Cesium-137 deposition and contamination of Japanese soils due to the Fukushima nuclear accident

Abstract

The largest concern on the cesium-137 (¹³⁷Cs) deposition and its soil contamination due to the emission from the Fukushima Daiichi Nuclear Power Plant (NPP) showed up after a massive quake on March 11, 2011. Cesium-137 (¹³⁷Cs) with a half-life of 30.1 y causes the largest concerns because of its deleterious effect on agriculture and stock farming, and, thus, human life for decades. Removal of ¹³⁷Cs contaminated soils or land use limitations in areas where removal is not possible is, therefore, an urgent issue. A challenge lies in the fact that estimates of ¹³⁷Cs emissions from the Fukushima NPP are extremely uncertain, therefore, the distribution of ¹³⁷Cs in the environment is poorly constrained. Here, we estimate total ¹³⁷Cs deposition by integrating daily observations of ¹³⁷Cs deposition in each prefecture in Japan with relative deposition distribution patterns from a Lagrangian particle dispersion model, FLEXPART. We show that ¹³⁷Cs strongly contaminated the soils in large areas of eastern and northeastern Japan, whereas western Japan was sheltered by mountain ranges. The soils around Fukushima NPP and neighboring prefectures have been extensively contaminated with depositions of more than 100,000 and 10,000 MBq km^-2, respectively. Total ¹³⁷Cs depositions over two domains: (i) the Japan Islands and the surrounding ocean (130–150 °E and 30–46 °N) and, (ii) the Japan Islands, were estimated to be more than 5.6 and 1.0 PBq, respectively. We hope our ¹³⁷Cs deposition maps will help to coordinate decontamination efforts and plan regulatory measures in Japan.

A catastrophic earthquake and tsunami occurred on March 11, 2011, which caused destruction in northeastern Japan and severely damaged the Fukushima Daiichi Nuclear Power Plant (NPP). This event led to emissions of radioactive materials from the NPP (1), albeit at unknown and likely strongly varying release rates (1–3). Among these materials, with a half-life of 30.1 y (4), cesium-137 (¹³⁷Cs) causes the largest concerns because of its deleterious effect on agriculture and stock farming, and, thus, human life for decades. Removal of ¹³⁷Cs-contaminated soils or land use limitations in areas where removal is not possible is, therefore, an urgent issue. The Japanese government, general public, and scientists have been waiting for the information of the spatial distributions of ¹³⁷Cs deposition and its soil contamination over all of Japan.

The aerosol-bound ¹³⁷Cs can be removed from the atmosphere and brought to the surface by dry or wet deposition. Analysis of data collected after the Chernobyl accident has shown that ¹³⁷Cs adsorbed in the top soil layer can remain there for many years (5, 6), restricting land use, e.g., for food production, of highly contaminated areas for a long time. To minimize the impacts on human health of soil contamination in Japan due to the Fukushima NPP accident, spatial maps of ¹³⁷Cs deposition and concentrations in soil are urgently needed. Sporadic sampling of the soils in and around Fukushima prefecture has been carried out after the NPP accident under the instruction by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (7) and others (Table S1). However, it is impossible to fully capture the distribution of ¹³⁷Cs deposition across Japan from a limited number of in situ measurements alone. On the other hand, reliable estimates using dispersion models are also not available because of the largely unknown source term. Not only is the total release of ¹³⁷Cs from the damaged NPP poorly known, but also its variation with time is even more uncertain. Although first attempts to estimate it have been made (2) and another study tried to estimate its deposition based on the previous study (2) over the limited areas around Fukushima prefecture with a regional chemical transport model (8), these estimates on the emission rate are highly uncertain and the discussion on deposition over all of Japan has not been made.

In this study we quantitatively estimate the spatial distribution of the ¹³⁷Cs deposition and its soil contamination over all of Japan. We take relative deposition distribution patterns from a Lagrangian particle dispersion model, FLEXPART (Materials and Methods and SI Text) (9), using a constant source term [as assumed also in some simulations because of high uncertainty of the emission amount (10–16)]. We fuse daily varying observations of ¹³⁷Cs deposition in each Japanese prefecture (17) into the modeled deposition fields to obtain quantitative deposition estimates.

Results

Our estimate of ¹³⁷Cs deposition is made for the period between March 20 and April 19 because no observations of ¹³⁷Cs deposition were made between March 12 and 17 and the dispersion model did not simulate any depositions at the observation locations on March 18 and 19 (Fig. S1). Thus, our quantitative estimates do not include the first 8 d after the NPP accident, yet for that period we provide relative contributions to the deposition. In Fig. 1A we show the relative contribution map of the deposition over the period when our estimate was not applicable. It shows that before March 20, potentially contaminated air masses were mainly transported toward the Pacific Ocean and ¹³⁷Cs deposition would have occurred mostly over the ocean, except for Fukushima prefecture and some neighboring provinces. Between March 20 and April 19 (Fig. 1B), a much wider area was effected by the deposition. In particular, eastern and northeastern parts of Japan had a greater potential for ¹³⁷Cs deposition, whereas in western Japan the potential for ¹³⁷Cs deposition was low. Overall, however, the highest potential deposition occurred over the Pacific Ocean, where a few observations of ¹³⁷Cs deposition exist (18), showing that winds were generally quite favorable and carried most radiation away from populated areas.

Fig. 1.

Cesium-137 deposition maps. (A) Relative deposition contributions between March 11 and 19, showing the areas potentially effected by ¹³⁷Cs before the start of measurements. The sums of the depositions during the period were divided by the maximum deposition in the accumulated field. (B) The same as in A, but for March 20–April 19. (C) An example of estimated daily deposition of ¹³⁷Cs on March 21. Squares in black denote the observation locations in each prefecture (Table S2). (D) Daily accumulated rainfall on March 21 by TRMM.

From March 20, we can estimate ¹³⁷Cs deposition fields over Japan more reliably because daily observations have been made in most Japanese prefectures (17) (Table S2). For days from March 20, we create deposition estimates using scaled model values [hereafter called, deposition ratio (DR); see Materials and Methods] from a constant source term simulation in conjunction with the measurements from the MEXT observation network (17). An example of a three-hourly DR animation map is available in Movie S1. For periods when the simulated daily DR value is close to zero yet deposition is observed, a threshold factor is employed in Eq. 2 (Materials and Methods and SI Text). In such cases, a minimum positive deposition ratio value, hereafter called DR threshold (DRT), needs to be used to derive the scaling (Materials and Methods and SI Text). The choice of this DRT is subjective and also effects the estimated deposition amount in our method. After performing comparisons with the observed depositions in each prefecture (SI Text, Fig. S2, and Table S3), we used a DRT of 0.005 for the best guess estimate of daily deposition between March 20 and April 19 (Movie S2) but we also report derived deposition estimates for other DRT values.

The simulated distribution of ¹³⁷Cs deposition is closely linked to precipitation as shown for the case of March 21 (Fig. 1 C and D). The highest deposition values downwind of the NPP are clearly aligned with satellite-observed precipitation by tropical rainfall measuring mission (TRMM, 3B42 V6 product) in a frontal rain band, which causes washout of the radionuclides (Movie S2 and S3). Comparison of daily observed precipitation fields with the estimated deposition maps shows that ¹³⁷Cs deposition is simulated mainly when frontal rain bands pass over Japan (Movie S2). It was reported previously (19) that around 90% of the total deposition of ¹³⁷Cs occurs with precipitation. Thus, the general agreement between observed precipitation and simulated deposition confirms that the model captures the main deposition events, which is also consistent with the discussion from the study by using a regional chemical transport model (8). The daily ¹³⁷Cs deposition was frequently detected at observatories (17) in the eastern and northeastern prefectures of Japan from March 20. Furthermore, large increases of atmospheric radioactivity in the prefectures around Fukushima were observed (20) on March 22 (Fig. S3), probably reflecting ground shine radiation from the radionuclides deposited during the rainfall event on March 21 (Fig. 1 C and D).

Maps of the total ¹³⁷Cs deposition between March 20 and April 19 are shown in Fig. 2A. As a general characteristic, most of the eastern parts of Japan were effected by a total ¹³⁷Cs deposition of more than 1,000 MBq km^-2. Our estimates show that the area around NPP in Fukushima, secondarily effected areas (Miyagi and Ibaraki prefectures), and other effected areas (Iwate, Yamagata, Tochigi, and Chiba prefectures) had ¹³⁷Cs depositions of more than 100,000, 25,000, and 10,000 MBq km^-2, respectively. Airborne and ground-based survey measurements jointly carried out by MEXT and the US Department of Energy (DOE) (21) show high ¹³⁷Cs deposition amounts were observed northwestward and up to a distance of 80 km from Fukushima NPP. It was estimated from the first measurement that by April 29, more than 600,000 MBq km^-2 had been deposited in the area, which is greater than our estimate of less than 500,000 MBq km^-2 (Fig. 2A), yet well within the range of uncertainty of our method (Fig. S4). Furthermore, because no observations on daily deposition were available before March 27 in Fukushima City (Fig. 2B) and no daily deposition observations around NPP, our estimates are expected to underestimate the total deposition in the vicinity of the NPP. In conducting sensitivity studies with DRT, our estimates provide values on the similar order of the MEXT/DOE observations using a DRT value of 0.001 (Fig. S4).

Fig. 2.

Total deposition of ¹³⁷Cs. (A) Gridded total ¹³⁷Cs deposition values for the period March 20–April 19 using our reference DRT value of 0.005. Outputs with 0.2° × 0.2° were interpolated to finer grid using cubic interpolation. Squares in black denote the observation locations in each prefecture (Table S2). (B) Comparisons between total observed depositions for the period March 20–April 19 and estimates at the grid point of each observatory location (Table S2) in the selected prefectures, using different DRT values to derive the scaling factor for the model output. Orange, gray, and black boxes denote no observation (Miyagi) and missing observations (Yamagata, between March 29 and April 3; Fukushima, before March 27 and April 4), respectively.

Using an approximate relationship between ¹³⁷Cs deposition and its topsoil concentration (22) [conversion coefficient (CC) of 53 ± 15 kg m^-2] (Fig. S5), we converted the estimated depositions into topsoil concentrations (Fig. 3, Movie S4, and Table S4).

Fig. 3.

The ¹³⁷Cs concentration in soil. We used DRT of 0.001 and CC of 53 kg m^-2. Outputs with 0.2° × 0.2° were interpolated to finer resolution using cubic interpolation. The Merged IBCAO/ETOPO5 Global Topographic Data Product (25) was used to mask out ocean area below 0 m above sea level (a.s.l.).

We compared ¹³⁷Cs concentrations in the topsoil derived from both observed ¹³⁷Cs deposition values (17) and from our estimates, with direct measurements of ¹³⁷Cs concentrations in soils and grasses with a soil-to-grass transfer factor of 0.13 (23) (SI Text and Table S1) (Fig. 4). The MEXT deposition-based soil contamination tends to be lower than the soil- and grass-based samplings because of the latter including the time period just after the NPP accident. Our scaled model deposition fields agree well with both the point measurements including the MEXT areal surveys and other available observations (Fig. 4), generally falling in the range between the two techniques. If the soil contaminations closer to the Fukushima NPP are included, the observed soil contamination has a wider range of 190–310,000 Bq kg^-1 with mean and median values of 20,575 and 5,750 Bq kg^-1, respectively. The observations in Fukushima City give a lower range of 620–21,000 Bq kg^-1 (mean, 5,969 Bq kg^-1; median, 4,200 Bq kg^-1). Our estimated distributions were within the range of the observations around the NPP, but close to the lower bound (Fig. 3 and Fig. 4A). From the sporadic observations (7, 21) (Table S1) and our estimate in this study, Fukushima prefecture as a whole is highly contaminated, especially to the northwest of the NPP, with soil ¹³⁷Cs concentrations above 1,000 Bq kg^-1 throughout the east prefecture from our estimate (Fig. 3). Neighboring prefectures are less contaminated, but ¹³⁷Cs concentrations there are still above 250 Bq kg^-1 in most of the areas. Thus our approach provides a necessary estimate of deposition beyond the immediate area of the Fukushima prefecture, where monitoring activities are less intensive.

Fig. 4.

Atmospheric, soil, and grass observation-based Cs-137 concentrations, and estimates based on the scaled model output and for different DRT values used for the scaling. (A) Comparisons in northern prefectures. Aomori and Miyagi prefectures had no ¹³⁷Cs detections on the daily deposition data and no measurements, respectively. The minimum value in Yamagata prefecture is no detection and no lower error bar is shown. (B) The same as in A, but around Kanto area. Lower and upper error bars denote minimum and maximum concentrations using CC of 68 and 38 kg m^-2 based on Fig. S5, respectively. Orange, gray, and black boxes denote no observation (Miyagi) and missing observations (Yamagata, between March 29 and April 3; Fukushima, before March 27 and April 4), respectively. A soil-to-grass transfer factor of 0.13 (23) was used to convert grass to soil contamination. For Fukushima prefecture, only the soil observations in Fukushima City were used, excluding other observations close to the Fukushima NPP. The data source for the comparisons are summarized in Table S1.

Discussion

There are many important agricultural regions in Japan. In Japan, the limit for the sum of ¹³⁴Cs and ¹³⁷Cs concentrations (as total cesium) in soil is 5,000 Bq kg^-1 under the Food Sanitation Law (24). Considering that about half (2,500 Bq kg^-1) of the total radioactive cesium deposition is due to ¹³⁷Cs, the east Fukushima prefecture exceeded this limit and some neighboring prefectures such as Miyagi, Tochigi, and Ibaraki are partially close to the limit under our upper bound estimate (Movie S4) and, therefore, local-scale exceedance is likely given the strong spatial variability of ¹³⁷Cs deposition. For those three prefectures, detailed soil sampling is recommended in the near future. Estimated and observed contaminations in the western parts of Japan were not as serious, even though some prefectures were likely effected to some extent (Fig. 3, Movie S4, and Table S4). Concentrations in these areas are below 25 Bq kg^-1, which is far below the threshold for farming. However, we strongly recommend each prefecture to quickly carry out some supplementary soil samplings at city levels to validate our estimates even if the concentrations are low.

The relatively low contamination levels over western Japan can be well explained by the Japanese topography. The eastern and northeastern parts of Japan are surrounded by mountain ranges such as the Kanto, Echigo, and Ohwu mountain ranges (Fig. S6) (25), which, to a large extent, sheltered the northwestern and western parts of Japan from the dispersion of radioactive material. It is worth noting, however, that relatively higher contamination levels can be seen over the Hida, Chugoku, and Shikoku mountain ranges (Fig. 3, Fig. S6, and Movie S4), probably due to orographic enhancement of precipitation and, thus, wet deposition of ¹³⁷Cs. In Hokkaido, to the north of Japan’s main island, both lower altitude and higher altitudes such as the Yubari and Hidaka mountain ranges are effected by ¹³⁷Cs deposition, partially due to direct transport from the Fukushima NPP via the Pacific Ocean as shown in Movies S1 and S2 and also as simulated by another atmospheric transport model (12).

We estimate that a total of more than 5.6 and 1.0 PBq ¹³⁷Cs were deposited over Japan and the surrounding ocean (130–150 °E and 30–46 °N), and the Japan Islands in this domain only, respectively (Fig. 2A). Although the estimate for the larger domain is quite uncertain because it is constrained only by measurements in Japan, these numbers are consistent with a suspected total release of about 12 PBq ¹³⁷Cs (2). Most of the deposition occurred over the Pacific Ocean, yet soil concentrations of ¹³⁷Cs are above 100 Bq kg^-1 over large areas of eastern Japan (Fig. 3). According to our results, food production in eastern Fukushima prefecture is likely severely impaired by the ¹³⁷Cs loads of more than 2,500 Bq kg^-1 (upper limit of farming) and also partially impacted in neighboring provinces such as Iwate, Miyagi, Yamagata, Niigata, Tochigi, Ibaraki, and Chiba, where values of more than 250 Bq kg^-1 cannot be excluded (Fig. 3 and Movie S4). Notice also that our estimates are based on a transport model driven with meteorological analysis data from a global model. Such a model cannot fully capture all complexities of the regional wind field over Japan and, in particular, does not resolve the high spatiotemporal variability of precipitation. Therefore, we expect the true soil contamination across Japan to be considerably more variable than in our estimate. Even in regions where we find relatively low soil contamination levels, hot spots with high concentrations (e.g., due to convective rain fall, orographic enhancement of rainfall, or fine-grain soil flow by rainwater on the ground) may be possible. In contrast, relatively clean patches may also be present in areas with high overall contamination levels. Despite these shortcomings, we expect our results to be useful for regulatory measures and for guiding monitoring activities toward areas with expected high ¹³⁷Cs burdens. We hope this study will contribute to understanding the contamination issue in Japan.

Materials and Methods

Observations of Cesium-137 Deposition and Concentration in Soil in each Prefecture.

From March 18, MEXT has been observing daily radioactivity levels in deposition in most of the prefecture (17). The exact coordinates of the sampling locations were individually accessible through our contacts to MEXT (Table S2). The deposition data between March 18 and 19 were not used in our estimate because of no depositions at observatories from the modeled DR maps as mentioned in the main text. In some prefectures, data were missing or unavailable [Miyagi, March 18–April 19 (completely no observations); Yamagata, March 29–April 3; Fukushima, March 18–March 26 and April 4; Gifu, March 24, 25, 27, 28, and 30; Nara, March 18–21 and April 15–18; Oita, March 22–26).

FLEXPART and Estimated ¹³⁷Cs Deposition.

FLEXPART (9) is a Lagrangian particle dispersion model simulating transport, diffusion, dry and wet deposition, and radioactive decay of radioactive materials such as ¹³¹I, ¹³⁷Cs, and ¹³³Xe (See http://transport.nilu.no for further details on FLEXPART). In this study, continuous emission from the Fukushima Daiichi NPP was assumed after 1800 hours coordinated universal time (UTC) on March 11, 2011. The simulation ended at 0000 hours UTC on April 20. FLEXPART was forced with the European Center for Medium-Range Weather Forecasts (ECMWF) operational analysis data with a global resolution of 1° × 1° and 0.18° × 0.18° for 120–168 °E and 25–50 °N. The output had a resolution of 0.2° × 0.2° and was recorded every 3 h (SI Text).

For each day, we first normalized the modeled daily accumulated deposition in each grid cell with the maximum accumulated deposition value for the model domain, hereafter called daily deposition ratio (DDR) maps:

[1]

where FPD_(x,y)i is the three-hourly modeled deposition in grid cell (x,y) and FPD_max is the maximum daily deposition value found in the entire model domain. T is the number of model output timesteps per day (T = 8). Daily gridded deposition values of ¹³⁷Cs were estimated by scaling the DDR map with available daily observed ¹³⁷Cs depositions in each prefecture by MEXT (17) by the following equation:

[2]

where Depo_(x,y) is the estimated daily total ¹³⁷Cs deposition in grid cell (x,y), Depo_i(Obs.) is the observed ¹³⁷Cs deposition at location i (Table S2), N ≤ 47 is the number of available counts on a certain day in Japan’s 47 prefectures, DDR_i(Obs.Loc.) is the DDR_(x,y) in the grid point where ¹³⁷Cs deposition was observed, and DDR_(x,y) is the DDR in grid cell (x,y). Only the cases with both the observed deposition and the DDR_i(Obs.Loc.) not equal to zero at each observatory location were used for counting N on each day. Because the Depo_i(Obs.) to DDR_i(Obs.Loc.) scaling factor in Eq. 2 becomes infinite when the simulated DDR value is close to zero but deposition is actually observed, a minimum positive DDR value, DRT, needs to be used to derive the scaling. Several DRTs of 0.001, 0.005, 0.007, 0.01, 0.05, and 0.1 for DDR_i(Obs.Loc.) within the simulation domain on each day were used to avoid abnormally high Depo_(x,y) values due to dividing by small values (SI Text). If DDR_i(Obs.Loc.) at a certain grid point was less than a DRT value, DDR_i(Obs.Loc.) was set to the DRT value.

For computing total ¹³⁷Cs deposition between March 20 and April 19, we corrected all values to April 19 using a half-life of ¹³⁷Cs of 30.1 y (4). The sum of all the daily observed or estimated ¹³⁷Cs depositions is the total ¹³⁷Cs deposition (Fig. 2 and Fig. S4A).

Observations on ¹³⁷Cs Concentrations in Soil and Grass.

For comparison with our estimates, measurements of ¹³⁷Cs concentrations in soil or grass were used (SI Text and Table S1). Mean transfer factor of soil-to-grass of 0.13, which was obtained from the observations in Japanese soil and grass, was used to convert grass contamination to soil equivalent contamination (grass contamination divided by the transfer factor) (23). The times and locations of those samplings varied. To cover the time period of our study (March 20–April 19), we also used some soil samples from later dates, but we did not use any data after May 19. Notice also that the soil samples were also effected by ¹³⁷Cs deposition before March 20 (SI Text). Some observatories measured total cesium concentration including both ¹³⁷Cs and ¹³⁴Cs. In that case, we assumed that half of the total Cs was ¹³⁷Cs.

To convert the ¹³⁷Cs deposition into soil concentration, soil depth and density information are needed. However, it is currently difficult to obtain this information across all of Japan. There is an empirical relationship on the ratio between ¹³⁷Cs concentration and deposition from 0 to 5 cm soil, paddy soil, and field soil samples (22) (Fig. S5). We considered the mean value of the ratio as CC of 53 ± 15 kg m^-2 reflecting the 5-cm depth soil information and its density. Our estimated CC value is close to the CC value of 65 kg m^-2 assumed by MEXT (26) with 5-cm soil and a soil density of 1,300 kg m^-3. Dividing our estimated deposition (MBq km^-2 = Bq m^-2) by the CCs, we empirically obtained the mean ¹³⁷Cs concentration in soil (Bq kg^-1).