Sinking velocity of particulate radiocesium in the northwestern North Pacific

Abstract

Sinking particles (SP) were collected by time series sediment traps at two depths in the northwestern Pacific before and after the Fukushima Daiichi Nuclear Power Plant accident, and accident‐derived particulate radiocesium was measured. Radiocesium (¹³⁷Cs) was first detected at 500 m (4810 m) about 2 weeks (1 month) after the accident. ¹³⁷Cs of SP collected over 1 year revealed that the time lag between two depths was larger than that for the first ¹³⁷Cs detection (about 2 weeks). We estimated the transient sinking velocity (SV) from the cumulative temporal ¹³⁷Cs flux and the time lags at the two depths. Although the SV of SP collected in very early period was large, the estimated SV of most particulate ¹³⁷Cs (about 80%) was about 50 m d⁻¹. Based on comparison of ¹³⁷Cs concentration in total SP with that in SP without organic materials, we suspect that most of the ¹³⁷Cs was likely incorporated into aluminosilicates.

1 Introduction

The Tokyo Electric Power Company‐Fukushima Daiichi Nuclear Power Plant (FNPP1) was seriously damaged by the 2011 Tohoku‐Oki earthquake and tsunami on 11 March 2011, and as a consequence, large quantities of artificial radionuclides, including radiocesium, were emitted. Oceanographic observations 1 month after the FNPP1 accident combined with a numerical simulation revealed that the release of contaminated water and eolian dust resulted in the rapid dispersal of FNPP1‐derived radiocesium to a broad area of the northwestern North Pacific [Honda et al., 2012, 2013, and references therein]. Most of radiocesium transported to the ocean exists in dissolved form [e.g., Whitehead et al., 1988], but a part of the radiocesium is likely to be absorbed by or adsorbed onto particulate materials, including living creatures, sinking particles (SP), and seafloor sediments. At the time of the FNPP1 accident, the Japan Agency for Marine‐Earth Science and Technology Center (JAMSTEC) had already deployed a time series sediment trap at station K2 in the northwestern North Pacific to collect SP. Using measurements of radiocesium in SP collected before and after the accident (between March and June 2011), Honda et al. [2013] reported that FNPP1‐derived radiocesium reached the deep sea (4810 m depth) at K2 within a month after the accident and that only a small proportion (~0.5%) of radiocesium supplied to upper ocean was transported to the deep layer in particulate form. In addition, from the time lag between the first detection of radiocesium at 500 m and that at 4810 m, they estimated the sinking velocity of particulate radiocesium to be >180 m d⁻¹. SP continued to be collected and the radiocesium to be measured at K2 from July 2011 to June 2012. Here using these additional data, we report estimates of the sinking velocity in the second phase of the postaccident stage and discuss the potential chemical form of particulate radiocesium.

2 Methods

At time series station K2 (47°N, 160°E; water depth, 5200 m, in the North Pacific western subarctic gyre), biogeochemical observations have been conducted since 2001. The horizontal distance from FNPP1 to station K2 is about 1870 km. Time series sediment traps (McLane 7G‐21) were deployed at 500 m and 4810 m at K2 during two periods: from November 2010 to June 2011 (first period), and from July 2011 to June 2012 (second period). The traps were deployed or recovered during three scientific cruises of the R/V Mirai (JAMSTEC): MR10‐06 (October 2010), MR11‐05 (June–August 2011), and MR12‐05 (July 2012). During the first period, SP were collected with interval of 12 days, and during the second period, they were collected with interval of 16 days. Before the traps were deployed, the collecting cups were filled with a seawater‐based buffered 10% formalin solution, used as a preservative.

In the laboratory, SP collected from the traps were sieved through a 1 mm plastic mesh, and then the filtrate was dried and pulverized. Radiocesium and major chemical components (organic materials (OM), biogenic opal, CaCO₃, and aluminosilicate) in the SP were measured by using gamma ray spectrometry, an elemental analyzer, and an inductively coupled plasma atomic emission spectrometry as described previously [Honda et al., 2013]. To estimate the trapping efficiency, radioactivity of ²¹⁰Pb (hereinafter ²¹⁰Pb concentration) in SP was measured simultaneously by gamma spectrometry. The counting time varied from 1 day to several days. The counting error was about 12% for 500 m sample and 3% for 4810 m sample on average. To determine whether ¹³⁷Cs was preferentially incorporated into organic materials or to other components, ¹³⁷Cs was also measured in SP after removal of the OM. After the concentration of ¹³⁷Cs had been measured in the total SP sample, the sample was boiled in a 10% hydrogen peroxide (H₂O₂) solution for about 30 min at 120°C on a hot plate to eliminate the OM, then filtered, dried, and weighed. The mean concentration (weight %) of OM in the SP was estimated to be 13 ± 1% (Table 1), which corresponds well to that estimated previously with an elemental analyzer [Honda et al., 2013].

Table 1. Mean Concentrations and Fluxes of ¹³⁷Cs in Total Samples and in Samples After Removal of Organic Materialsa^a The analytical error is based on 1 sigma of counting statistics.

Total Sample					Sample Without OM
Sample Name	Trap Depth (m)	Sampling Period	Sample Weight (mg)	^137Cs (Bq g^‐1)	Sample Weight (mg)	^137Cs (Bq g^−1)	Concentration OM (%)	^137Cs in OM (%)	^137Cs in Others (%)
K2‐4810 m 1‐9	4810	5 Jun 2011 to 17 Jun 2011	85.0	0.285 ± 0.020	73.6	0.292 ± 0.021	13.4	11.1 ± 8.9	88.9 ± 8.9
K2‐500 m 1‐3	500	25 Mar 2011 to 6 Apr 2011	86.6	0.397 ± 0.026	75.9	0.485 ± 0.030	12.4	−7.1 ± 9.7	107.1 ± 9.7

• ^a The analytical error is based on 1 sigma of counting statistics.

3 Results

Honda et al. [2013] previously reported the fluxes and concentrations of radiocesium (¹³⁴Cs and ¹³⁷Cs) and the major chemical components in SP collected during the first period.

The radiocesium isotope ¹³⁴Cs was practically not present in this region just before the FNPP1 accident, so it would potentially be a good short‐term tracer of FNPP1‐derived radionuclides. However, owing to its short half‐life (2.06 years), the small SP sample volume, and the relatively low sensitivity of gamma ray spectrometry to this isotope, no ¹³⁴Cs was detected during the major period of the second collection phase. Although some ¹³⁷Cs was present in the ocean just before the accident, in the SP collected before the accident its concentration was below the detection limit. Thus, we assumed that any detectable ¹³⁷Cs in SP was FNPP1‐derived, and in this report we present and discuss only ¹³⁷Cs.

3.1 Total Mass Flux and the Chemical Composition of SP

The total mass flux (TMF) and the chemical composition of SP at 500 m and 4810 m changed over time (Figure 1). At 500 m, TMF began to increase in late May 2011, and a TMF peak, composed mainly of biogenic opal, was observed in July 2011 (Figure 1a). Thereafter, TMF decreased toward spring 2012, although large TMF spikes, composed dominantly of CaCO₃, were observed in late September 2011 and in late December 2011/early January 2012. In March 2012, TMF again began to increase, peaking in April 2012. This spring SP peak was composed primarily of biogenic opal.

Figure 1

Total mass flux and chemical composition at (a) 500 m and (b) 4810 m at station K2. OM, OP, CA, and LM are organic matter, biogenic opal, CaCO₃, and lithogenic material (aluminosilicates), respectively.

At 4810 m, TMF began to increase in June 2011 (Figure 1b), after a month‐long hiatus, and it reached a maximum in late November 2011. Although biogenic opal was dominant during July and September, CaCO₃ became dominant in SP after October. In late November, TMF began to decrease gradually, and it was small from early February to mid‐March 2012. Thereafter, TMF again increased, peaking in late April 2012. At this time, biogenic opal was the dominant component. It is noteworthy that annual peak at 4810 m in 2011 appeared in late autumn while annual peak usually appears in late spring and summer this region [e.g., Honda et al., 2009; Honda and Watanabe, 2010]. Based on almost decadal time series observation of SP this region, late peak appeared in 1998, 2008, and 2011 (M. C. Honda, unpublished data, 2013). These periods seems to overlap negative phase of Nino 3.4 index. However, its relation has not been analyzed statistically.

The small TMF increases observed between March and April 2011 and the large increases observed between March and May 2012 were approximately synchronous at 500 m and 4810 m, with a small time lag. Between August and early September 2011, TMF decreased greatly at 500 m (Figure 1a). During this period, the water depth of the top of the mooring system was deepened by about 15 m relative to its normal depth (210 m) [MIRAI MR12‐02 Preliminary Cruise Report, 2012]. Thus, it is possible that stronger current or internal wave lowered trapping efficiency of the 500 m sediment trap. If so, then the TMF at 500 m would have actually been larger than that observed during this period. Therefore, if a consecutive increase in TMF from May to late September 2011 at 500 m was presumed, it would be synchronous with the observed increase from June to November at 4810 m, with a time lag of at most two sampling intervals (i.e., 32 days). Thus, if the time lag of the flux variability between the two measurement depths can be considered to reflect the sinking velocity, the sinking velocity between the two measurement depths can be estimated as >134 m d⁻¹ (4310 m/32 days).

Between November 2010 and May 2012, ²¹⁰Pb concentration in SP at depths of 500 and 4810 m was 0.07–1.7 and 1.2–12.2 Bq g⁻¹, respectively, and the average fluxes of excess ²¹⁰Pb were estimated to be about 50 and 400 mBq m⁻² d⁻¹, respectively.

3.2 ¹³⁷Cs Flux and Concentration

As reported by Honda et al. [2013], ¹³⁷Cs was detected for the first time at 500 m in SP collected between 25 March and 6 April 2011 (Figure 2a). In that sample, the ¹³⁷Cs flux was at its maximum of about 31 mBq m⁻² d⁻¹ during the entire first observation period (March 2011 to June 2012). Thereafter, the ¹³⁷Cs flux gradually decreased until September 2011, except for a large temporary decrease in early May 2011. After October 2011, the ¹³⁷Cs flux was small or not detectable, and it dropped below the detection limit for good in February 2012. The total ¹³⁷Cs flux was about 2.25 Bq m⁻² at 500 m (Table S1 in the supporting information). At 4810 m, ¹³⁷Cs was initially detected in SP collected between 6 April and 18 April 2011, 12 days after it was first detected at 500 m (Figure 2b). The ¹³⁷Cs flux was about 5 mBq m⁻² d⁻¹, and it remained almost constant until early June 2011. Between 5 June and 17 June 2011, the ¹³⁷Cs flux increased suddenly by about 17 mBq m⁻² d⁻¹, after which it again remained almost constant until early October 2011. From mid‐October 2011, the ¹³⁷Cs flux gradually decreased until April 2012, and from late April, it was below the detection limit. The total ¹³⁷Cs flux at 4810 m was about 2.61 Bq m⁻², about 20% larger than that at 500 m. Honda et al. [2013] estimated the input of eolian ¹³⁷Cs derived from the FNPP1 accident at K2 in April 2011 from the ¹³⁷Cs concentrations in surface seawater and at the mixed layer depth to be about 450 Bq m⁻². Thus, removal rate of ¹³⁷Cs from surface layer is estimated to be 0.5–0.6% yr⁻¹ (2.25–2.61 Bq m⁻² yr⁻¹/450 Bq m⁻² × 100). This removal rate is slightly lower than that estimated previously (0.5–1.5% yr⁻¹, Honda et al. [2013]). Buesseler et al. [1987, 1990] also reported that removal rate was declined with time. This is attributed to that most of radiocesium is scavenged to sinking particle in early stage while radiocesium input takes place in only the early stage, i.e., average ¹³⁷Cs flux decrease with time while ¹³⁷Cs input is constant.

Figure 2

¹³⁷Cs flux at (a) 500 m and (b) 4810 m at station K2. (c) ¹³⁷Cs concentrations at 500 m (open circles) and 4810 m (closed circles) at K2. Arrows indicate 11 March 2011, when the Tohoku‐oki earthquake occurred.

At 500 m, the ¹³⁷Cs concentration in the SP sample in which ¹³⁷Cs was first detected (25 March and 6 April 2011) was the largest during the entire observation period (about 0.41 Bq g⁻¹, Figure 2c), and the concentration decreased thereafter until summer 2011. After August 2011, the ¹³⁷Cs concentration remained less than about 0.03 Bq g⁻¹, and after February 2012, it was below the detection limit, when the flux was also undetectable. At 4810 m, in contrast, the ¹³⁷Cs concentration increased from the time it was first detectable in SP and reached its maximum of about 0.33 Bq g⁻¹ in SP collected between 24 May and 5 June 2011. Thereafter, the concentration generally decreased, with small fluctuations, until October 2011. After October, the concentration remained less than 0.05 Bq g⁻¹, but it continued to be detectable until mid‐April 2012.

4 Discussion

4.1 New Sinking Velocity Estimate

Honda et al. [2013] estimated the sinking velocity of SP (particulate radiocesium) to be >180 m d⁻¹ from the distance (4310 m) between the two traps and the time lag (24 days at maximum) between the first detection of radiocesium at each trap. This estimate is comparable to previously reported values based on time series sediment trap observations [e.g., Berelson, 2002; Honda et al., 2009]. However, after a longer observation period, the time lag between the two depths apparently increased. The highest ¹³⁷Cs concentration at 500 m was observed in SP collected between 25 March and 6 April 2011, whereas the highest ¹³⁷Cs concentration at 4810 m was observed in SP collected between 24 May and 5 June 2011. In addition, at 500 m, ¹³⁷Cs was below the detection limit after February 2012, whereas at 4810 m, ¹³⁷Cs could be detected until mid‐April 2012. On the basis of these observations, we estimated the time lag between the two depths to be about 60 days and the sinking velocity to be about 70 m d⁻¹ (4310 m/60 days). This result suggests that the sinking velocity of most particulate radiocesium was considerably smaller than the previous estimate of >180 m d⁻¹.

We therefore used the following method to obtain another estimate of the sinking velocity of particulate radiocesium. First, we assumed that the total ¹³⁷Cs flux at 500 m, which was about 20% smaller than that at 4810 m, was an artifact caused by low trapping efficiency at 500 m, and that there had been no lateral ¹³⁷Cs input to the 4810 m sediment trap. This presumption of low trapping efficiency by the 500 m sediment trap was supported by the observed excess ²¹⁰Pb flux in SP, on average about 50 mBq m⁻² d⁻¹ at 500 m and 400 mBq m⁻² d⁻¹ at 4810 m. We estimated the potential ²¹⁰Pb flux in the water column of the northern North Pacific by using data for ²¹⁰Pb and ²²⁶Ra collected at Geochemical Ocean Sections Study station 217 (46°36′N, 176°50′W) [Chung and Craig, 1980, 1983]. We assumed the deposition rate of ²¹⁰Pb from the atmosphere to be 5000 dpm (disintegrations per minute) m⁻² y⁻¹ [Harada and Tsunogai, 1986]. Then, from the assumed eolian ²¹⁰Pb input and the ²¹⁰Pb produced in the water column, we estimated that the excess ²¹⁰Pb fluxes at 500 and 4810 m should be about 210 and 940 mBq m⁻² d⁻¹, respectively. By applying these estimates to the observed ²¹⁰Pb fluxes at station K2, we estimated the trapping efficiency of the sediment traps at 500 m and 4810 m to be about 24% (50/210 mBq m⁻² d⁻¹ × 100) and 43% (400/940 mBq m⁻² d⁻¹ × 100), respectively. Though these trapping efficiency estimates have large uncertainties, taking also into account for the observation that 500 m sediment trap experienced to be deepened temporally during deployment (see section 3.1), we believe that our assumption of a relatively lower trapping efficiency at 500 m than at 4810 m is qualitatively reasonable.

First, to estimate fluxes during the hiatuses from 25 October to 28 December 2011 at 500 m and from 17 June to 5 July 2011 at 4810 m, we interpolated the data collected before and after those dates. Then we calculated the relative ¹³⁷Cs fluxes to the total ¹³⁷Cs flux, including the interpolated data, for each sampling interval and each depth (Figure 3, solid lines), and cumulated relative ¹³⁷Cs fluxes from 25 March 2011 (Figure 3, broken lines). Note that we averaged the relative ¹³⁷Cs flux over each sampling interval under the assumption that the relative ¹³⁷Cs flux increased linearly from the beginning to the end of each sampling interval. For instances, from these calculations, we estimated that the cumulative ¹³⁷Cs flux at 500 m reached 5% of the total ¹³⁷Cs flux about 4.2 days after 25 March 2011, whereas at 4810 m, the cumulative ¹³⁷Cs flux reached 5% about 42.5 days after 25 March 2011. Thus, we estimated the time lag for the first 5% of the total ¹³⁷Cs flux between the two depths as 38.3 days. In addition, we estimated that 50% of the total ¹³⁷Cs flux was reached about 76.4 days after 25 March 2011 at 500 m and about 148.5 days after 25 March 2011 at 4810 m. We therefore estimated the time lag for the first 50% of the total ¹³⁷Cs flux to be 72.1 days.

Figure 3

Relative ¹³⁷Cs fluxes at 500 m (thin solid line) and 4810 m (thick solid line) at station K2, and the cumulative ¹³⁷Cs flux at 500 m (thin broken line) and 4810 m (thick broken line) at K2. The hatching indicates interpolated data (see text).

From time lags for respective fractions of the total ¹³⁷Cs flux between the two depths, we estimated the sinking velocity of each fraction of total particulate radiocesium (Figure 4). We thus estimated the sinking velocities of the first 1%, 3%, and 5% fractions of the total ¹³⁷Cs flux to be about 257, 156, and 113 m d⁻¹, respectively. The sinking velocity continued to decrease as the fraction of the total increased, reaching about 66 m d⁻¹ for the first 10%. Once the first 20% of the total ¹³⁷Cs flux (about 3 months after 25 March 2011 at 4810 m) was reached, the sinking velocity remained relatively stable with only small fluctuations (50 ± 10 m d⁻¹). For 100% of the total ¹³⁷Cs flux, we estimated the sinking velocity (i.e., the average sinking velocity of the particulate ¹³⁷Cs flux) to be about 54 m d⁻¹. This estimated sinking velocity for SP in the deep sea is considerably smaller than the sinking velocity we reported previously [Honda et al., 2013]. Kaneyasu et al. [2012] reported that most of radioactively contaminated eolian dust was sulfate based on land observation. If this is the case, eolian radiocesium should be easily dissolved in the ocean. One potential scenario about decrease of sinking velocity with time is that, in the early stage just after ¹³⁷Cs was supplied to K2 as eolian dust and ¹³⁷Cs concentration in upper seawater was high, even big particles with high sinking velocity and with low specific surface might have an opportunity to meet ambient dissolved radiocesium and can scavenge these. On the other hand, in the later stage when ¹³⁷Cs concentration became low due to diffusion and advection, only small particle with low sinking velocity and with high specific surface might be able to scavenge ambient diluted radiocesium. Another possibility is the early eolian ¹³⁷Cs was insoluble and fast sinking large particle. Adachi et al. [2013] reported that, just after the accident, atmospheric radioactivity increased twice (14–15 March and 20–21 March 2011) at Tsukuba located about 170 km south‐southwest of the FNPP1. Based on analysis using an imaging plate, a scanning electron microscope, and an energy dispersive X‐ray spectrometer, they found that radioactively contaminated eolian dust in the first peak was relatively large spherical particle (diameter: about 2.5 µm) including zinc and iron while eolian dust in the second peak was relatively small (diameter: about 0.5 µm) and mainly sulfate as reported by Kaneyasu et al. [2012]. Although K2 is about 2000 km away from the FNPP1, long distance transport of ¹³⁷Cs‐bearing large particle might be supported by the fact that, in the early stage (middle and late March 2011), radiocesium was detected from large particle and ¹³⁷Cs‐bearing particle size decreased with time in European countries [Masson et al., 2013].

Figure 4

Sinking velocity for respective fraction of particulate ¹³⁷Cs flux. Sinking velocity is estimated to be time lag between arrival times of each fraction of total ¹³⁷Cs flux at both depths.

4.2 Potential Chemical Form of Cs in SP

Because our estimate of the sinking velocity of particulate ¹³⁷Cs was smaller than that estimated by an ordinary “peak to peak” comparison of major components observed in time series sediment traps, the chemical form of such slowly sinking particulate ¹³⁷Cs is of interest. In terrestrial sediments, ¹³⁷Cs is mainly adsorbed onto aluminosilicates [e.g., Qin et al., 2012], and in seafloor sediments, ¹³⁷Cs is also irreversibly adsorbed onto aluminosilicates [Otosaka and Kobayashi, 2012]. In contrast, Ono et al. [2013] reported higher ¹³⁷Cs concentrations in OM than in aluminosilicates in seafloor sediment, suggesting that ¹³⁷Cs may be preferentially incorporated into organic material.

To examine whether ¹³⁷Cs was incorporated into organic material or aluminosilicates in SP, we measured ¹³⁷Cs concentrations in SP after removal of the OM and found that they were higher than those in total SP (Table 1). We estimated the fraction of ¹³⁷Cs in OM to be only 1.8 ± 4.4% on average. Although the analytical error was large because of the small sample size, this result indicates that most ¹³⁷Cs was very likely incorporated not into OM but into other components of the SP (biogenic opal, CaCO₃, or aluminosilicate). We suspect that most of ¹³⁷Cs are incorporated to aluminosilicate as reported by Otosaka and Kobayashi [2012].

If so, it is noteworthy that maximum radiocesium concentrations (~0.35 Bq g⁻¹) of SP are as high as those in seafloor sediment of the coastal regions near Fukushima [e.g., Kusakabe et al., 2013; Otosaka and Kobayashi, 2013] although aluminosilicates are a minor fraction in SP (10–30%). It might be attributed to that their Cs concentration is average of upper 3 cm seafloor sediment, and thus, higher Cs concentration near surface seafloor sediment is diluted. This interpretation is supported by higher Cs concentration observed upper 1 cm of seafloor sediment near Fukushima [Otosaka and Kato, 2014].

Acknowledgments

We deeply appreciate anonymous reviewers for their constructive comments. This work was partially supported by a grant‐in‐aid for Scientific research on Innovative Areas from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (KAKENHI), Interdisciplinary Study on Environmental Transfer of radionuclides from the Fukushima Daiichi NPP Accident (#24110004).

The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.