1. Introduction

Various radiological surveys have been conducted to investigate the effects of radioactive material released into the environment due to the Fukushima Daiichi Nuclear Power Station (FDNPS) accident. A decade has passed since the accident, and radiological surveys have been continuously conducted and used as primary data for policymaking, such as lifting the evacuation order zone and decontamination. In 2017, the local government designated a specific reconstruction reproduction base area to accelerate the difficultto- return areas’ recovery and revitalization (Fig. 1) [1]. The central and local governments are together to create an environment conducive for residents to return to their homes and lift the specific reconstruction reproduction base area between 2022 and the spring of 2023. In response to local governments’ requests and recommendations from the ruling party, it was decided at a joint meeting of the 30th Reconstruction Promotion Council and the 55th Nuclear Emergency Response Headquarters held in August 2021 that decontamination and removing evacuation orders will be conducted in the 2020s to return to the areas. Continuous and dependable radiological surveys are required to lift the remaining evacuation zones.

Fig. 1

Difficult-to-return zones as of April 2022 and specific reconstruction and revitalization sites for the early lifting of restrictions. The background map is of ambient dose rates in FY2020.

Various new technologies have been proposed to meet the need for radiation surveys to visualize radiation distribution quickly, easily, and accurately in the environment, which did not exist before the accident. Immediately after the accident, crewed helicopters monitored the area to identify the diffusion range of released radioactive materials. The US Department of Energy (DOE) and the US military [2], conducted the first helicopter monitoring after the accident, and the technology was transferred to the Japan Atomic Energy Agency (JAEA) [3], and the JAEA is currently conducting regular monitoring [4]. A technology that combines global position system (GPS) and radiation information converted into ground-level value have been developed as a packaged technology combined with mapping and visualization technology using a geographic information system (GIS). This technology package has been applied to radiation survey technology using unmanned aerial vehicles (UAVs), which have been developed remarkably, and various methods have been proposed. Currently, large, unmanned helicopters, a type of UAV, are used to regularly measure the ambient dose equivalent rate distribution (ambient dose rates) around the FDNPS site. Unmanned helicopters are superior regarding maintainability and safety because they have been developed and refined domestically. The unmanned helicopter is used for ambient dose rate surveys near the high-dose-rate FDNPS sites [5] and riverbeds [6].

A national project is underway to combine the airborne radiation survey technique, which can measure a wide area, with a conventional ground-based survey [7]. This project visualizes the environmental radiation within 80 km of the FDNPS. It includes the airborne survey described above, a fixed-point survey using a sodium iodide (NaI) survey meter [8], a car-borne survey [9] and a backpack survey [10] using a dedicated radiation detector linked to a GPS and in situ measurement using germanium (Ge) detectors to measure the inventory of radiocesium on the ground [8]. This project is conducted yearly to evaluate the effective half-life, a parameter describing the decreasing trends of ambient dose rate [7]. The effective half-life combines both the physical decay and ecological decreasing such as migration or weathering. Furthermore, the location factor, a parameter characterizing the environmental ambient dose rates due to land use, has been assessed based on the differences in measurement methods [11]. The location factor is the exposure ratio at a given location to the exposure at 1 m above an infinite smooth lawn and has been used to characterize external exposure in some environments [12]. Furthermore, an integrated map technique is applied to integrate data from various measurement methods using a statistical approach and present them on the same map [13]. The past decade’s big data are available in a database on the JAEA home page [14].

This detailed radiation survey information serves as primary data for governmental decisions on decontamination policies such as lifting evacuation zones. These environmental radiation surveys of such a large-scale accident in Japan provide crucial insights for post-accident responses. This paper reviews the national radiation survey projects of the past decade since the accident.

2. Environmental Radiological Survey Methods and Results

2.1 Large-Scale Radiation Survey Project

After the accident, local governments, and designated public organizations-initiated radiation monitoring, focusing on measuring the ambient dose rates, and data were obtained over time, although the measured area was limited and the geographical density was rough. The airborne monitoring by the U.S. revealed full picture of contaminated area, and the necessity of comprehensive monitoring on the ground level in wider area was claimed. The Emergency Response Center (ERC), which oversaw monitoring at the time of the accident, started the “Survey and Research on Distribution of Radioactive Substances” as a national project, to establish a system to obtain and analyze environmental monitoring data systematically. We added the objective of investigating the detailed distribution of radioactive materials based on the information from the systematic soil sampling and ambient dose rate measurement with reliable and standardized methods. During the implementation period, more than 90 organizations (universities, research institutes and private companies) in addition to JAEA cooperated in this soil sampling campaign, and measurements of radioactivity concentration in the soil samples were shared among 21 universities, research institutes that own radioactive measuring instruments. The radionuclides detected at a sufficient site for mapping in the June 2011 survey were ¹³⁴Cs, ¹³⁷Cs, ¹³¹I, ^129mTe, ^110mAg, ⁸⁹Sr, ⁹⁰Sr, ²³⁸Pu, and ^239+240Pu. More than 99% of the contribution to the ambient dose rate came from radiocesium (¹³⁴Cs and ¹³⁷Cs) [15,16]. This result confirms the importance of radiocesium for long-term exposure doses. This project is referred to as the “Map Project” and is still being implemented as a Nuclear Regulation Authority project while streamlining the project contents.

However, the US DOE had been conducting aircraft monitoring using crewed airplanes and helicopters in cooperation with the military immediately after the accident. Since such technologies and systems were not available in Japan at that time, JAEA received the technology transfer, and conducted the monitoring while acquiring the technologies. Immediately after the accident, the Japan Self- Defense Forces and each municipality’s fire and disaster prevention teams used helicopters with radiation detectors borrowed from the U.S. for monitoring. However, in September 2011, the Ministry of Education, Culture, Sports, Science and Technology commissioned the JAEA to conduct a nationwide radiological survey from Hokkaido to Okinawa using private helicopters as a government project [17]. The government used the survey results in policymaking, such as setting evacuation zones and decontamination areas, because it was possible to visualize the entire picture of contamination.

2.2 Airborne Radiation Survey

Airborne surveys are being conducted in national projects using crewed and unmanned helicopters to survey large areas efficiently. Since crewed helicopters carry radiation detectors on board, aircraft without fuel tanks on the fuselage’s bottom, such as Bell412 and Bell430 (Bell Textron Inc. TX, USA), are selected. An unmanned helicopter, the FAZER-R manufactured by Yamaha Motor Co. (Shizuoka, Japan; Fig. 2), is used. The advantage of airborne radiation surveys is that ground surface γ-ray data can be collected with the same accuracy over a wide area, including mountainous areas.

Fig. 2

Photo of an uncrewed helicopter (FAZER-R) used for environmental radiation surveys.

Fig. 3 shows an image of the data acquisition and analysis method. Scintillation detectors and dedicated devices have been developed for each helicopter type. Currently, a commercial system (RSX-3, Radiation solution Inc., Canada) is used for crewed helicopters, and a proprietary system developed by the JAEA is used for unmanned helicopters. It acquires a once-per-second readout of spectrometers to produce a 1024-channel energy spectrum at 3 keV per channel. The readings are synchronized with time from a GPS receiver. The spectrum and the GPS data (date, time, latitude, longitude, and height above ellipsoid) are recorded every second. The spectrum data are processed using parameters, such as an attenuation factor and an ambient dose rate conversion factor, to retrieve the ambient dose rate at ground level.

Fig. 3

Airborne survey methodology.

The detailed analytical technique was according to the International Atomic Energy Agency technical standards [18]. We calculated the altitude distribution of ambient dose rates as follows. First, we superimposed raster data from a digital elevation model on those from the ambient dose rate of the airborne survey embedded in GIS software [19]. Raster data comprise cells (pixels) arranged in a grid (grid-like) of rows and columns. Second, we combined the digital elevation model (DEM) data with the ambient dose rate data on the nearest grid point. The spatial resolution is 0.25 km grid for the crewed helicopter and 0.05 km for the unmanned helicopter. DEM data are typically in a 0.01 km grid resolution. Mapping was performed by supplementing unmeasured areas by interpolating the measured results. Even though various methods, such as kriging and spline approaches, have been proposed for interpolation, the inverse distance weighted (IDW) method assigns weights to the measurement point values linearly and in inverse proportion to the distance and is applied to the airborne survey data. The IDW method is easy to use when analyzing a large amount of data because the parameter setting is straightforward [19]. This interpolation was conducted using ArcGIS software (Environmental Systems Research Institute Inc., CA, USA).

Fig. 4 shows the results from November 2011 after the accident and the latest airborne survey. The results for 2011 and 2021 are from Sanada et al, 2018 [4] and the latest NRA survey [20], respectively. Both are shown on the same map with decay correction on the date of measurement completion. Areas with higher ambient dose rates than 1 μSv h that once extended 50–80 km from the FDNPS, where populous cities, such as Fukushima and Koriyama are located, were shrinking in 2021. However, there are still areas around the FDNPS that exceed 1 μSv h⁻¹, and these areas are designated as difficult-to-return zones (Fig. 1). Kato et al. created an inventory map of ¹³⁷Cs by correcting the ambient dose rate information obtained from the aircraft survey for each land use [21]. Fig. 5 shows the corrected radiocesium map color-coded in the same range as the ¹³⁷Cs map around the Chernobyl Nuclear Power Plant, often used in international reports [22]. Although the environmental cesium contamination from the FDNPS accident was smaller than that of the Chernobyl Nuclear Power Plant accident, the maximum deposition levels were comparable.

Fig. 4

A contour map of ambient dose rate distribution on the topography map. (a) The tap as corrected on November 5, 2011, and (b) on October 25, 2021. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

Fig. 5

A contour map of ambient dose rate distribution on the topography map as corrected on July 2, 2011. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

Unmanned helicopters conduct airborne surveys in areas around the FDNPS where Japanese aviation law prohibits crewed flights. Because the Unmanned helicopter’s flight altitude is low (80 m above ground level), a distribution map with higher positional resolution than that of a crewed helicopter can be created. Fig. 6 shows the ambient dose rate map obtained from the airborne survey in the current difficult-to-return area. The area surrounded by a black dashed line around the FDNPS is the Unmanned helicopter’s measurement area. These results are an update of the results reported in Sanada et al. 2018 [4]. The distribution of crewed and Unmanned helicopters can be seamlessly drawn on the same map. Furthermore, the currently designated difficult-to-return zones have moderately high ambient dose rates of 1 μSv h⁻¹ or higher. The ambient dose rates are decreasing, and the results for 2021 show a few locations where the ambient dose rates exceed 10 μSv h⁻¹.

Fig. 6

Ambient dose rate maps in hard-to-return zones. (a) The map as corrected on October 20, 2012, (b) October 22, 2015, (c) October 16, 2018, and (d) October 3, 2021. The black boxed area results from an airborne survey using an uncrewed helicopter, and the rest of the areas are crewed helicopters. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

Visualizing the changes in ambient dose rates is crucial for future projections and assessing the impact of anthropogenic activities. Sanada et al. 2019 visualized the ratios of the latest airborne survey results and showed their relationship to land use [4]. Fig. 7 compares the visualization of the ratio of airborne surveys in 2011 and 2021 with the residential and agricultural areas. Thus, the areas with a 10% or less ratio over a decade are consistent with residential areas. Areas with a high decrease rate can be found near the FDNPS, indicating that the decrease in ambient dose rates in residential areas is faster than in mountainous and forested areas where human activities are less frequent, indicating the effects of decontamination and human activities.

Fig. 7

The ratio image of the resulting ambient dose rates in 2011 and 2021. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

2.3 Ground-based Survey

In the fixed-point surveys, the ambient dose rates at 1 m above the ground were measured at approximately 6,000 locations in the 80 km zone of each campaign using a handheld NaI(Tl) scintillation survey meter (TCS-171B or TCS172B; Nippon RayTech Co. Ltd., Tokyo, Japan) and ionization chamber–type survey meters (ICS-323C; Nippon RayTech Co. Ltd.) according to the magnitude of the ambient dose rate to be measured. The ionization chambers were used when the ambient dose rates were higher than the upper limit of reliable measurements using the NaI(Tl) scintillation survey meters (30 μSv h). Survey meters calibrated within a year preceding the measurements were also used. The device manufacturers or registered organizations conducted the calibration using standard radiation sources or fields traceable to national standards. The criteria for selecting the measurement locations, details about the survey meters’ measurement conditions, and the data recording procedures were the same as those reported in a previous study [8].

In the national projects, a dedicated system linked to GPS called the Kyoto University RAdiation MApping system (KURAMA) is used for continuous vehicle and backpack surveys. KURAMA-II (Matsuura Denkosha Co. Ltd., Ishikawa, Japan) is a simple survey system that can be used even without technical knowledge of radiation measurement, such as on connecting measuring devices or starting the measurement system. The KURAMA-II system, developed for car-borne surveys, has been used for backpack surveys around the FDNPS [10]. KURAMA-II can measure ambient dose rates and GPS positions every three seconds and automatically transmit these data to an Internet cloud server. An automatic data processing system was used with the KURAMA-II system for car-borne surveys to correct the measured data’s GPS positions based on roadmap data. Our backpack survey measured the ambient dose rate approximately 1 m above the ground with the measurement point’s positional information (coordinate). A CsI(Tl) scintillation detector (C12137; Hamamatsu Photonics, Hamamatsu, Japan) was used for KURAMA-II. The monitored ambient dose rates were based on the G(E) function (spectrum-dose conversion operator), which can accurately calculate ambient dose rates from a measured pulse height distribution even if significant changes occur in the energy spectrum [23]. The measurement regions were divided into 100 m × 100 m meshes for the car-borne survey and 20 m × 20 m for the backpack survey. Statistical errors were reduced by averaging the raw data over the 100 m meshes for the car-borne survey and the 20 m meshes for the backpack survey.

In the national project, the car-borne survey surveyed approximately 85,000 km (at most) of roadway per measurement campaign over a broad area of eastern Japan. Fig. 8 compares the results from June 2011 and the latest carborne survey. These results are an update of those reported by Andoh et al. 2016 [9]. As with the airborne survey, the ambient dose rates decreased overall. The backpack survey was conducted on the sidewalks in the residential area. Fig. 9 shows the latest backpack survey results around the current difficult-to-return areas. These results are an update of those reported by Andoh et al. 2019 [10]. Fig. 9 shows the enlarged results of the measurements around the railway station in Okuma-machi. The areas at the northern and southern ends have moderately high ambient dose rates, and a gradient in ambient dose rates can be observed even on the same straight sidewalk. The forest near the sidewalk was responsible for the high-dose areas in this patch. In this way, backpack surveys are more time-consuming than airborne surveys but provide a detailed picture of the ambient dose rate situation in the living area.

Fig. 8

A contour map of ambient dose rate distribution by car-borne surveys. (a) The map as corrected on June 13, 2012, and (b) December 3, 2021. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

Fig. 9

A contour map of ambient dose rate distribution by backpack surveys as corrected on November 18, 2021. (a) the entire area where the backpack survey was conducted, (b) expands on the central part of Okuma town as an example. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

2.4 Intercomparison of Methodology

Understanding each radiation survey method’s characteristics is crucial to determining the distribution of ambient dose rates. Fig. 10 shows an image and characteristics of each survey method. Specifically, since car-borne surveys are conducted on roads, the ambient dose rates tend to be lower than those of other surveys. However, airborne surveys have a broad view, including forested and mountainous areas. However, airborne survey data are coarse in positional resolution because of the long distance between the detector and the object to be measured. Extracting a continuous dataset of the same locations from the big data is necessary to accurately assess the temporal changes in ambient dose rates measured using various methods quantitatively. Here, we compared each survey’s results with the most recent ones using the method of Sanada et al. 2022 [11], outlined below.

Fig. 10

Image and characteristics of each measurement method.

The ambient dose rates of the national project included the background dose rates derived from natural radionuclides. Therefore, the background dose rate should be distinguished from these ambient dose rates to evaluate the severity of the FDNPS accident. Sanada et al. 2020 [24] used the results of several airborne surveys conducted after the accident to assess pre-accident ambient dose rates derived from natural radionuclides. These data’s accuracy was evaluated to compare the ground-based data acquired using a portable Ge detector. The background dose rates averaged over a 250 m × 250 m mesh were subtracted from each survey result. An analysis dataset was created by first establishing several 1 × 1 km² meshes within 80 km of the FDNPS and comparing the different survey results. The survey data within each mesh was the average from each survey method. Next, an analysis dataset was created from all survey meshes, and changes in ambient dose rates over time were analyzed. The selected mesh included data from all surveys conducted throughout the period (airborne, fixed-point, backpack, and car-borne surveys). For example, the results of the 2014 and 2021 national project campaigns were applied.

The survey results were compared based on the relative deviation (RD). Specifically, using a scatter plot, the fixedpoint survey data (D_a) were compared with the other survey data (D_b) in the same mesh. The RD of each survey mesh was calculated using Equation (1) to evaluate the accuracy of this study’s scheme, and the calculated RDs were used to estimate the total error and statistical uncertainty, displayed as a frequency histogram.

All survey methods’ results in this study were compared, and the scatter plots obtained from comparing the survey campaigns in 2014 and 2021 are given as an example in Fig. 11. These scatter diagrams show that the fixed-point survey positively correlated with each of the three other survey methods, and the RD histograms reveal the unevenness of each pair of surveys being compared. Table 1 summarizes the statistical figures for RD. A positive RD in this figure and table indicates that the survey method under comparison tends to overestimate and a negative RD indicates that the survey method under comparison tends to underestimate compared to a fixed-point survey, respectively. Airborne survey tended to be higher compared to fixed-point surveys, while car-borne and backpack surveys tend to be lower. This trend confirmed the same relationship regardless of the time elapsed since the accident, as the results for 2014 and 2021 shown as examples [11]. According to simulation studies, the field of view of radiation detection using airborne surveys is more expansive than those by ground-based surveys because the distance between the detector and the ground in airborne surveys is longer [25]. The car-borne and backpack survey results were lower than those of the fixed-point surveys. Statistics obtained from such large-scale monitoring are important in the post-accident context, as they can be used as coefficients to estimate environmental dose rates in other land conditions, similar to survey meter measurements on unpaved land.

Fig. 11

Statistical comparison of the survey results by measurement method. (a)–(c) The survey results on FY2014 and (d)–(f) the survey results on FY2021.

Table 1

Example of intercomparison results for each measurement method

3. Discussion

3.1 Temporal Change of Ambient Dose Rate

Understanding the decreasing trend in ambient dose rates is crucial for assessing the current state of environmental contamination around the FDNPS and for postaccident knowledge. A ground-based survey conducted by Yoshimura et al. 2020 [26] showed that the decreasing ambient dose rate on paved surfaces is faster than that of unpaved surfaces. Golikov et al. 2002 [27] evaluated the location factor based on ground-based survey data obtained around the Chernobyl Nuclear Power Plant accident area because it is a suitable indicator of the correlation between the ambient dose rate and land use. Sanada et al. 2021 obtained a location factor using large-scale survey data to assess the reducing ambient dose rates per land use [11]. Nakama et al. 2019 [28] statistically calculated the changes from continuous survey results before and after radioactive decontamination. This section presents the characteristics of temporal change for each survey method from the latest survey results.

We examined trends in ambient dose rate changes by unifying measurement locations as much as possible by extracting reference area meshes where ambient dose rate data from all measurement methods are available (carborne, backpack, fixed-point, and airborne) and examined each measurement method’s characteristics and trends. In preparing the analysis dataset, we first set a series of 1 × 1 km² mesh within the 80 km zone from the FDNPS to compare the different survey results. The survey data of ambient dose rate within each mesh were the average value of each survey method’s data. Among the national project data from 2011 to 2021, there are 1,340 meshes, primarily in inhabitable areas, where all four data types exist. Each measurement method has the following characteristics.

Fig. 12 shows each measurement method’s geometric mean ambient dose rates and smoothed approximation curves at 1,340 locations. The ambient dose rates increase in the order of airborne, fixed-point, backpack, and carborne surveys. The decreasing trends of the airborne and fixed-point surveys were similar until the second year after the accident, but from the third year onward, the difference between the two surveys widened, and the fixed-point survey became lower. This result is because the decontamination work started in earnest after the third year and the airborne survey measures the radiation from 300 m above the ground; therefore, the radiation gradient in a narrow area cannot be traced because it measures the average radiation value in the area within a 300 m radius directly below the ground. The relationship between the measurement altitude and the source field of view was evident from computational evaluations [29]. Furthermore, the backpack survey is lower than the fixed-point and aircraft surveys because the decontamination focused on the areas around houses and the washing away of radioactive cesium on paved roads by rainwater and other means. The car-borne survey is lower than the other three measurement methods.

Fig. 12

Survey methods and temporal changes in ambient dose rates.

Human activities significantly affect these differences in measurement targets. Yoshimura (2021) reviewed continuous survey data and examples of studies obtained from urban areas around the FDNPS and showed that ambient dose rates in urban areas decreased faster than in other land uses due to high ¹³⁷Cs washout on paved roads and anthropogenic influences [30]. The dynamics of radionuclides in the terrestrial environment are critical for environmental remediation and radiological protection after a severe nuclear disaster.

3.2 Integration of Different Survey Results

Because each ambient dose rate survey method has unique characteristics, the measured data are evaluated and discussed individually. The algorithm for applying the hierarchical Bayesian model, used in the field of spatial statistics to create maps by integrating spatial ambient dose rate distributions obtained using different survey methods, was developed in collaboration between the JAEA and the Lawrence Berkeley National Laboratory in the US and was validated for a small area around Fukushima City [13]. The above paper details the methodology, and an overview is presented here.

The method statistically estimates the most reliable ambient dose rate distribution y, when the backpack survey data set z_w, the car-borne survey data set z_c, and the airborne survey data set z_A are obtained from contemporaneous measurements covering the same area. Specifically, y and its variance are estimated so that the conditional probability p(y|z_w,z_c,z_A), expressed by the following equation (2), is maximized. p(y|z_w,z_c,z_A) means the probability that the spatial dose rate distribution becomes y under the condition that z_w, z_c and z_A are observed. p(y|z_w,z_c,z_A) can be written separately by Bayes’ theorem as follows,

where, p(z_A|y) is a probability of observing z_A when the ambient dose rate distribution is y, p(z_c|y) is a probability of observing z_c when the ambient dose rate distribution is y, and p(y|z_w) is a probability that the ambient dose rate distribution becomes y when z_w is observed. p(y|z_w) is called the prior probability, and the obtained spatial dose rate distribution y based on the observed value z_w is called the prior distribution. p(y|z_w,z_c,z_A) is called the posterior probability, and the obtained ambient dose rate distribution y is called the posterior distribution. Since the backpack survey provides ambient dose rates that are closely related to human activities, it is assumed that the ambient dose rates obtained by the backpack survey are the actual ambient dose rates. For locations where backpack survey data are not available, the ambient dose rates are estimated assuming backpack surveys were conducted at those locations. To obtain y satisfying the conditions of equation (2), the underlying data for the three probability functions on the right-hand side are obtained by a statistical analysis of the measured data. The data models associated with p(z_A|y) and p(z_c|y) are obtained by analyzing the correlation between the aircraft survey and the backpack survey, and the correlation between the car-borne survey and the backpack survey, respectively. The process model associated with p(y|z_w) is obtained from the analysis of spatial patterns of backpack survey results. Specifically, how the spatial dose rate between two distant measurement points varies with distance in the backpack survey is analyzed. Since it is known that correlations between surveys, which form the basis of the data models and the process model, vary depending on the land use, statistical analysis is conducted for each land use status. Using these relationships, y and its confidence interval (standard deviation) is estimated.

Fig. 13 shows an integrated map of the 80 km radius that integrates measurement data from the backpack survey, car-borne survey, and airborne survey conducted in 2021. When the depicted area around the FDNPS is enlarged, the complexly represented dose rate distribution can be seen. Currently, the area around the FDNPS is undergoing decontamination and the construction of an interim waste storage site, and the distribution of dose rates has been affected by such human activities. An integrated map can not only provide an accurate dose rate information to residents, but also be effective in assessing residents’ exposure.

Fig. 13

Ambient dose rate maps with statistical integration of the four survey results for 2021. The map was created using ESRI ArcGIS 10.7.1. Basemap sources: Digital Map 25000 (Map image) published by the Geospatial Information Authority of Japan.

3.3 Publication of Databases

The JAEA regularly collects and publishes environmental survey data related to the FDNPS accident published by various organizations [14] via an online database. The gathered data are converted into a standard file format (CSV format) suitable for analyzing the spatial distributions and temporal changes of radioactive materials. The JAEA has provided collected data to our website (https://emdb.jaea. go.jp/emdb/). The website was accessed over 100,000 times per month on average, but it had small functions because it was a stable website. Furthermore, registering new data was laborious; therefore, the manual work was extensive for data conversion and preparing map images and graphic data. [31]. The JAEA upgraded the website to have a dynamic system based on web application and background relational database systems. A background relational database system is a structure where a single piece of data is expressed as a set of values for multiple attributes, and the data are stored by enumerating the sets. The collected data were segregated into ambient dose rate and radioactivity, with radioactivity segregated into 12 categories: (1) soil depth distribution, (2) seawater, (3) marine soil, (4) deposition soil and environment samples, (5) airborne particles, (6) fallout, (7) river water, (8) river soil, (9) drinking water, (10) wild animals and aquatic organisms, (11) ship-towing survey, and (12) food. These sorting were considered to display the same units on the same map. These data were 59,322,151 as of May 2022. These CSV data were registered in a database, and a web-based application was built to control the search and display functions using a structured query language. The data map display is automatically averaged within a predefined mesh according to the scale one wishes to display. Various maps and graphs showing the distributions of radioactive materials are available on the website to promote intuitive understanding for users (fig. 14).

Fig. 14

An overview of the JAEA’s website for publishing survey data.

In addition to the map visualization function, this database has trend display, search, and download functions. Downloaded files can be selected from CSV, text, and shape file, the standard features of ArcGIS. For soil depth distribution, a unique vertical distribution graph can be displayed. The big data can be used as machine learning supervisory data for more complex analysis. Furthermore, standardizing the data characteristics from different measurement methods and creating an integrated map will enable a more accurate understanding of the current situation. For more advanced analysis, we plan to register the integrated map data that are statistically integrated from different survey data, such as airborne surveys, car-borne surveys, backpack surveys, and monitoring posts (fixed station radiation monitoring systems).

4. Summary

This paper reviews the national radiation survey projects conducted over the past decade since the FDNPS accident. A decade after the accident, radiological surveys are still ongoing. Many evacuation zones within a 30 km radius around the FDNPS after the accident have now been lifted. In response to requests from local governments and recommendations from the ruling party, decontamination and removing evacuation orders will be conducted in the 2020s to return to the areas. Survey data, which has continued since the accident, have been used as basic information for policymaking on lifting such zones. Environmental radiation surveys of large-scale accidents provide crucial insights for post-accident responses. The characteristics of various environmental radiation survey tools mentioned in this review should be understood in normal circumstances, and practical training is essential. Publicizing survey data is crucial to promote a correct understanding of the population and foreign countries. Furthermore, the large-scale radiation monitoring and mapping technologies that developed after the accident, such as UAV-based radiological survey technology, should be preserved for future generations. Further development is expected with GIS technology to visualize the radiation distribution and this measurement technology.