In an early stage after the accident, personal dosimeters were not widely available for the public. As noted above, how occupancy factors (behavior patterns) are assumed is one of the influential factors for dose estimation from ambient dose rate. Some evacuees had complex behaviors including multiple moves. Therefore, obtaining records of individual behavior (such as post-disaster evacuation behavior) from residents, and combining this information with ambient dose rate was considered to estimate external dose.

This was conducted as the “Basic Survey”, a part of the Fukushima Health Management Survey [13]. Details of the Basic Survey are described elsewhere [14]. Questionnaires asking behavior patterns for the first 4 months after the accident were sent to all persons registered as Fukushima Prefecture residents (around 2 million), including residents of non-evacuated areas. The Basic Survey estimates individual external effective doses using the digitized behavior records and a calculation program which includes 2×2 km² mesh daily ambient dose rate maps.

Individual external doses for the first 4 months were estimated for about 475,000 persons. Although the maximum dose was 25 mSv, 94% of the doses were less than 2 mSv. The Basic Survey also revealed actual occupancy factors after the accident. In the case of Iitate Village, average time spent outdoors per day was around 2 hours [15], which was much shorter than the 8 hours used by the government model. According to the Basic Survey, age dependence of individual effective doses was not so large: the 4-month doses for infants (<6 years old) and children (6–15 years old) were 1.1 times and 1.04 times larger than that for adults (>15 years old) on average in non-evacuated areas [16]. The ratios were smaller than those estimated by the UNSCEAR 2013 Report: around 1.7 times and 1.4 times smaller for infants and children for the first-year dose.

The UNSCEAR 2013 Report estimated municipality-average doses using this approach for non-evacuated areas [4]. The dose estimation was based on (1) ambient dose rate calculated from measured deposition density data and (2) several occupancy factors depending on age groups and occupation (outdoor worker/indoor worker). For evacuees, UNSCEAR used 18 typical evacuation patterns [17] to estimate their doses.

Dose estimation based on some typical behavior patterns by age group or occupation could give a more realistic dose than that estimated using a fixed pattern like 8 hours outdoors and 16 hours indoors. Takahara et al. [18, 19] investigated behavior patterns for four population groups: Fukushima City Office staff, senior citizens’ club members, contractors’ association members, and agricultural cooperative members. Then, probabilistic dose assessments were made for these population groups by considering the spatial variability of ambient dose rate and interpopulation differences resulting from behavior patterns. Mori et al. [20] applied the probabilistic approach to estimate the external doses that children would receive after returning to evacuated areas. Application of this approach is described in another paper of this journal in more detail [8].

Among the three factors affecting individual external dose assessment (Fig. 2), the Basic Survey has an advantage for the second point (obtaining actual behavior records), but it has disadvantages for the first point. In the Basic Survey, outdoor ambient dose rate was assumed to be uniform within 2 km×2 km mesh areas, but strictly speaking, that would not be true. Also, the Basic Survey used a fixed shielding factor for each type of building (e.g., 0.4 for wooden houses) to estimate indoor dose rate from outdoor dose rate, but the factor could differ from house to house [12].

Even for the second point, the Basic Survey was considered to have a potential weakness: it relies on people’s memories for the behavior records and “recall bias” may affect the individual doses. However, a comparison between dose estimates based on behavior records collected before and those collected recently indicated that the effect of recall bias could be small [21].

Another issue is that the Basic Survey is a volunteer-based survey, while the dose assessment using typical behavior patterns can be applied to a whole population group. Due to this concern, representativeness of individual doses estimated by the Basic Survey was investigated [22]. Although the response rate to the Basic Survey questionnaires was around 27%, the dose distribution was considered to be representative for the whole prefecture, according to a statistical test.

Although glass badges are useful in large-scale personal monitoring, their disadvantages are that (1) they generally need a measurement period of a few months and (2) it is not possible to determine when and where the person is exposed (at home, at work, etc.). A personal dosimeter, D-shuttle, overcoming these disadvantages was developed in 2013 by AIST (National Institute of Advanced Industrial Science and Technology) and it was made commercially available by Chiyoda Technol Corporation [35]. D-shuttle enables users to record hourly personal doses and to read out the hourly doses anytime by using a dedicated device for its administration.

Studies using the D-shuttle in combination with GPS systems allowed identification of exposure levels, locations, and times. In particular, it was possible to find the dominant exposure in the total exposure. Identifying source contributions to the total dose is important in determining effective dose reduction measures. In this respect, it has been broadly used for risk communication in Fukushima Prefecture [36, 37].

As described previously, personal doses for hometown returnees were obtained using integrating-type personal dosimeters. Monitoring for those people also has been done using the D-shuttle [38, 39]. A 2020 publication analyzed doses for 239 voluntary participants among returnees across ten municipalities [40]. Monte Carlo simulations used for quantifying both uncertainty and population variability of observed data demonstrated that the mean of the annual dose in 2019 (including natural background doses) was 0.93 mSv (95% uncertainty interval, 0.53–1.76 mSv), with limited variation among municipalities.

Immediately after the accident, no whole-body counters were available in Fukushima Prefecture. Thus, initially whole-body counters located outside the prefecture were utilized for measuring ¹³⁴Cs and ¹³⁷Cs body contents for residents from Fukushima Prefecture.

Morita et al. [47] reported internal contamination level for evacuees from Fukushima after the accident using a whole-body counter at Nagasaki University. Measurable ¹³⁴Cs and ¹³⁷Cs amounts were detected in 49 out of 196 people who were in Fukushima Prefecture at any time during March 11 to April 20, 2011. Among these 49 people, the 90^th percentile CED value was 0.06 mSv.

The Fukushima Prefecture government organized WBC measurements by a committed institute, the National Institute of Radiological Sciences, which started at the end of June 2011 [48]. A total of 174 subjects mostly from evacuated areas were measured until the end of July. The 90^th percentile CED value for adults was around 0.1 mSv and the maximum CED (0.63 mSv) was found in an elderly male.

Another committed institute, the Japan Atomic Energy Agency (JAEA), started WBC measurements on July 11, 2011. A total of 9,927 subjects were measured until the end of January 2012. Most of these subjects were residents of evacuated areas. The median CED values were 0.02 mSv and 0.025 mSv for subjects aged 13–17 years and for subjects aged >17 years, respectively [49].

Several months after the accident, installation of whole-body counters in Fukushima Prefecture was started. Some of them were a mobile type. The installed whole-body counters were operated either by (i) Fukushima Prefecture including its commissioned organizations such as JAEA; (ii) local governments (municipalities); or (iii) organizations independent of the prefectural or local governments. Data on measurements conducted by operators of category (i) have been collected by the Fukushima Prefectural government and they are periodically reported on webpages of Fukushima Prefecture [50]. According to these webpages, persons with CEDs greater than or equal to 1 mSv numbered 26 among more than 340,000 subjects measured. Also, no subject with a CED greater than or equal to 1 mSv was found after March 2012 (1 year after the accident).

Some results obtained by operators for categories (ii) and (iii) have been reported as scientific papers: these papers included subjects of Minamisoma City [51, 52], Iwaki City (including its suburbs) [53, 54], Namie Town [55], and Miharu Town [56], and subjects in and around Fukushima Prefecture [57, 58]. All papers reported very low doses as follows: (1) among 9,498 residents measured between September 26, 2011 and March 31, 2012, CEDs were less than 1 mSv except for 1 resident (1.07 mSv) [51]; (2) CEDs of 566 high-risk residents measured about 4 months after the accident were found to be less than 1 mSv [52]; (3) the maximum CEDs in two studies were both less than 0.1 mSv [53, 54]; (4) the average CED for residents with detectable radioactivity of ¹³⁷Cs or ¹³⁴Cs was 0.025 mSv [55]; (5) no child was found to exceed the ¹³⁷Cs detection limit of 300 Bq per body (corresponding to the CED of 0.04 mSv due to ¹³⁷Cs for a 6-year old) in 2012 and 2013 [56]; (6) among 2,700 babies, none had detectable levels of ¹³⁴Cs or ¹³⁷Cs (the detectable level of ¹³⁷Cs corresponds to 0.016 mSv per year including the contribution from ¹³⁴Cs) [57]; and (7) between 12 to 20 months after the accident, the ¹³⁷Cs detection frequency was 1.0% among all ages (0.09% among children) [58].

As mentioned before, personal dosimeters were utilized for estimating external doses for returnees to the former evacuation order areas. Most municipalities in such areas organize opportunities for persons to undergo WBC so that they can check their own internal contamination, if they wish. WBC results for returnees to Kawauchi Village were reported by Tsubokura et al. [59]. Although returnees had higher chances of consuming locally produced vegetables, proportions of individuals with internal radiation exposure above the ¹³⁷Cs detection limit were 3.4%, 1.6%, and 3.1% for the returnees, commuters and non-returnees among the Kawauchi villagers, respectively. A Cochran-Mantel-Haenszel (CMH) test showed the level of internal radiation exposure for the returnees was not significantly higher than that in the two other groups. This indicated that it was possible to maintain internal exposure at very low levels even in a highly contaminated region.

The UNSCEAR 2013 Report adopted this approach. It estimated ingestion dose based on a food database with some conservative assumptions. For example, many ¹³⁴Cs and ¹³⁷Cs concentrations in the database were shown as below the limits of detection and in these cases, it was generally assumed that ¹³⁴Cs and ¹³⁷Cs concentrations were each 10 Bq·kg⁻¹. The 2013 Report estimated inhalation dose based on atmospheric transport dispersion and deposition models for radioactive materials released from the FDNPP accident. The first-year dose due to ingestion was estimated to be 0.94 mSv for adults in non-evacuated areas. The first-year dose due to inhalation was estimated to be from 0 to 0.47 mSv for adults in non-evacuated areas, depending on location. Thus, the municipality-average first-year dose due to internal exposure ranged from 0.94 to 1.41 mSv for adults. Comparison with the WBC results mentioned in previous section indicates the dose was likely to be overestimated.

Actually, restriction orders for food supplies such as contaminated vegetables and milk, and intake of tap water were implemented within several days after the major release of radionuclides on March 15, 2011 [60]. In addition, collapse in supply chains, i.e., due to damage to distribution facilities, lack of transportation vehicles or electricity, and the closure of earthquake-damaged retail stores, contributed to a situation where contaminated food or supplies were not consumed in large quantities in general, even before the food restriction orders [61]. Due to this situation, ingestion of highly contaminated food was not likely to occur.

Harada et al. [62] estimated that the median CED was 23 μSv due to dietary intake of ¹³⁴Cs and ¹³⁷Cs for the year, based on food-duplicate samples collected in December 2011. Their CED was estimated assuming that the dietary intake of ¹³⁴Cs and ¹³⁷Cs was constant throughout the year. Koizumi et al. [63] purchased fifty-five sets of meals, each representing one person’s daily intake, in local towns in Fukushima Prefecture in July 2011 and analyzed them while including a daily tap water intake. The median CED was estimated to be 3.0 μSv per year (ranging from not detectable to 83.1 μSv per year), with the assumption that the dietary intake of ¹³⁴Cs and ¹³⁷Cs was constant throughout the year. Since 2011, ¹³⁴Cs and ¹³⁷Cs contents in food have been monitored and they seem to remain very low. Analyses by the duplicate diet method covering 100 families throughout Fukushima Prefecture showed that CED did not exceed 0.1 mSv [64]. Based on 7,668 food samples collected in Kawauchi Village in 2013 and 2014, Orita et al. [65] estimated that CED due to 1-year ingestion ranged from 24.4 to 42.7 μSv for males and from 21.7 to 43.4 μSv for females.

In this general situation that ingestion of highly contaminated food was not likely to occur, attention should be directed to residents who consume homegrown produce without radiation inspection, and who often collect wild mushrooms or cultivate their own mushrooms on bed-logs. Tsubokura et al. [66] advised those residents (n=9) to consume the distributed food mainly and to refrain from consuming potentially contaminated foods without a radiation inspection and local products under shipment restrictions such as mushrooms, mountain vegetables, and meat of wild game. A few months after the intervention, re-examination of ¹³⁴Cs and ¹³⁷Cs levels revealed remarkable reduction of internal contamination in all the nine residents.

Technical issues raised for internal dose assessment by WBC of Fukushima Prefecture residents were well summarized by Kurihara et al. [67]. The issues can be categorized as follows: (1) setting a suitable intake scenario to estimate intake amount from measured ¹³⁴Cs and ¹³⁷Cs body content (acute or chronic, inhalation or ingestion); (2) avoiding interference by the natural radionuclide ²¹⁴Bi in ¹³⁴Cs detection (both radionuclides emit similar energies of gamma rays, which makes spectrum analysis difficult); (3) standardizing calibration and measurement procedures; (4) establishing a method for estimating ¹³⁴Cs and ¹³⁷Cs body contents of children (whole-body counters were originally designed for measuring adult radiation workers); and (5) avoiding effects on measurements due to body surface contamination.

The first issue is associated with how to estimate the intake amount from ¹³⁴Cs and ¹³⁷Cs body content (see Fig. 3), while the four other issues are related to measurement accuracy of ¹³⁴Cs and ¹³⁷Cs body content. Although the four issues on measurement accuracy have been settled to some extent, how to treat the first issue in estimating early intake remains somewhat controversial [68–71].

Kunishima et al. [68] analyzed the relationship between evacuation behavior data obtained from 112 out of the 174 subjects who underwent WBC [48] and their CEDs. Most subjects living in municipalities near the FDNPP started evacuation promptly. The percentage of persons remaining within the 20-km radius area of the FDNPP was found to be 100% at 16:00 on March 12, 2011 and 42.9% at 0:00 on March 15 for those with CEDs >0.1 mSv, whereas the corresponding percentages were much lower for those with CEDs ≤0.1 mSv. Igarashi et al. [69] demonstrated that the ¹³⁴Cs and ¹³⁷Cs detection rates in the WBC results were both several times higher in the late evacuees (who evacuated outside the 20-km radius of the FDNPP at 15:00 on March 12 or later) compared to the prompt evacuees (who evacuated before 15:00 on March 12). These differences in ¹³⁴Cs and ¹³⁷Cs detection rates would be caused by exposure to the radioactive plume on the afternoon of March 12, which was likely to influence the late evacuees. Time-related changes in rate of positive detection of radiocesium (either or both of ¹³⁴Cs and ¹³⁷Cs) in WBC also supported that acute intake was the major contributor of ¹³⁴Cs and ¹³⁷Cs body contents [70].

On the other hand, Nomura et al. [71] showed that individuals who evacuated to areas outside Fukushima Prefecture had similar contamination levels of ¹³⁴Cs to individuals who stayed in Fukushima (relative risk, 0.86; 95% confidence interval, 0.74–0.99). They found that time spent outdoors had no significant relationship with contamination levels and they considered that effects of inhalation from radioactive plumes on total internal radiation contamination might be so low as to be undetectable by the WBC unit used to examine participants.

As mentioned in previous subsection, “interpretation of personal dose measurement results”, representativeness of subjects is one of the concerns about volunteer-based monitoring. Representativeness of WBC subjects was investigated by Nomura et al. [72]. Based on regression projection techniques, separate probabilities for ¹³⁷Cs and ¹³⁴Cs detection for a whole population were simulated. They were compared with ¹³⁴Cs and ¹³⁷Cs levels measured from October 2011 to March 2015 for voluntary participants. They found sufficient agreement between simulated and measured ¹³⁴Cs and ¹³⁷Cs levels except for ¹³⁴Cs in October 2011 and concluded that the voluntary monitoring participant group was a good representative sample [72].

Doses reported by the publications reviewed in the present paper were summarized in Tables 1 and 2. Table 1 summarizes doses estimated for the first-year (or first 4-month) doses or doses estimated based on measurements within the first year after the accident. Table 2 summarizes doses estimated for periods after the first year. The doses in both tables are shown with information on methodology for dose estimation, target population (age groups and residential areas) and year of measurement.

Internal doses were basically shown as CEDs due to intake within the periods of interest. The CEDs in the reviewed publications were calculated based on either acute or chronic intake scenario. The scenario adopted for each publication is shown in Tables 1 and 2. The acute intake scenario assumed that the intake occurred just after the accident and no further intake occurred after that. As described previously, the acute intake scenario was considered to be reasonable especially in the evacuated areas [68–70]. On the other hand, the chronic intake scenario assumed the same amount of consecutive intake for the period of interest in both approaches of food analysis and WBC. In the case of WBC, measured amounts of radiocesium (either or both of ¹³⁴Cs and ¹³⁷Cs) were assumed to be in an equilibrium state between consecutive ingestion and excretion throughout 1 year. The CEDs estimated by this scenario can be interpreted as resulting from ingestion that occurred within the year. An exception was the CED reported by Orita et al. [53] who calculated CED due to chronic ingestion from the day following the accident start to the day of measurement (more than 2 years as the longest period of assumed chronic ingestion).

The different methodologies shown in these tables can be compared in the following way. Regarding external dose estimation, personal dosimeters were not available on a large scale for up to around 6 months from March 2011. Thus, external dose up to then had to be estimated from ambient dose rate. As described before, Ishikawa et al. [14] estimated external doses taking actual behavior patterns into account, while WHO and UNSCEAR assumed typical behavior patterns. WHO’s estimation was based on conservative assumptions that people in deliberate evacuation areas stayed there for the first four months although the inhabitants were subjected to relocation at different times during the four-month period. Also, Ohba et al. [73] showed that the actual behavior patterns of residents differed from the typical behavior patterns used by UNSCEAR, although they analyzed only residents aged less than 20 years. In this respect, the estimation by using actual behavior patterns could give more realistic doses.

As described in previous section, internal dose estimation can be categorized into two approaches: estimation from personal monitoring (WBC) and estimation from monitoring of environmental samples such as radioactivity concentration in food and drinking water. If the latter approach is adopted with conservative assumptions, it results in overestimation of dose. This was the case for UNSCEAR and WHO’s dose estimations. That is, WHO assumed that all the food monitored was on the market although the monitoring data set included the results of food samples that were collected for monitoring purposes and were not allowed on the market [1, 3]. In the food database which UNSCEAR used, many ¹³⁴Cs and ¹³⁷Cs concentrations were shown as below the limits of detection and in these cases, it was generally assumed that ¹³⁴Cs and ¹³⁷Cs concentrations were each 10 Bq·kg⁻¹ [4]. These assumptions clearly lead to overestimation of ingestion doses. Thus, the first-year internal doses estimated for non-evacuees by UNSCEAR were much higher than those estimated for evacuees by WBC (Table 1). Considering such conservative assumptions, internal doses estimated by WBC could be more reliable. Although estimation from WBC has issues related to the intake scenarios, the acute and chronic intake scenarios were considered both rational and appropriate assumptions for estimating individual internal doses to large populations from a single WBC measurement result [67].

The doses based on the most reliable methodology discussed above can be used to estimate the first-year effective doses in the following way. As examples of two affected areas, Namie Town and Iitate Village, effective doses for the first year can be inferred as follows: as the first four-month doses, the Basic Survey results showed that average doses for residents in the two affected areas were around 4 mSv and 1 mSv for all age groups, respectively [14]. Almost all the residents in these areas were evacuated within the first 4 months after the accident. Thus, subsequent doses for them could be inferred from personal dosimeter measurements conducted in non-evacuated areas [25]. Municipality averages were around 0.5 mSv per 3 months in 2011 even in relatively higher ambient dose areas among non-evacuated areas, such as Date City and Nihonmatsu City. Although there remains the uncertainty as to items discussed in previous section, “interpretation of personal dose measurement results”, the 8-month dose can be roughly inferred as 1.5 mSv from this result. Internal dose due to the acute intake of ¹³⁴Cs and ¹³⁷Cs was mostly below 0.1 mSv even for residents in contaminated areas [48, 49].

Thus, the average values for the first-year effective doses for residents in the two affected areas are not likely to reach 10 mSv, the lower end of doses estimated by the WHO’s first report appearing in 2012. They would be even lower than the district averages of the UNSCEAR estimations (7.8–8.0 mSv for adult evacuees in Iitate Village and 5.0–7.0 mSv for adult evacuees in Namie Town, depending on evacuation routes) [4].

For non-evacuated areas, the maximum municipality-average first-year effective dose could be around 3 mSv (4-month external dose, 1.5 mSv; the subsequent 8-month external dose, 1.5 mSv; internal dose due to ¹³⁴Cs and ¹³⁷Cs, less than 0.1 mSv). The maximum dose of 3 mSv due to external exposure for non-evacuated areas was consistent with that estimated by the UNSCEAR 2013 Report (adults). The municipality-average internal doses for the first year estimated by UNSCEAR (0.94–1.41 mSv, adults in non-evacuated areas) must be overestimated.

Since the accident, individual external dose levels in inhabitant areas have changed due to human factors as well as radioactive decay and weathering. For example, (1) decontamination of soil has been conducted in the inhabitant areas in Fukushima Prefecture and (2) some residents in former evacuation areas have begun to return to their hometown. Also, people’s daily behaviors (occupancy factors) may have changed (or returned to their normal behaviors before the accident) over time. Due to these human factors, individual external doses in a rehabilitation phase cannot be correctly predicted only by model calculations [1, 3, 4] based on deposited radionuclides after the accident. In this respect, monitoring of individual external doses by personal dosimeters is of significance and such results are worthy of publication.

Some of the personal dosimeter/WBC measurements conducted in Fukushima Prefecture (e.g., [23, 24, 50]) have not yet been presented in scientific papers, although residents were notified of their own results and in most cases, all results have been openly available to the residents. These measurements were originally conducted as a healthcare service for the residents. In order to disseminate these data to global scientific communities, however, ethical issues including consensus from residents and municipalities must be overcome.