the Environmental Protection Agency in 2006 (Nielsen, 2006) includes the water quality produced by 262 DWTPs from 5 regions but no sample was observed to exceed the parametric value for radon, although this parameter was only measured in 36 samples.
Radon measurements from drinking water have been also made in Cyprus (Louizi et al., 2004), where the upper levels of the concentration range from 0.4-150 Bq/l, sometimes exceeding the parametric values for radon in the Euratom Drinking Water Directive 2013. Rn-222 in water issue has been addressed very thoroughly in Romania. Rn-222 concentrations above 100 Bq/l have been found in spring and well water in Transylvania, in regions where the soil contains granitic formations: Hateg region and the Codru-Moma Mountains (Cosma et al., 2008). However, the mean Rn-222 concentration in Romanian drinking water is only 9.8 Bq/l (Burghele et al., 2019) and the parametric value of 100 Bq/l is only exceeded in 0.7% of cases.
Most drinking water supplies in United Kingdom have very low levels of radon (Gibbons & Kalin, 1997;Henshaw et al., 1993). Elevated radon levels can occur in private water supplies that come from groundwater sources. E.g., high levels of radon in some water supplies can be found in south west England. Radon concentrations in surface water on the granite moors of Devon and Cornwall can be well in excess of 100 Bq/l in some places, and the radon levels in water from boreholes and wells are generally higher than in surface water (Bowring & Banks, 1995).
According to extended data in Supplementary Material, in case of tap water the median values are below 100 Bq/l, and higher values can be seen for well waters, with the highest values in drilled bedrock waters. It can be noted that radon concentration in water is most likely to be high in drinking water for private consumers who obtain water from personal drilled or dug wells without any prior purification. This is evident in the studies carried out in Finland, where high values have been reported for such type of well waters. In general, the majority of the Rn-222 median values in the studied papers remains below 100 Bq/l. High values (over 100 Bq/l) have been determined in several cases, but these make up a small percentage from the whole dataset and belong dominantly to drilled or dug well waters.
To date, the available epidemiological studies have not shown an association between consumption of drinking water and increased risk of stomach cancer due to the presence of Rn-222 (World Health Organization [WHO, 2017]). Based on a report by the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR), on average, 90% of the dose due to Rn-222 is caused by inhalation rather than ingestion. It has been estimated (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2000) that an Rn-222 concentration of 1,000 Bq/l in drinking water will increase the indoor air radon concentration by 100 Bq/m 3 . Swedish survey on private wells revealed that in case of high radon concentrations, Pb-210 and Po-210 might contribute to a higher radiation dose than radon itself (Ek et al., 2008). ID can be 3-4 times higher due to Pb-210 and Po-210. The same study concluded that only 33% of the private wells in Sweden have a Rn-222 concentration lower than 100 Bq/l, while radon concentration exceeds 1,000 Bq/l in 8% of private wells. Radon concentration in Swedish municipal water supplies is in general low, with a median radon concentration of 10 Bq/l, except in few cases where groundwater is used.

Conclusions
Drinking water can be a source of natural radioactivity for humans, but public awareness about this issue is hardly widespread. This paper reviews two types of information to compile the existing information about the topic and describe the situation in Europe related with naturally occurring radionuclides in drinking water. On one hand, bibliographic information from more than 100 scientific papers about the characterization of different types of drinking water (untreated, treated and mineral bottled water) is compiled and summarized. On the other hand, radiological data provided by different public and private bodies that manage the purification and supplying of drinking water from 12 European countries was collected to complete this work and is summarized herein.
In general, the exposure to natural radioactivity through the consumption of drinking water in Europe is mostly within the legal limits. Drinking water coming from surface sources such as rivers and lakes carries negligible levels of natural radioactivity. Groundwater is more likely to contain naturally occurring radionuclides, namely when this water is in contact with basement rocks containing natural radionuclides such as uranium and thorium. This phenomenon makes drinking water to occasionally exceed the recommended radioactivity levels in localized centers of population throughout the continent.
The amount of information freely available for citizens strongly varies depending on the country. The bibliographic review leads to the conclusions that mineral bottled water and, above all, well water, are the types of drinking water with the highest levels of natural radioactivity. This later case is especially remarkable for water coming from drilled wells that is consumed without any purification treatment, what may lead to higher exposure to ionizing radiation among the consumers who get their water from private wells. Consequently, increasing public awareness about this issue in order to avoid potentially increased health risks is necessary.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
The authors gratefully acknowledge support of this work by the LIFE programme from the European Commission through the agreement LIFE16 ENV/ES/000437-LIFE ALCHEMIA project.