Citizen measurement of ambient dose rate: challenges and opportunities

1 Introduction

Citizen-led measurements (CM) of radioactivity appeared in Belarus in the late 1990′ in the form of ambient dose rate measurements as a mean for local population to evaluate the radiological quality of their environment and to develop specific radioprotection actions in response to their questionings about daily living conditions (Hériard-Dubreuil et al., 1999; Lepicard and Heriard-Dubreuil, 2001). These early actions led to the development of the concept of practical radiation protection culture, which increased the confidence in daily living condition and decreased adverse health and psychosocial consequences of nuclear accidents (Hériard-Dubreuil, 2006; Lochard, 2007). Citizen ambient dose rate measurements received a renewed interest following the Tohoku catastrophe and the resulting Fukushima Daiichi Nuclear accident (Ando, 2015), mainly due to a loss of confidence in national authorities and a concomitant wish to understand the prevailing situation (Lemarchand, 2016; Berti-Suman, 2021). These initiatives in Japan were supported as soon as 2012 by the international network Safecast (www.safecast.org). Thanks to the increasing availability of internet and social medias, radioactivity measurement results were widely shared, especially through interactive mapping, supporting adversarial debates and dialogue between citizen and authorities (Brown et al., 2016; Kenens et al., 2020). However, the objectives of these CMs remained the same, i.e., allowing everyone to evaluate their own risks and implement self-protective actions, in addition to making measurement data available in real time. ¹

These initiatives attracted interest of European countries engaged in research actions to develop emergency preparedness and response (EPR&R) policies through several European projects such as SHAMISEN (Ohba et al., 2021). More specifically, citizen measurement of ambient dose rate was recognized as a key lever to help people to implement self-protective actions (Maître et al., 2021). The importance of transfer, continuity and sustainability of these actions was also identified as essential (Maître et al., 2021). As a result, and based on these lessons, the international commission for radiological protection (ICRP) recommended to support citizen initiatives of measurements as a mean to foster dialogue between population and authorities (ICRP, 2020), especially through a co-expertise process, a concept developed in the aftermath of the Fukushima accident (Lochard et al., 2019; Lochard et al., 2020; Lochard et al., 2026).

Several citizen networks for radioactivity measurement appeared in Europe since 2015, some of these being created in response to a societal demand, driven by both an interest in environmental quality check especially around nuclear plants and the rapidly evolving situation in Eastern Europe with the Russian war in Ukraine. These networks are CzechRad (lauched in 2015) supported by the PIANOFORTE European Project (https://www.zhavamista.cz/czechrad) CITISTRA (https://pianoforte-partnership.eu/citistra/), OpenRadiation in France (Bottollier-Depois et al., 2017; Bottollier-Depois et al., 2019), FIfF e.V./TDRM in Germany (https://tdrm.fiff.de/index.php/en/), OpenRed in Spain (https://ibercivis.es/en/openred-eng/), and more recently, Saveecobot in Ukraine (https://www.saveecobot.com/en). All these European projects present some common features and difficulties that were highlighted and discussed in a recent workshop (Bertho et al., 2026), and path toward sustainability and useability of CM data are proposed. Two key issues were identified. The first one is the difficulties linked to the sustainability of the current citizen networks. The second one is the useability of CM database for scientific or expertise studies. An overview of current European citizen-led radioactivity measurement is proposed in this review.

2 How does it work?

Whatever the network, the basic principle remains the same: to provide citizen with detectors, to help them to use the detectors, and to provide means for publishing and sharing their results. This definition of citizen-led radioactivity measurement networks translates into a system with three components (Figure 1A): A detector, a communication tool eventually monitored by the citizen and an interactive map for publishing measurement results. What makes the difference between networks is the complexity of the detector proposed. Broadly speaking, there are two main families of detectors. The first family is autonomous detectors which incorporate all the functionalities needed: GPS chip, internal clock, communication module through GSM networks, temperature sensor in addition to the electronic needed for managing the detector functioning. Some of these models may also include solar panels so that they are autonomous in energy, thus useable as fixed, autonomous beacons. This is the case of the detectors developed by Safecast and adapted for the SaveEcoBot network in Ukraine, mainly for users’ safety reasons. The second family of detectors is the simplest one, containing the detector, the associated electronic, a battery and a Bluetooth communication module. This family of detectors is then controlled by a smartphone application through Bluetooth. These detectors are much cheaper, since the metadata are collected from the smartphone, thus avoiding costly modules inside the detector. Moreover, this family of detectors, thanks to the use of the smartphone, allows to collect more metadata, such as descriptive fields, pictures or literal description of the environment at the place of measurement (Figure 1B). As a whole, many metadata are associated with each measurement result, providing a large array of information (Figure 1C).

Another essential point is that all these networks are working according to the European citizen science association (ECSA) ten principles of citizen science (Haklay et al., 2020). Among others, a fundamental principle is the openness of citizen science, which translate into free access to data, but also to source codes of applications and web sites, and to electronic diagrams of detectors, so that anyone can reproduce the whole system, or can get the whole or only a part of the database to make specific analyses. The sole limitation is the citation of sources, under the CC-BY licence requirements. This principle of openness is essential to maintain trust from citizen engaging in these networks.

Beside these networks which uses specific detectors of ionizing radiation, one must note that there are some smartphone applications which uses the CMOS camera of the smartphone to detect and measure ionizing radiations. This is the cheapest system, since this is an all-in-one system. Such applications ask for obturating the camera, so that the camera chip is in fully dark mode. This allows the detection of ionizing radiations that light up pixels with excess charge (Odenwald, 2019). However, the sensitivity and the specificity of these CMOS cameras to ionizing radiation is low, especially in the low dose rate range, i.e., less than 0.5 µSv.h^-1. Moreover, a specific calibration curve is needed for each CMOS camera type, since ionizing radiation detection is highly dependent on the size and nature of the CMOS cameras. As a whole, it is not recommended to use these applications especially in areas with a usual background (Odenwald, 2019; IRSN, 2019b; Johary et al., 2021). Nevertheless, such applications, with specific noise removing function and calibration, may help detecting cosmic rays, especially during commercial flights (Takano et al., 2025).

3 Common features and challenges

The idea for the Openradiation project was conceived in 2013 in the wake of the Fukushima disaster and was officially launched in 2017. The aim was to develop a simple, low-cost system for measuring ambient equivalent dose rate that would allow for widespread data sharing. Furthermore, the system had to be suitable for primarily educational use in normal circumstances, whilst being robust and reliable enough to be used in a radiological emergency. From the outset, therefore, there was a clearly identified challenge regarding the long-term sustainability of the project. This was one of the reasons behind the creation of the OpenRadiation consortium, which is responsible for steering the initial project and managing the current network. This consortium brings together Planète sciences, an NGO dedicated to science education for children, IFFO-RME, an NGO specialising in continuous education on major risks, the National Association of Local Information Committees and Commissions (ANCCLI), which gather the local information committees existing in France around each major nuclear installation, the Sorbonne University FabLab, dedicated to new educational methods through the creation of connected devices and the French authority for nuclear safety and radiation protection (ASNR), which provides scientific, technical and financial support to the project. The formation of this consortium clearly focuses activities on educational initiatives and risk awareness-raising actions targeting various audiences, including school and university students, residents within the CLI’s 20 km perimeter of action, and individuals interested in the scientific approach to measurement. The situation is similar for the other European networks and other initiatives elsewhere (Takano et al., 2025), except for the SaveEcoBot network, which was created to provide an environmental radiation monitoring network in the context of the Russian war in Ukraine, responding both to the questioning of the Ukrainian population and as an early warning system in case of nuclear emergency in a safe way for contributors. One have to note that SaveEcoBot is the first citizen network officially accredited for radiation environmental monitoring in the world (https://safecast.org/safecast-radnote-solar-powered-realtime-radiation-sensors-certified-for-official-use-in-ukraine/), underlining the recognition of the potential usefulness of citizen-led measurement networks.

3.1 Detection technology and reliability of measurements results

The principles of citizen science ask for reliable measurement results and rigorous protocols to be used by citizen, so that the results can be used in scientific studies, as underlined by the ECSA (Haklay et al., 2020). Moreover, citizen embarked into such measurement activities must be confident into the measurement results they produce in relation to the potential health issues. For ambient dose rate measurement, there is also a need for reliable, robust, low-cost and easy-to-use detectors, financially accessible to citizen. Most of the existing networks use a Geiger-Muller (GM) technology, although with some variations in the geometry of the detector. This technology is robust and reliable, provided that GM detectors are tested before use. For instance, OpenRadiation network select only GM tubes that fall within a tolerance range of approximately 10% of a reference dose rate. A recent campaign showed that the 366 selected tubes provided a response in the range of ± 6.2% of the reference value and that the rejection rate was 86%. However, other type of detectors, using for instance CsI(Tl) or NaI scintillator crystal, can be used, especially with an objective of spectrometry analysis. This is for instance the case of the FIfF e.V./TDRM network, aiming at registering spectrometric profiles to identify radioactive releases from nuclear power plants (NPP). In all cases, the European networks are associated with expert institutes allowing to test and characterize the detector response.

Very few intercomparison studies with citizen detectors have been published. An intercomparison was made in 2019 (IRSN, 2019a) in an area with locally high concentrations of naturally occurring radionuclides (mainly potassium-40, thorium and uranium), resulting in elevated ambient dose rates. The site was first mapped using a high precision NaI detector of 2 × 2 litres in an all-terrain vehicle at one meter above the ground. This detector has a range of energy detection from 25 keV to 3 MeV, a linear response from 1 nSv.h^-1–5 µSv.h^-1 for H(10) with a continuous acquisition mode. On the basis of the mapping made with this embarked detector, 6 reference points were chosen (Figure 2). These points were measured at one metre above the ground using a ionisation chamber (Model RSS131, Reuter Strokes) with an energy range of 40keV-3 MeV and a linear response over a range of 10 nSv.h^-1 – 1Sv.h^-1 in H(10). 15 acquisitions were made at each point with 1 min duration. The resulting mean was used as a reference for all detectors tested in the field.

Several commercially available professional detectors and citizen detectors were then tested at these reference points. For each citizen detector, 7 to 20 measurements were made and were compared to measurements made with two professional gas detectors (GD1, n = 15–40 measurements per point and GD2 n = 50–80 measurements per point). The results (Figure 3), limited to the comparison of professional gas detectors with citizen GM detectors for consistency, demonstrated that measurement results were comparable at five out of six field points. At field point 2, where was the highest dose rate, a large variability of measurement results was observed. A hypothesis to explain this dispersion was the heating of detectors at temperatures above 50 °C due to the rising sun, which induced aberrant responses of citizen detectors.

Another study was conducted in Japan over the period 2019–2023 to compare the Kyoto University radiation mapping (KURAMA) system and the Safecast citizen mapping system (Yasmint et al., 2026). This study first showed that the geographical coverage of the KURAMA is larger than the coverage of the Safecast network, but the Safecast network was able to catch local situations (residential areas, public facilities) that Kurama does not monitor. Results also showed systematic bias in uncorrected Safecast data, although the regression analysis showed a strong linear correlation between the two datasets (Yasmint et al., 2026). This study supports the concept that citizen measurement data might be useful if integrated into environmental monitoring and policymaking (Berti-Suman (2021).

Overall, if these field studies show a good agreement in measurement results between professional and citizen detectors, there is a need for other studies, made in laboratory conditions and on a larger range of dose rates. In fact, the dose rates encountered in a post-accidental situation are much higher than the ones encountered with naturally occurring radionuclides, thus questioning the linearity of the response of citizen detectors at dose rates in the range of 1–10 µSv.h^-1. This range of dose rates is frequently encountered in Japan in areas close to the Fukushima-Daiichi NPP, even 15 years after the accident. Moreover, such studies should also delineate citizen detector response in the low-dose range (>0.15 µSv.h-1, the most likely encountered dose rate range), reproducibility of measurements and stability of response according to time. Although all citizen networks are confident with the quality of the measurement results published, the availability of such studies is essential for the credibility of measurement results and their integration into processes of environmental monitoring and policymaking.

Reliability of reported data is also linked to the measurement uncertainty, as defined by the number of counts registered for each measurement. This measurement uncertainty is managed in two different ways according to the type of detector but results in any case in a measurement uncertainty less than 15%. To reach this level of uncertainty, a minimal number of 50 counts has to be detected. Safecast detectors, deployed mainly in Europe by the SaveEcoBot and networks, is using a large gas detector in a “pancake” format, and measurement results are integrated for one minute. The large size of the detector allows to detect more than 50 counts in one minute, except for measurements in very low background areas. OpenRadiation choose a different strategy by preventing the measurement from stopping and being recorded until at least 50 counts have been recorded. Moreover, OpenRadiation recommend to end-users to accumulate as many counts as possible, depending on the circumstances, to improve the accuracy of the results. Citizen measurement results are thus reliable, from the metrological point of view, although calibration and sensitivity studies are not available. Nevertheless, other sources of uncertainties exist that may limit the usefulness of citizen measurement data.

3.2 Reliability of measurement reports

To be comparable, measurement results should be done using recommended protocols and reported with consistent descriptions of the environment. For instance, the OpenRadiation network recommend taking measurements at 1 m above the ground, the usually recommended heigh for ambient dose rate measurements. This recommendation is made commonly by all measurement networks and is generally well-followed by keeping the detector at waist height. Some contributors developed spontaneously wood-made detector holders so that the measurement heigh was well controlled (Figure 4) (Ando, 2015). This demonstrates that citizens are able to develop skills and practical tools to ensure that recommended measurement protocols are followed.

As well, it is recommended that users describe the environment of the measurement. This is made through the collection of metadata associated with the measurement. For the OpenRadiation system, supplemental metadata are collected manually, before registering the measurement (Figure 1B) and are mainly made for a better description of the environmental conditions prevailing at the place of measurement.

However, the fundamental principle of citizen science is that people are free to measure what they want, where they want, when they want and as they want. As a result, people are governed by their own primary interests (Kenens et al., 2020; Berti-Suman, 2021). In line with this, a study made in Belarus in 2018 with OpenRadiation detectors took advantage of the GPS location and timeline of published measurements to understand the way people are organising their measurements (Bertho et al., 2019). The study consisted in providing radiation detectors to a group of 17 pupils (8 girls and 9 boys) for a one-month period, to measure ambient dose rate in their village. Each pupil got the detector for a duration of 3–5 days and was free to organize its own measurement. A total of 645 measurements were made in 1 month (Figure 5A) with dose rate results mostly in the range of 0.05–0.18 µSv.h^-1, similar to observed ambient dose rates elsewhere in the world. Three hot spots were also identified, in each case associated with fireplaces using local, contaminated wood. More importantly, it was possible to delineate the way pupils organize themselves. A first observation was that most of pupils made their first measurement at home, although some participants made their first measurement outside, as indicated by the metadata and confirmed by interviews (Figure 5B). This highlights the importance of the home as a safe place. A second observation is that each participant chooses the places to be measured in a different way. One participant makes a tight grid covering his closest neighbourhood with some supplemental places of interest in his daily life (Figure 5C, participant 13), while another one makes a walk around the settlement (Figure 5C, participant 8). Such individual choices induce a variability in the measurement places, which represent a source of uncertainty. Other sources of uncertainties are also linked to the heigh of the measurement, the weather, variations in background values, or the measurement of radioactive objects such as street furniture made of naturally occurring radioactive materials (marbles, granite, etc.), naturally radioactive rocks in the environment or radioactive sources used for calibration, pedagogic uses or other purposes, including wish to publish controversial data.

These uncertainties highlight the need for a moderation of the published results. As with any citizen science initiative, and regardless of the quality of the data provided, any publication of data requires critical scrutiny, particularly in the case of data measuring exposure to radioactivity, which, when published and widely shared, undeniably takes on a political dimension (Carmes, 2020; Kenens et al., 2022). Here again, each network uses its own system for assessing the quality of published data, either beforehand or afterwards. Data acquired using Safecast detectors (as the Citistra network does), because they are stored in the detector’s memory, must be pre-processed before they can be published on a map via the internet. This approach allows the quality of the data to be checked before publication, through the work of moderators. Ultimately, this helps prevent the publication of erroneous data. Conversely, this a priori moderation system may not suit users who wish to retain control over the measurement and publication process.

Unlike the Citistra network, OpenRadiation has opted for a post-publication review of data, entrusting responsibility for the quality of published information to each contributor. Also, OpenRadiation network refrains from removing any results, to prioritise maintaining contributors’ trust. Nevertheless, this requires the implementation of a post-publication management process for results of an atypical nature. To this end and following a preliminary study conducted in 2020 (G. Darley, personal communication), a verification threshold has been established: any measurement exceeding 0.3 µSv.h^-1 is labelled ‘atypical’ to alert users viewing the map and automatically generates a message send to the moderators. The published measurement is then processed to attempt understanding the origin of the high level of radiation exposure reported, and a comment is added to the measurement to explain the most plausible reason for the value and the origin of the published result. OpenRadiation has therefore opted for an educational approach to explain the origin of unusual values, particularly in the case of measurements taken near a radioactive object (minerals enriched with natural radionuclides or heritage objects).

Moreover, this way of moderation encourages contributors to acquire a critical look on the results published on the map. Visitors on the map spontaneously look for “red dots”, these measurement data which are unusually high. This helps foster a culture of radiation protection among contributors whilst making them aware of the importance of the quality and representativeness of the published measurements and associated data.

3.3 Absence of radiological issue: difficulties in maintaining interest

The main difficulty for networks other that SaveEcoBot roots in this orientation towards pedagogic activities and risk awareness-raising actions, which are not sufficient to maintain interest of the citizen, nor to obtain financial support on the long term. Indeed, one of the intended functions of these networks is to provide real-time field data in the event of a radiological or post-accident emergency situations - which, fortunately, are rare. It is therefore necessary, with a view to long-term sustainability, to develop other activities. Moreover, an analysis of users’ contribution to published data on the OpenRadiation map shows that a very limited number of users provides the majority of measurement results and conversely, the vast majority of users publish results occasionally or even makes only one measurement (Figure 6A). This picture is very frequently observed in citizen science, whatever the topics is (Sauermann and Franzoni, 2015). This behaviour is encouraged by the observation of dose rate values that vary in a short range, between 0.05 and 0.15 µSv.h^-1 in most cases (Figure 6B). Conversely, this encourages the emergence of contributors who seek out hotspots. It is thus necessary to find out other activities aiming at both maintaining interests of current contributors and increasing interest from other potential users.

Two kinds of activities could be developed. The first kind seek to increase the number of contributors (i.e., to increase the community of users), through activities linked to the interests of users. This can be developed based on the numerous survey and interview studies conducted recently aiming at defining the motivations of contributors (Land-Zandstra et al., 2021). Most of these studies underline the importance of the feeling to belong to a community, which is supported by communication and exchanges through different media (website, newsletters, reports on specific actions, etc.) and actions (challenges, in-person meetings, etc.) (Rüfenacht et al., 2021). Nevertheless, motivations to participate to a project can be classified as intrinsic factors and extrinsic factors (West and Pateman, 2016; Perko et al., personal communication). Intrinsic factors are linked to personal interest for environment, health issues, participation into a scientific project, while extrinsic factors rely on recognition, reputation, etc. In any case, the participation of users to specific actions has to be recognized and acknowledged, whatever the nature of the action is.

The second kind of activities is the use of the database by external users, which, when communicated to the contributors, could represent an important motivation through the participation to scientific projects. For that, it is first necessary to delineate the useability of the databases generated by contributors, especially in the view of uncertainties identified in chapters 3.1 and 3.2.

3.4 Usefulness and useability of CM of radioactivity.

The current OpenRadiation database collected more than 1.6 million measurement results, with approximately 100,000 collected per year, mostly in Europe. The Safecast network collected more than 250 million measurements worldwide, with a particular focus on Japan. These databases thus contain a huge amount of data that could be used for specific purposes, such as environmental monitoring, identification of high-background areas and provides datasets for local studies. These CM data could also be useful in decision-making process during a radiological emergency of for the management of the mong-term phase, as proposed by ICRP (ICRP, 2020). In fact, CM data are able to capture local situations that are not detected by official monitoring networks, mainly because of the different territorial grids (Yasmint et al., 2026). For instance, the French ambient dose rate monitoring network, the Teleray network, concentrates its beacons around nuclear power plants with less dense coverage in between. By contrast, CM data are mainly driven by personal interest, leading to a higher density of measurements in living areas such as cities and villages. Some examples of such application of citizen measurements in post accidental situation are sparse but highly demonstrative of the usefulness of CM of radioactivity (Ando, 2015; Bertho et al., 2019; Ando et al., 2026). However, to be useable, CM data need to be first analysed for their consistency and coherence.

To evaluate the quality of CM data, a study was conducted on a sample dataset from the OpenRadiation database corresponding to the calendar period between April 1st and 31 December 2024. This period was chosen because of the high proportion of alert messages received during this period, with measurement results above the 0.3 µSv.h^-1 threshold (Bertho et al., 2025).

During this period of April to December 2024, 75,505 measurement results were published on the website, of which 30,613 measurement results generated an alert (40.5%). These 30,613 alerts were first analysed using the measurement mode, i.e., “ground level” or “in plane”. Results (Figure 7) showed that 29,393 of alerts (96.01%) were made on board of a plane, which well-explain the high ambient dose rates, up to 6.5 µSv.h^-1 of this group of measurements. This was also confirmed by the altitude of these measurements systematically over 3,000 m. However, this also shows that most of the alerts were due to mistakes in the use of the measurement mode. By using such metadata analysis, it was possible to identify several reasons for these high ambient doses rates published during this period among the 1,220 remaining alerts. 9 of these measurements were made at ground level but in mountains at altitudes over 3,000 metres, and two measurements with manual reporting used a false calibration curve.

The second step of analysis is to look at the location of the measurements directly on the map. This visual inspection indicated that 128 measurements were made in areas with a high background due to naturally occurring radionuclides, and 63 measurements were made in Japan, in areas close to the Fukushima-Daiichi NPP. This result illustrates the behaviour of some contributors seeking hot spots: numerous people traveling to the Fukushima prefecture are making numerous measurements, expecting seeing high levels of ambient dose rates. However, thanks to the huge efforts of decontamination conducted by Japanese authorities, these areas of high ambient dose rates are limited to the vicinity of the NPP and some forest areas. This second step of analysis allowed to explain a total of 191 alerts (15.6% of alerts made at ground level).

The third step of analysis was conducted through contacts with the contributors, over the 1,018 remaining alerts. Unfortunately, 11 of these measurements were published by non-registered users and 6 users did not respond to the contact, preventing any identification of a plausible cause. 22 measurements were made close to urban objects enriched in naturally occurring materials or patrimonial objects used during pedagogic activities (science festivals, etc.). The remaining 979 measurements were due to 4 contributors, who made test studies with reference radioactive sources.

This study on a sample of the OpenRadiation database shows that the proportion of published measurements above 0.3 µSv.h^-1 which remains unexplained is very low, 0.022% in this sample. Results also show that the main source of uncertainty is the way results are reported by the users, especially when measurements are made in specific circumstances, such as measurements close to a radioactive source, whatever the reason is. However, all the metadata included in the OpenRadiation database were not analysed in this study. For instance, users can publish comments with their results and text analysis of these comments may help understand the reasons of these elevated dose rate results. As well, environment tags (see Figure 1) were not analysed while they can provide additional explanations. Moreover, data were analysed step by step, while a multiparameter analysis should provide additional information. Anyhow, these preliminary results provide a first flowchart for an analysis of the OpenRadiation database.

4 Meeting the challenges

This study is limited to a specific calendar period and examine only measurement with a result above the alert threshold 0.3 µSv.h^-1, with the specific objective to explain these high-level measurements. Moreover, the completion of data was not examined. Some data may be incomplete (lack of or incorrect GPS location, date or time) or may include physically non-realistic results such as altitude lower than the sea level, negative dose rate or counts equal to 0. Thus, supplemental cleaning of data is needed. However, this study also demonstrates that CM measurements can capture local situations of increased background levels of radioactivity, which could be useful for environmental monitoring on a local scale. This was already observed in Japan (Yasmint et al., 2026). For instance, a contributor measured in 2023 high levels of radioactivity on a beach in the western part of France. Most of the measurements made on that beach showed ambient dose rates between 0.4 and 2.63 µSv.h^-1, with a hot spot detected in the cliff over the beach at 78,4 µSv.h^-1, while the usual background in this area is generally lower that 0.3 µSv.h^-1 (Figure 8A). Such a situation cannot be detected by the Teleray network, since the nearest beacons are located more than 30 km away from the beach (Figure 8B). In fact, the Teleray network is intended to atmospheric measurements of ambient dose rates in case of nuclear or radiological emergency, able to provide an early warning, but is not designed to detect such local singularities. An extended study was then conducted by ASNR to characterize this radiological singularity and to evaluate the radiological exposure of population using the beach for recreative activities. Results of this study (ANSR, 2025) showed that the origin of these high dose rates were due to the presence of Monazite in the cliff, a red crystal enriched in thorium and uranium that produces “red sands” thanks to erosion. More importantly, the exposure evaluation showed an absence of risk for the population, based on protective hypotheses. This example of detection of a local radiological anomaly completed by an expert study is the demonstration that CM and expert monitoring can provide complementary data in environmental monitoring. As well, the usefulness of CM measurements in the post-accidental phase of a nuclear accident at the local scale was already demonstrated (Lepicard and Heriard-Dubreuil, 2001; Ando et al., 2026; Bertho et al., 2019).

So, the question that comes is why not using the CM data in environmental monitoring, either in normal situations or in emergency situation? Some examples are depicted hereafter.

4.1 Past and ongoing research projects using CM of radioactivity

The best example of such application of CM networks is of course the Safecast radiation monitoring network that developed in Japan. The long-term engagement and effectiveness of citizens was supported by several critical factors including data accuracy, knowledge production and citizen-related outcomes (Poudel et al., 2025). However, these local CM networks already work apart from local and national authorities yet operating in the same environment (Kenens et al., 2022). This independence of CM networks is crucial as they are able providing a different point of view on actual situation in the Fukushima prefecture and play a role in reassurance of local population yet influencing local policies.

Some research projects already used CM data from OpenRadiation since several years. The first one is CosmicOnAir (www.cosmiconair.org) which tracks solar flares through radioactivity measurements during flights (Trompier et al., 2022). This project is brought by a consortium of astrophysicists who uses these measurements to better understand the physics of the sun and the generation of solar flares with a potential output for radiation protection for flying crews, frequent flyers and astronauts. Data are generated by citizen who make measurements during flight using OpenRadiation detectors with “on board a plane” setting, which means that a specific calibration function is used to calculate ambient dose rate on board. The objectives of this project is clearly a scientific one, embarking citizen for data collection.

The second project, named CORALE (Laurent et al., 2021), aims at defining the radiation exposome of a sub-cohort of the French CONSTANCES epidemiological cohort (Zins et al., 2015). The radiological exposome is defined as the cumulated life-long exposure of a person, taking into account both internal and external exposure on the basis of life history, surveys and environmental monitoring. In this sub-cohort, some of the participants lived abroad for defined period of time, in countries where ambient dose rates are not measured or not easily available. The CORALE study uses a citizen science approach, asking travellers, either for professional or for personal reasons, to make some measurements abroad in different places targeted by the CORALE project. This helps in part to fill the gap in missing exposure data for this population embarked in the CORALE study.

In addition with the TERRITORIES project (see chapter 3.2) (Bertho et al., 2019), these three scientific studies demonstrate that CM of radioactivity data are fully useable for different purposes, according to the needs, in research activities. Moreover, these studies could represent a strong motivation to participate in CM networks for contributors.

Another example in the field of radiation protection is the recent EU-funded initiative used a citizen science approach to test and mitigate radon exposure, the RadoNorm project (Kulka et al., 2022). Previous actions have shown that top-down communication have limited effectiveness in motivating citizen engagement in radon mitigation (Perko and Hevey, 2024). This project developed an incubator for citizen science projects aiming at evaluating and developing mitigation actions adapted to the local context in various European countries. The development of these local, citizen-based projects was sustained in the long term (i.e., several years) by the health issues associated with radon exposure, the emergence of citizen-driven new radon dosimeters and innovative mitigation techniques. Public awareness also increased, resulting in improved local policies and expanded school curricula (Martell and Perko, 2026).

In addition, two main area of work were identified in a recent workshop (Bertho et al., 2026) that may be important to develop the use of CM data in research, expertise and environmental monitoring. The first one is to manage the uncertainties identified in CM datasets (as exemplified in chapter 3.4), and the second one is to enhance the ability of expert institutes and decision makers to be prepared to use CM datasets. These two areas are currently studied in two different European projects, the CIThARA project and the ORION project, respectively.

4.2 The CIThARA project

The overall concept of the CIThARA project, funded under the third PIANOFORTE open call, topic 3: “Use of Artificial Intelligence and Big Data techniques to improve emergency response” (https://pianoforte-partnership.eu/cithara/) is to develop and evaluate artificial intelligence (AI)-based methodologies to enhance specific thematic areas related to nuclear Emergency Preparedness and Response (EPR), with a special focus on thematic areas where there is a lack of AI applications, but with potential of considerable benefits. In this framework, the CIThARA project develop a specific objective, centred on the development of AI-based tools for radiological CM dataset analysis to enhance their quality, to demonstrate their added value, to render feasible their integration with data from official environmental monitoring networks so that they will be useable in the modelling system developed in other work packages, while ensuring the privacy of citizen personal data. The CIThARA project started in January 2026, for a three-year period.

A general strategy was already developed to identify and explain outliers, based on the analysis of the Openradiation database as fixed on 1 January 2026, with 1.3 million individual measurement results. The objective is to assess data availability, completeness and representativeness across variables. This will allow to identify main intrinsic parameters that are source of uncertainty, to investigate the presence and origin of outliers and to quantify the role of statistical uncertainties. This will allow to associate a confidence probability to each measurement based on duration, number of hits, and other metadata. The dataset was first cleaned by converting data to consistent formats, then to exclude incomplete data (lack of start time, lack of geolocation lack of values (counts, dose rate, etc.). The dataset was then filtered to select reliable ground-level measurements. This generated a dataset of 1,185,871 lines suitable for background analysis, representing more than 90% of useable data from the original dataset.

The data analysis was then started by looking at the distribution of data according to available parameters in the database (intrinsic parameters) such as duration of measurement, number of hits, environment, height above the ground, apparatus, etc. This analysis demonstrated that the identification of outliers needs a multiparametric approach, including temporal and spatial structuring. This step of the analysis will allow to identify low, medium and high background areas, in turn allowing to identify outliers. External sources of information such as geological mapping to identify soil nature in high background areas and contextual mapping to identify places of potential anthropic sources of radioactivity will be used to assess the reasons for locally elevated dose rates. The overall objective of these analyses is to ascribe a confidence index to each data that will allow on the first hand to identify outliers and potential explanations for these outliers and on the other hand to provide cleaned dataset for expert studies. The precise AI tools to be used in this project are not yet fully defined.

In parallel, reflexions already started on the comparison and integration of CM data and official monitoring network data to evaluate consistency and useability of CM data. For that purpose, the project will take advantage of existing environmental monitoring networks such as Teleray in France and RESORAD in Italy. The first step was to select areas of interest in both countries, based on availability of CM data and a sufficient number of network beacons. For that purpose, a library of French nuclear sites with corresponding Teleray fixed beacons and available CM data was created. This allowed to select 4 sites with the highest number of available CM data in a 20–30 km radius together with at least 16 Teleray beacons. These four French nuclear sites will serve as experimental datasets for CM data integration. Regarding Italy, the choice was made to look at the Piemonte and Val d’Aosta regions, since these regions are well-covered both by official networks and CM data.

The next step is then to extract data from the CM database, to standardize data formats, measurement units, geographic and temporal annotations. Thereafter, and once data from the official monitoring networks will be obtained, it will be necessary to align the different datasets for temporal resolution and spatial matching, thus defining necessary matching rules that will be generalized in a larger scale.

Another site was also selected because of a low number of CM data available and a high number of Teleray beacon and will serve as an experimental area for citizen engagement in a measurement challenge. This action will be supported by OpenRadiation and the local information committee of the NPP, which gather local representatives and stakeholders. The principle is to ask people living in the vicinity of this NPP to make as much as possible measurements in a limited time frame and to publish these results on the OpenRadiation map. The primary objective is to gather a significant dataset (<5,000 individual measurements) in a reduced period of time (2–3 months) and to compare it to the Teleray results in a unique time and space frame. The second objective is to study motivations and behaviours of participants through survey and targeted interviews. This action will be conducted in a full citizen science approach, meaning that participants will be implicated in all steps of the study, from data collection to data interpretation and exploitation in the frame of the CIThARA project. Such an action will allow to define conditions of citizen engagement in field action such as baseline determination of a site, detection of radiological singularities, or possibility to participate in radiological mapping in post-accident situations. Moreover, such studies provide a strong motivation to participate in citizen measurement activities on the long term, provided that the citizen input is fully recognized, both quantitatively and qualitatively (Pelacho et al., 2021; Berti-Suman, 2021).

4.3 The ORION project

The overall objective of the ORION project, funded under the EURAMET metrology partnership call 2025 (https://www.euramet.org/about-euramet) is to propose recommendations for a better consideration and integration of uncertainties in decision-making in nuclear or radiological emergency. In fact, decisions on protective actions requires prompt actions within a limited time frame but are based on incomplete and uncertain data, that may have a direct impact on effectiveness of protective actions and safety of citizens. The ORION project will identify gaps in measurement strategies to reduce uncertainties in emergency response decision-making and will issue recommendations for improvement of these measurement strategies.

The specific objectives of the project are to investigate early warning networks, aerosol sampling and environmental strategies, to analyse and improve measurement methods on food, drinking water and industrial products to reduce harm to trade and to improve harmonization of protective actions across borders and to provide recommendations on how to use measurements performed by citizens and drones during an emergency for a better inclusion of these non-classical measurement methods into decision making. These specific objectives will support the implementation of European regulations and international standards and facilitate the take up of project results by end users.

Regarding CM, the ORION project will analyse the current citizen science initiatives and instruments used in them. Measurement uncertainties and calibration practices will be analysed. Usability of mobile phone applications will be investigated despite their limitations (see chapter 2), and applications will be tested in the laboratory to see the current state of the art in this quickly evolving field. Measurement uncertainties will be analysed and recommendations on the usability of the data will be given. Practical use in selected exposure scenarios will be analysed (e.g., NPP accident, use of nuclear weapons) and recommendations will be provided on how to use these data in decision-making. Possibilities for users to do a self-check (verification) for the instrument’s reading to increase the representativeness of results and to have an estimate of uncertainty in measurement will be proposed. This project will start in early July 2026.

The generalizability of studies about data analysis is probably limited, due to the limited numbers of CM networks and their location mainly in Europe. However, the results of the CIThARA project may serve as a basis for the creation of a repository of CM data in a harmonized way, making these data available both to the scientific community and policy and decision makers. Moreover, the knowledge gained about the main sources of uncertainties encountered in CM data will serve as a basis for recommendations to be issued by the ORION project about quality management of measurement results, protocols and reports. Such recommendations will be fully generalizable to any citizen initiative aiming at measuring environmental radioactivity.

5 Conclusion

The current European projects as described above represents a significant change in the way CM are considered. Until recently, measurement of radioactivity by citizen were considered useful mainly for educational purposes or awareness-raising actions about risks in the absence of radiological issue, and in post-accident situation, mainly for the development of a local radiation protection culture in affected populations (Ando, 2015; Bertho et al., 2019; Hériard-Dubreuil, 2006) in the frame of the co-expertise process (Lochard et al., 2026), as recommended by ICRP (ICRP, 2020) and the SHAMISEN European project (Ohba et al., 2021; Liutsko et al., 2021). The use of CM was not considered by expert institutes and decision-makers mainly because of the underlying uncertainties existing in CM, both from the technical (calibration, verification), measurement protocol and reporting points of views. However, the landscape is changing, due in part to the rapidly evolving situation in Eastern Europe which re-activated fear of nuclear weapons use in the battlefield, but also due to the large amount of data accumulated over time by these CM networks, which make them attractive for expertise work. In the meantime, some research projects started to use CM data, demonstrating their usefulness.

In addition, two recent European projects CIThARA and ORION embark both technical supporting organisations (TSO) and nuclear regulatory organizations (NRO), along with scientific institutes, universities and CM networks: this is a potential game changer. In fact, both environmental surveillance and the management of emergency and post-accidental situations require a large amount of data, and CM networks are able to provide it from the field in a very short time. For instance, Safecast detectors can collect 1 measurement per minute, and OpenRadiation or OpenRed systems can collect 15–20 measurement data per hour in a normal background area. Moreover, geographic coverage is complementary to the one provided by official measurement networks (see Figure 8). Experience feedback suggests that, in emergency situations, one would expect thousands, if not hundreds of thousands CM to emerge rapidly once the information about a potential or real nuclear or radiological accident is broadcasted. In these conditions, TSO and NRO have a high interest in being prepared to use these CM data in emergency situation. This means that a double path emerges, on the one hand by citizen keen on radioactivity measurement to improve the quality of their measurements and on the other hand by TSO and NRO to use CM data, even if it does not reach the quality level of expert measurements. Dialogue between these different actors is therefore the key for future developments in CM use.

Nevertheless, CM networks have to remain independent from public authorities and technical supporting authorities, since it remains an essential driver of citizen confidence. This means that the use of CM data by these authorities should not be considered as anything else than a recognition of the quality of CM data, making them useful for expertise and decision-making in a continuous dialogue between CM networks and regulators (Berti-Suman, 2021). Moreover, CM network also advocate that radioactivity measurements using small detectors, adapted to citizen use, and eventually provided as do-it-yourself kits, is the best way to educate about radioactivity and risks associated with, and more generally to educate to science, which is a key resilience feature for the population, should an accident occur. Such educative actions should remain at the core of activities of CM network values.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Nadja Zeleznik, Milan Vidmar Electric Power Research Institute, Slovenia

Reviewed by: Abdus Sattar Mollah, Military Institute of Science and Technology (MIST), Bangladesh

Judelyn Patero, Surigao del Sur State University, Philippines

Footnotes

On behalf of the Open Radiation consortium composed of national association of local information committees (ANCCLI), the institute for continuous education to major risks (IFFO-RME), The FabLab from Sorbonne university, the Association Planète Sciences, and the French nuclear nuclear safety and radiation protection authority (ASNR).