Temporal variations of 90Sr and 137Cs in atmospheric depositions after the Fukushima Daiichi Nuclear Power Plant accident with long-term observations

Changes in radioactivity in atmospheric depositions after the accident

In March 2011, ¹³⁴Cs was detected with the same activity as that of ¹³⁷Cs. As ¹³⁴Cs had not been detected before the accident except during the emission period resulting from the Chernobyl accident in 1986^11,40,41, the observed ¹³⁴Cs/¹³⁷Cs ratio verified that the only source of ¹³⁴Cs and ¹³⁷Cs was the FDNPP (Supplementary Fig. S2). Our atmospheric aerosol samples indicated that at least three plumes resulting from the FDNPP accident passed across site A (Supplementary Fig. S3). When these plumes arrived at site A, the activities of ⁹⁰Sr and ¹³⁷Cs in atmospheric deposition increased to 2.7 × 10³ and 3.2 × 10⁶ times, respectively, higher than those before the accident (between July 2009 and June 2010) (Fig. 1). The ¹³⁷Cs/⁹⁰Sr activity ratio calculated from our observational results in March 2011 was 4.5 × 10³. This large difference in the rate of increase between ⁹⁰Sr and ¹³⁷Cs reflects the discrepancy between their emission rates, i.e., the total amounts of ⁹⁰Sr and ¹³⁷Cs released were estimated as 0.02 PBq³⁹ and 14.5 PBq²³, respectively. These estimations indicated that the ⁹⁰Sr emission level was much lower than that of ¹³⁷Cs. The monthly ¹³⁷Cs deposition peak due to the FDNPP accident (2.31 × 10⁴ Bq m⁻²) was much higher than those resulting from nuclear weapon tests (548 Bq m⁻²; June 1963) and the Chernobyl accident (131 Bq m⁻²; May 1986) (Fig. 1a). On the other hand, the ⁹⁰Sr activity due to the FDNPP accident (5.2 Bq m⁻²) was lower than that due to the nuclear tests in the 1960s (170 Bq m⁻²; June 1963) (Fig. 1b). For comparison, the average ¹³⁷Cs values in atmospheric depositions before the FDNPP accident (between July 2009 and June 2010) were 7.0 (1.2–22.5) mBq m⁻² at site A and 25.0 (6.1–76.4) mBq m⁻² at site B, while those for ⁹⁰Sr amounted to 1.9 (ranging from not detectable (N.D.)–6.0) mBq m⁻² at site A and 26.0 (6.5–116.8) mBq m⁻² at site B. The possible causes of the higher depositions rates at site B than those at site A are the differences in altitude (site A: 40 m; site B: ~ 1390 m) and local environmental effects (site A: open area; site B: surrounded by forestland).

The activity of ⁹⁰Sr and ¹³⁷Cs in atmospheric depositions and that of ¹³⁷Cs in aerosol samples rapidly decreased after the first surge in March 2011 (Fig. 2 and Supplementary Fig. S3). The decrease rate of radioactivity in atmospheric depositions at site A was due to the change in radionuclide emission, transport, and deposition processes²⁹. We classify the period after the FDNPP accident into three phases. The first phase is dominated by direct emissions (March 2011), the second phase is dominated by tropospheric circulation and removal (from April to December 2011), and the third phase is dominated by resuspension (after January 2012). In the first phase, the direct discharge/emission of radioactive materials during the FDNPP accident and meteorological conditions governed the radionuclide concentration in the environment^29,42,43,44. In the second phase, tropospheric transport of the radioactive materials remaining in the atmosphere after the FDNPP accident and their removal processes dominated atmospheric depositions^17,29. The third phase (after January 2012) mainly depended on the resuspension of radioactive materials^{29,30,31,33,34,35,36}. For comparison, the corresponding decrease rates (first, second, and third phases) resulting from the Chernobyl accident were shorter than those resulting from the FDNPP accident (for more discussion details, please refer to Supplementary Fig. S4 and the text). More discussions regarding the first and second phases were also presented in previous studies^29,34, and hence the scope of the present study is restricted to the third phase.

The latest average monthly ¹³⁷Cs atmospheric depositions in 2018 at sites A and B reached ~ 1/8100 (2.9 Bq m⁻²) and ~ 1/4500 (3.0 Bq m⁻²), respectively, with regard to the peak levels after the accident. But these levels were still ~ 400 and ~ 130 times, respectively, higher than those before the accident (Figs. 1a and 2a, respectively). On the other hand, the ⁹⁰Sr depositions in 2018 amounted to 3.0 (1.2–10.5) mBq m⁻² and 33.8 (3.1–117) mBq m⁻² at sites A and B, respectively (Figs. 1b and 2b, respectively). These ⁹⁰Sr deposition levels were almost at the same level as the preaccident deposition levels, and we concluded that the ⁹⁰Sr deposition levels at our observation sites had returned to the preaccident levels in at least 2015 (Fig. 2b).

Before the FDNPP accident, the ⁹⁰Sr and ¹³⁷Cs activity in atmospheric deposition showed seasonal variations (Fig. 3 and Supplementary Figs. S5 and S6). The ¹³⁷Cs deposition value peaks in spring (April) at site A. On the other hand, it peaks twice in May and September at site B (Supplementary Figs. S4 and S5). Similarly, ⁹⁰Sr deposition reaches peak values during the spring season (March and April) at site A and during the fall season (September and October) at site B (Fig. 3). Studies have suggested that the ⁹⁰Sr and ¹³⁷Cs deposition peaks during the spring season at site A are caused by local and long-range transported dust particles^{14,15,34,45,46}.

After the FDNPP accident, direct emissions and their tropospheric removal processes governed the ⁹⁰Sr and ¹³⁷Cs activity in atmospheric depositions at sites A and B, and seasonal variations were not apparent in the first and second phases (Fig. 2). After 2012 (in the third phase), although the mean ¹³⁷Cs deposition value at site A had slightly increased in spring (peaking in April), no seasonal variations in ¹³⁷Cs at either site were observed (Fig. 2 and Supplementary Figs. S5 and S6). After 2014, in contrast, the seasonal variations in the ⁹⁰Sr radioactivity in atmospheric deposition at both sites showed similar trends to those before the accident (Figs. 2 and 3).

Possible carriers of ⁹⁰Sr and ¹³⁷Cs at sites A and B

The radionuclides in the atmosphere are generally carried by aerosol particles (host particles) emitted through, for example, geochemical and biological cycles. The correlations between dust components (e.g., Al and Fe) and radionuclides (⁹⁰Sr and ¹³⁷Cs) within the collected samples before the accident suggest that mineral dust particles are the dominant carriers of these radionuclides at site A (Fig. 4a). Previous studies have also demonstrated that the sources of these radionuclides are mainly resuspension of contaminated dust originating from long-range transport (large-scale phenomenon) and neighboring areas (local-scale phenomenon)^{14,15,33,34,45,46}. After the accident, chemical analysis results indicate that dust particles are the dominant carriers of ⁹⁰Sr and ¹³⁷Cs at site A, except from 2012 to 2014 for ⁹⁰Sr when the contributions from the accident were high (Fig. 2).

The correlations between the dust components and radionuclides after the accident at site B were poor (Fig. 4b). However, the ⁹⁰Sr activity showed correlations with inorganic salts such as Mg, K, and Ca at site B. Scanning electron microscopy (SEM) observation exhibited the presence of inorganic salt particles including KCl, NaCl, and CaSO₄ in dried deposition samples (Supplementary Fig. S7). Although these salt particles had possibly crystallized during the preparation of the atmospheric deposition samples, it is probable that ⁹⁰Sr coexists with these components in the environment as they are abundantly present in the samples. Studies have indicated that the biological cycle may be a source of these inorganic elements in forests^47,48,49, i.e., the Mg, K, and Ca concentrations in throughfall depositions increase in forests due to foliar leaching. As Sr exhibits a similar geochemical behavior to that of Ca, the occurrence of Sr could be synchronous to that of Ca in the neighboring forest.

Before the accident, the ¹³⁷Cs activity at site B showed positive correlations with major mineral dust components such as Mg, Mn, Ca, K, Fe, and Al (Fig. 4), suggesting that dust particles were the dominant host particles for ¹³⁷Cs. However, no significant correlation was detected between mineral dust and the ¹³⁷Cs activity after 2014. Previous studies have suggested that bioaerosols contribute to the resuspension of ¹³⁷Cs at forest sites in the contaminated area within the evacuation zone in Fukushima Prefecture^35,36. Thus, it is possible that bioaerosols carry ¹³⁷Cs at site B. The differences between the possible carriers may cause the observed differences in the activity ratios of sites B and A (R_B/A) for ⁹⁰Sr and ¹³⁷Cs deposition after the accident (Supplementary Fig. S7).

Estimation of the environmental decrease in ¹³⁷Cs

With the use of regression curve fitting of the activity of ¹³⁷Cs in atmospheric deposition, we estimated its effective half-life due to radioactive decay and environmental removal processes (Fig. 2). We adopted a single-exponential function before the accident from January 1990 to July 2010 and a multiple exponential function after the accident (2012–2018; the resuspension phase). The detailed method of the calculation is described in the Supplementary Information.

The effective half-lives of the short- and long-lived components (t₁ and t₂, respectively) of the ¹³⁷Cs deposition were 195 days and 4.7 years, respectively, at site A, and those at site B were 148 days and 5.9 years, respectively. Interchange of the predominant short- and long-lived components occurred during the period between September and December 2013 (Fig. 2). Our estimation of the effective half-life of the long-lived component at site A is longer than the estimation by the previous study (~ 1.1 years)²⁹ possibly because our estimation 1) excluded the direct emission period and 2) extended observation data by the end of 2018. The effective half-life of the long-lived component of ¹³⁷Cs at site A after the FDNPP accident is shorter than that before the accident (8.5 years). There are two possible reasons for the difference between the effective half-lives before and after the accident. First, the dominant resuspension processes are different before and after the accident. Second, the elapsed time after contaminations is different between the pre and postaccident periods, i.e., more than 30 years had passed for the analysis period before the accident since the last atmospheric nuclear test, on the other hand, only 8 years had passed since the significant pollution after the FDNPP accident.

The above estimated effective half-lives imply that, based on the atmospheric ¹³⁷Cs deposition level, ~ 42 and ~ 48 years will be required from the year of the accident to reach the preaccident level at sites A and B, respectively. These estimates contain uncertainties due to the short observation period compared to the effective half-life before the FDNPP accident (8.5 years). A better understanding of the carriers, resuspension processes, and environmental circulation conditions of radionuclides is needed to confirm the above estimates. The radionuclide decreasing trend may change in the future if resuspension process, biological recycling, and their carriers changed. Finally, our observations only pertain to atmospheric deposition and provide limited features of the environmental radionuclide cycle. The contamination in other fields, such as surface soils, forests, and oceans, will exhibit different effective half-lives. Nevertheless, our continuous observations of the radionuclides in atmospheric deposition before and after the accident enable the evaluation of the atmospheric phases and the changes in various processes to regain the environmental conditions before the nuclear power plant accident.