Deep-Sea Research I 80 (2013) 37–46

Contents lists available at SciVerse ScienceDirect

Deep-Sea Research I

0967-06http://d

n CorrE-m

vincent

journal homepage: www.elsevier.com/locate/dsri

Multi-decadal projections of surface and interior pathwaysof the Fukushima Cesium-137 radioactive plume

Vincent Rossi a,b,n, Erik Van Sebille b,c, Alexander Sen Gupta b,c, Véronique Garçon d,Matthew H. England b,c

a IFISC (Institute for Cross-Disciplinary Physics and Complex Systems), CSIC-UIB, Palma de Mallorca, 07122 Spainb Climate Change Research Centre, University of New South Wales, Sydney 2052, Australiac ARC Centre of Excellence for Climate System Science, University of New South Wales, Sydney 2052, Australiad Laboratoire d’Etude en Géophysique et Océanographie Spatiales, CNRS/UPS/IRD/CNES, 14 av. E. Belin, Toulouse 31400, France

a r t i c l e i n f o

Article history:Received 8 February 2013Received in revised form16 May 2013Accepted 29 May 2013Available online 15 June 2013

Keywords:Radioactive tracersNorth PacificOcean circulationMode water formationFukushima nuclear disaster3D Lagrangian modeling

37/$ - see front matter & 2013 Elsevier Ltd. Ax.doi.org/10.1016/j.dsr.2013.05.015

esponding author. Tel.: +34 971 172 915.ail addresses: vincent.rossi.ocean@gmail.com,@ifisc.uib-csic.es.(V. Rossi)

a b s t r a c t

Following the March 2011 Fukushima disaster, large amounts of water contaminated with radionuclides,including Cesium-137, were released into the Pacific Ocean. With a half-life of 30.1 years, Cs-137 has thepotential to travel large distances within the ocean. Using an ensemble of regional eddy-resolvingsimulations, this study investigates the long-term ventilation pathways of the leaked Cs-137 in the NorthPacific Ocean. The simulations suggest that the contaminated plume would have been rapidly dilutedbelow 10,000 Bq/m3 by the energetic Kuroshio Current and Kurushio Extension by July 2011. Based onour source function of 22 Bq/m3, which sits at the upper range of the published estimates, waters withCs-137 concentrations 410 Bq/m3 are projected to reach the northwestern American coast and theHawaiian archipelago by early 2014. Driven by quasi-zonal oceanic jets, shelf waters north of 451Nexperience Cs-137 levels of 10–30 Bq/m3 between 2014 and 2020, while the Californian coast is projectedto see lower concentrations (10–20 Bq/m3) slightly later (2016–2025). This late but prolonged exposure isrelated to subsurface pathways of mode waters, where Cs-137 is subducted toward the subtropics beforebeing upwelled from deeper sources along the southern Californian coast. The model suggests thatFukushima-derived Cs-137 will penetrate the interior ocean and spread to other oceanic basins over thenext two decades and beyond. The sensitivity of our results to uncertainties in the source function and tointer-annual to multi-decadal variability is discussed.

& 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The Tohoku earthquake and the associated tsunami on 11March 2011 caused enormous human losses and infrastructuredamage. It also triggered a nuclear disaster in which the Fukush-ima Daiichi nuclear plant released large amounts of radionuclidesinto the atmosphere and ocean (Buesseler et al., 2011; Chino et al.,2011; Stohl et al., 2011), including Cesium‐137 (Cs-137). Recentstudies suggest that the Fukushima disaster caused the largest-ever direct release of anthropogenic radionuclides into the ocean(Bailly du Bois et al., 2012). Although several attempts have beenmade to quantify the total amount of Cs-137 released (Buesseleret al., 2012 and references therein), considerable uncertaintyremains related to the duration and intensity of release. Earlyreports found a strong impact of the contaminated water on localmarine life (Garnier-Laplace et al., 2011), while a recent study

ll rights reserved.

performed off Japan suggested that radiation risks due to Csisotopes had dropped below the levels generally consideredharmful to marine animals and humans within three months ofthe accident (Buesseler et al., 2012).

However, the poorly known long-term effects of these radio-nuclides on marine biota (Garnier-Laplace et al., 2011) combinedwith the unknown factors relating to the magnitude of release andthe redistribution of Cs-137 via oceanic circulation prevent a robustassessment of the potential impacts. Behrens et al. (2012) havestudied the evolution of the Cs-137 plume in the North Pacific forthe first decade following the release using passive tracer calcula-tions in the NEMO ocean circulation model. Recently, Nakano andPovinec (2012) used a coarse resolution model (21) with climato-logical forcing to study the long-term dispersal of the radioactiveplume. Here, we use Lagrangian particles within the eddy-permitting Ocean General Circulation Model For the Earth Simu-lator (OFES) to focus on multi-decadal (up to 30 years) evolution,confirming some of the results reported by Behrens et al. (2012) andNakano and Povinec (2012) and analyzing further the surface andsub-surface pathways for an extended period of time.

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–4638

Cs-137 is of purely anthropogenic origin and is released into theatmosphere and the ocean via nuclear weapons testing, nuclearplants releases, nuclear accidents, and the dumping of radioactivewastes. Even though Cs-137 has only been present in the environ-ment for about six decades, it has already penetrated to the deepocean (Livingston and Povinec, 2000). Background oceanic con-centrations vary between 10−4 and 10−1 Bq/m3 in the southernhemisphere to a maximum of a few Bq/m3 in the North Pacific(Aoyama et al., 2011). Since Cs-137 behaves as a passive tracer inseawater and has a 30.1 years half-life, it can aptly be used to studythe decadal-scale circulation pathways and ventilation time-scalesof water masses in the global ocean (Aoyama et al., 2011 andreferences therein). Hence, the future distribution of Fukushima-derived Cs-137 in the ocean provides an opportunity to validateoceanic numerical models and their parameters over decadaltime-scales (Dietze and Kriest, 2011), provided the source functioncan be tightly constrained.

Here, we focus our analysis on the simulated long-term andlarge-scale evolution of Cs-137 in the North Pacific as a result ofthe Fukushima radioactive incident, without re-evaluating theinitial near-coastal sub-annual spreading, neither the releasescenario, as it has been pursued in previous studies (Buesseleret al., 2012; Kawamura et al., 2011; Tsumune et al., 2012; Estournelet al., 2012). Using an ensemble member mean distribution from27 Lagrangian regional simulations, the aim of this study is toestimate the dilution and spreading pathways of the Cs-137radioactive plume in the North Pacific over the next few decades.In addition we examine the location, timing and concentrations ofCs-137 contaminated waters reaching selected coastal areas in theNorth Pacific. The vertical evolution of the Cs-137 tracer is alsoanalyzed along specific sections across the basin providing a fullthree dimensional analysis of the decadal pathways.

2. Methods

To simulate the spatial evolution of Cs-137 in the North PacificOcean, we advect a large number of neutrally buoyant Lagrangianparticles from a region off Fukushima using 3D velocities from aglobal Oceanic Global Circulation Model. The Lagrangian techniquehas been shown to be well suited to problems in which highcontaminant gradients are involved, since spurious numericaldiffusion is not introduced (Nakano et al., 2010; Monte et al.,2009). Cs-137 is considered as a conservative dissolved tracer asscavenging, sedimentation, re-suspension from the sediment,biological uptake (or bioaccumulation) and food-chain transfer(or biomagnification) are negligible. This assumption is supportedby laboratory studies that demonstrate this inert behavior (Vogeland Fisher, 2010; IAEA, 2004; Heldal et al., 2001). Given the largeuncertainties involved in atmospheric transport patterns andestimates of total release, atmospheric deposition at the oceansurface has not been taken into account here (Buesseler et al.,2012; Stohl et al., 2011; Yasunari et al., 2011). Similarly, river inputsand background pre-Fukushima concentrations of Cs-137 havebeen neglected, with initial concentrations set to zero. Thesimulations thus only incorporate the material newly releasedinto seawater from the Fukushima disaster (see also the discussionsection).

The numerical model used to derive the Lagrangian advectionfields is the Ocean model For the Earth Simulator (OFES) based onthe MOM3 ocean model (Masumoto et al., 2004; Sasaki et al.,2008). OFES is initialized from the World Ocean Atlas 1998climatological fields, and then run for 57 years using NCEP forcing.It is an eddy-resolving global ocean model with a horizontalresolution of 1/101 and 54 vertical layers. The fine resolution ofthe Lagrangian Cs-137 simulations is a key feature of this study,

since mesoscale eddies have considerable importance in thetransport of Cs-137 (Behrens et al., 2012; Tsumune et al., 2011).Furthermore, the OFES model has been comprehensively assessedin the Kuroshio region and the Northern Pacific: the outputs usedhere (taken from years 1980 to 2006) have been shown toadequately capture the seasonal, interannual and multi-decadalvariability of the surface and sub-surface circulation in this region(Taguchi et al., 2010, 2007; Nonaka et al., 2012; Sasaki et al., 2012).OFES has also been validated against mode water formation andsubduction rates in the North Pacific (Aoki et al., 2007; Qu andChen, 2009) as well as against the deep limb of the Atlanticmeridional overturning circulation (Van Sebille et al., 2011).

In our approach, the Lagrangian particle concentrations can bescaled to a Cs-137 concentration that gives an estimate of the totalquantity of tracer introduced from Fukushima, as done in anEulerian study by Hazell and England (2003). Both approacheshave the advantage of being adaptable to any new estimates oftotal release and allows for comparison with future in-situ mea-surements when adding background oceanic concentrations.Latest estimates suggest a major release of 22 (712) PBq of Cs-137 over one month starting mid-March (Bailly du Bois et al.,2012). Although some regional studies estimates lower amounts(Tsumune et al., 2012; Estournel et al., 2012), an analysis based onboth in-situ data and numerical simulations confirmed that thesource term used here (22 PBq over 1 month, see also discussionsection) is the most probable one (Buesseler et al., 2012).

To simulate the release of waters contaminated with dissolvedCs-137, an ensemble of experiments is performed, in each of whicha total of 100,000 passive Lagrangian particles are released fromMarch 13 to April 13. Specifically, we release 10,000 particles atthe surface ocean off the Fukushima plant every 3 days over thismonth, i.e. during the first month following the Fukushimadisaster. A non-local source function is used by choosing randomlythe spatial location of release of each particle from a two-dimensional Gaussian distribution that is centered on the locationof the Fukushima plant and has a decay length-scale of 30 km,mimicking the mixing effect of non-resolved near-coastal currents(Behrens et al., 2012). In order to assess the sensitivity of the tracerevolution to the initial oceanic conditions, an ensemble of 27release experiments is performed, each simulation starting in adifferent year (from 1980 to 2006).

Particles are advected with a 4th order Runge-Kutta integrationscheme, using the Connectivity Modeling System described inParis et al. (2013), for 30 years using 3-day snapshots of OFESvelocity data for the North Pacific region. As is typical in thesekinds of multi-decadal basin-scale Lagrangian experiments ineddy-resolving models, there is no need for an additional randomwalk term, as mesoscale eddy variability accounts for an effectivehorizontal and vertical diffusion of the particles (see also thediscussion section).

In order to integrate the particles trajectories for a 30-yearperiod, velocity fields are looped, so that after 2006 the integrationcontinues using the 1980 fields. Similarly to what was done in VanSebille et al. (2012), we tested the impact of looping the velocityfield in this way by comparing a standard looped simulation (i.e.from 1980 to 2006 and back) with a short loop simulation (1980–1990, looped over these 10 years), and no significant difference inthe tracer evolution was found (not shown). While the simulationsproject forward in time by three decades based on the circulationfields of 1980–2006, the variability across the ensemble set gives ameasure of uncertainty into the future under the assumption thatocean circulation variability over the next few decades will not besubstantially different from that seen over the past few decades.This is a reasonable assumption given that over this time periodchanges in circulation related to natural variability are likelyto dominate over any potential changes driven by global warming

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–46 39

(e.g. (Hawkins and Sutton, 2009)). The integration of a particle isstopped when it leaves the North Pacific domain (0–601N and1201E–1001W). Rather than adjusting the scaling of the radio-activity carried by the particles over time, we have chosen here tosimulate the natural radioactive decay of the Cs-137 isotope byrandomly removing particles over time with a probability corre-sponding to a half-life of 30.1 years. This allows for an interpreta-tion of the set of particles as a subsample of the Cs-137 atoms, andkeeps us clear from the issue of mixing that is inherent in theinterpretation where each particle is a finite-size water mass (seealso p. 66 of Van Sebille (2009)).

Throughout this study, we present the 27-member ensemble-mean distribution (i.e. corresponding to a total of 27�100,000Lagrangian trajectories) and highlight the robust pathways asso-ciated with the ocean circulation. Where applicable, an examina-tion of the spread in the distributions provides information on thesensitivity of the pathways and time-scales to interannual varia-tions of the circulation.

3. Results

3.1. Surface pathways through the North Pacific

Fig. 1 shows snapshots of Cs-137 concentrations at differenttimes in the surface ocean (0–200 m averages) and the corre-sponding concentrations along 37.51N (which approximates thecore of the eastward plume of Cs-137). Along the zonal section(Fig. 1e), the spread in concentrations associated with different

Fig. 1. Surface (0–200 m) Cs-137 concentrations (Bq/m3). (a) In April 2012, (b) April 2014(black dashed line in panels a–d). Error bars in (e) represent the standard deviation over tthe black line along the North American west coast represent our areas of interest. Whitevarious locations (for interpretation of the colors in this figure, the reader is referred to

ensemble members is relatively small (less than 10% of theconcentrations reported in Fig. 1e). This suggests that at largescales, the pathways are relatively insensitive to the initial oceanconditions and interannual-decadal variability in the circulation.Indeed, a recent modeling study by Behrens et al. (2012) foundthat the impact of the initial ocean condition would largely fadeafter the first 2–3 years. As such, the lack of real-time circulationfields during the initial release is unlikely to significantly affect ourresults on interannual time-scales and beyond.

The plume is quickly advected away almost zonally from Japandue to the vigorous Kuroshio Current and Kuroshio Extension as itpasses over the Izu-Ogasawara Ridge (∼1421E). Within this path-way there is a rapid dilution of Cs-137 with time. Concentrationsfall below 10,000 Bq/m3 everywhere by the end of July 2011, froma maximum concentration of about 150,000 Bq/m3 near theJapanese coast at the end of March 2011, in good agreement withcoastal studies (Buesseler et al., 2011; Tsumune et al., 2012). Mostof this decay is related to the intense eddy field of the KuroshioCurrent and Kuroshio Extension that leads to strong horizontaland vertical mixing. Further east, the Kuroshio Extension encoun-ters successive topographic features such as the Shatsky Rise(1601E) and the Emperor Seamounts (1701E), which have beenassociated with meandering of the main flow (Niiler et al., 2003).It then rejoins the interior gyre circulation as the North PacificCurrent (NPC), which advects the plume eastward towards thecoast of North America. The NPC is not a uniform eastward flowbut is punctuated by zonal fronts where intensified flow occurs atrelatively stable latitudes, despite strong seasonal and interannualvariations in forcing (e.g. Oka and Qiu, 2012). The strong eastward

(c) April 2016, (d) April 2021 and (e) along 37.51N at the latitude of Fukushima planthe ensemble of 27 simulations. The black square around the Hawaii archipelago andflow vectors represent an illustrative sense of the large-scale surface circulation atthe web version of this article).

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–4640

transport at 451N in April 2012 (Fig. 1a) might correspond to aquasi-stationary jet associated with the Polar Front, sometimescoinciding with the Subarctic front (Oka and Qiu, 2012). Furthersouth, an intensified eastward advection is also visible at ∼32.51N,at the location of the Kuroshio Extension jet (Qiu and Chen, 2005).Note that although most of the surface plume is contained in thesubpolar gyre, Cs-137 is gradually transported into the subtropicalgyre, likely due to the mesoscale activity of the Kuroshio Extensionjet and to the Ekman transport under the subpolar westerly winds(Fig. 1a and b). After reaching the American continental marginaround April 2014 at 451N, the ensemble-mean of the contami-nated water plume then splits into two branches. One branchflows northwards to feed the Alaska Current, thereby entrainingpart of the contaminated waters into the Subarctic Gyre. The otherbranch flows southwards feeding the California Current. South of401N, the offshore Ekman drift at the coast associated with coastalupwelling helps to prevent Cs-137 enriched surface waters fromdirectly reaching the coast. The persistent upwelling off Californiaforced by the prevailing northeasterly winds maintains Cs-137-contaminated waters slightly offshore the shelf for upto 5 yearsafter release (Fig. 1b and c).

In the western Pacific, the North Equatorial Current recirculatesCs-137 enriched water back westward at around 301N (Fig. 1b).After 10 years (Fig. 1d) contaminated surface waters have beendiluted almost everywhere to concentrations below 10 Bq/m3,with a noticeable local maximum in the surface Eastern Subtropi-cal North Pacific (15–351N and 135–1201W).

When the westward-flowing North Equatorial Current entersthe Philippines Sea, the long-term pathways become more com-plex. Part of the water flows northward and re-enters the KuroshioCurrent again, closing the recirculation of contaminated waterwithin the subtropical gyre. Much of the remainder moves south-west and exits the model domain. This occurs as water is entrainedinto the southward-flowing Mindanao Current south of 151N, andthen either through the Celebes Sea to feed the IndonesianThroughflow, or toward the Equator to enter the South Pacificvia the Equatorial Undercurrent (Nakano et al., 2010). Anotheralternative route is via the South China Sea north of 151N and thenvia the Indonesian Throughflow to reach the Indian Ocean. InAoyama et al. (2011), a core of high concentration Cs-137 wasfound in the eastern part of the Indian Ocean which was believedto have traveled via the Indonesian Throughflow after originatingin the Pacific region 10–20 years earlier by atmospheric depositionin the 1970s. Evaluating the relative importance of these exitroutes is beyond the scope of this study but may constitute animportant pathway of contaminated water over the next fewdecades.

After 30 years (in 2041) about 25% of the total initial Cs-137released remains within the North Pacific gyre, 25% has left thegyre, and the remaining 50% has disappeared by natural radio-active decay. Of the amount that left the North Pacific, a very smallfraction has been lost northward toward the Bering Sea, with thelargest part injected into the southern hemisphere through thedifferent routes discussed above and in Tsumune et al. (2011),Nakano et al. (2010).

3.2. Intrusion of the plume onto coastal areas of the Eastern NorthPacific

To investigate the penetration of contaminated water onto theNorth American shelf, Cs-137 concentrations (0–200 m) in a coastalband parallel to the coastline are examined. Note that the intra-ensemble spread for the selected latitudes, which accounts foruncertainty with regards to interannual variability in the circulation,is relatively small (upto 20% in Fig. 2b, see also the possible changesdue to multi-decadal variability in the discussion section). The first

waters characterized by Cs-137 levels above 1 Bq/m3 are projected toreach the US west coast at around 451N in mid-2013, while rapidlyincreasing upto 5 Bq/m3 from early 2014 (Fig. 2a and b). Maximumconcentrations between 20 and 30 Bq/m3 occur from mid-2014 (3years after the initial release) to mid-2018 off Oregon/British Colum-bia (45–551N), in good agreement with the simulations by Behrenset al. (2012). Cs-137 concentrations above 10 Bq/m3 in surface shelfwaters near Vancouver (491N) are modeled from 2014 to mid-2021,i.e. over �6 years. The early and strong rise in Cs-137 concentrationsseems related to the Subarctic/Polar front jet that rapidly advectswaters toward the shelf (Oka and Qiu, 2012). The first Cs-137contaminated waters start to reach the Californian coast from mid-2014, with local maxima at 28–321N (Baja California/San Diego) and35–401N (Northern California). Maximum concentrations are lowerthan in the north, ranging from 10 to 20 Bq/m3, due to longer transittimes associated to larger dilution and natural radioactive decay.Concentrations above 10 Bq/m3 near the Californian coast occur from2016, some 2 years after the northern coast (Fig. 2b), as coastaldivergence in the Californian upwelling prevents contaminated sur-face waters from reaching the coast. While the maximum concentra-tions off California are weaker than those north of 401N, elevated Cs-137 concentrations remain evident from 2016 until around 2025,about 3 years longer than along the northern coastline (Fig. 2a). Whilethe northern areas are mostly affected by the first impact of theplume, the southern areas experience prolonged input probably dueto the persistent upwelling of contaminated waters that originatefrom the subsurface.

In the centre of the North Pacific, the evolution of the Cs-137plume evolution is studied by averaging surface concentrationsover the area 18–231N and 160–1551W, covering all islands of theHawaiian Archipelago (Fig. 2b). Slightly contaminated waters startto reach the area by 2013, but Cs-137 concentrations above 5 Bq/m3

only arrive in early 2015 and remain above this concentration untilearly 2025. Maximum concentrations of about 8 Bq/m3 occur frommid-2017 to mid-2021. The archipelago is not situated along the mainpathway of the plume and so is only affected by water thatrecirculates or mixes southward (Fig. 1a–c).

3.3. Vertical distribution of Cs-137 and interior pathways

As the contaminated water spreads laterally across the NorthPacific basin, there is also substantial vertical transport (Figs. 3–6).By 2016 (Fig. 3), i.e. 5 years after the initial release, about 42% oftotal remaining Cs-137 is at depths of 200–600 m, with 5% atgreater depths (600–1500 m). By 2031, Cs-137 has penetrated toconsiderable depth with only about 28% of the remaining Cs-137in surface layers, 48% in central water masses (200–600 m depth)and 22% within intermediate waters (600–1500 m depth), indicat-ing that Fukushima derived Cs-137 is likely to be found in Modeand Intermediate Waters in the future. However, because there isno significant Deep-Water formation in the North Pacific, only avery small proportion, less than 3%, penetrates deeper than1500 m by the year 2030.

Diapycnal mixing and vertical water mass movements, forexample within late winter pycnostads, are likely to be the mainfactors driving the deep intrusion of Fukushima derived Cs-137,favored by the period of release; namely, at the end of winterwhen mixed layers in the North Pacific are relatively deep (upto200–300 m). Using a Cs-137 labeled water mass analysis, Aoyamaet al. (2008) showed that Central Mode Waters and North PacificSubtropical Mode Waters are formed by subduction of surfacewaters in the subarctic regions of the western North Pacific beforebeing transported into the ocean interior to lower latitudes. On thewestern side of the basin, the model simulates intense subductionand vertical mixing of Cs-137 enriched waters already within thefirst 2–4 years of release (2013–2014). The subduction area extends

Fig. 2. Surface (0–200 m) Cs-137 concentrations (Bq/m3) along selected continental shelves. (a) Hovmöller diagram (latitude versus time) of Cs-137 concentrations (Bq/m3)along the North American western continental shelves (concentrations are averaged over a ∼300 km wide coastal band as shown by the black line in Fig. 1. (b) Cs-137concentrations at 301N, 491N (indicated by the two dotted white lines on panel a) and around the Hawaii archipelago (see the black square in Fig. 1a–d), error barsrepresenting the standard deviation over the ensemble simulations (for interpretation of the colors in this figure, the reader is referred to the web version of this article).

Fig. 3. Vertical distribution of Cs-137 over time in the North Pacific basin. “Surface”waters are defined from 0 to 200 m, “Central” from 200 to 600 m, “Intermediate” from 600to 1500 m and “Deep” below 1500 m. Percentage is given over the total remaining part of Cs-137, taking into account its natural decay (for interpretation of the colors in thisfigure, the reader is referred to the web version of this article).

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–46 41

from 30 to 451N/1451E–1601W (Fig. 4a) and reaches to depths of∼1000 m (Fig. 4b). This appears to be associated with the forma-tion of three distinct water masses: Denser Central Mode Water,Lighter Central Mode Water and Subtropical Mode Water, asdocumented by Oka and Qiu (2012). The southernmost (25–301N) region of subduction, related to the North Pacific SubtropicalMode Water formation (s0∼25 kg/m3), is not likely to entrainelevated Cs-137 waters, since the Fukushima release occurredfurther north. However, subduction of both Denser (s0∼27 kg/m3

within 37.5–451N) and Lighter (s0�26 kg/m3 within 30–37.51N)Central Mode Waters (Fig. 4a) must entrain high Cs-137

concentrations into the deep ocean (Oka and Qiu, 2012; Aoyamaet al., 2008). This is also evidenced by the asymmetrical bowlshape structure along 1651E (Fig. 4c) that can be compared to theCs-137 profile from in-situ data collected in 2002 by Aoyama et al.(2008) (Fig. 1). These observed spatial patterns are well repro-duced by the model, suggesting that the mechanisms driving theelevated deep concentrations are realistically simulated. Althoughthe meridional slope of the stratification appears almost symme-trical north and south of 301N, Cs-137 penetrates down to�1000 m between 30 and 451N (Lighter and Denser Central ModeWater) but only down to ∼400 m between 20 and 301N (North

Fig. 4. (a) Subsurface (400–600 m) Cs-137 concentrations (Bq/m3) in April 2014. (b–d) Vertical sections of Cs-137 concentrations (Bq/m3) in April 2014 at 301N (b), along1651E (c) and 1401W (d). Black contours in (a) represent the mean depth (in m) of the s0¼26 kg/m3 isopycnal, while the mean stratification is shown in (b–d), as derivedfrom the OFES ensemble. The white dotted lines indicate the position of the respective sections (for interpretation of the colors in this figure, the reader is referred to the webversion of this article).

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–4642

Pacific Subtropical Mode Waters), possibly due to the Kuroshio andKuroshio Extension that act as barriers to the southward penetra-tion of the tracer (Behrens et al., 2012).

Further east and in April 2014, the two regions of elevatedlevels of Cs-137 (20–60 Bq/m3) centered at 321N and 451N (Fig. 4dat 1401W) correspond to the two zonal jets mentioned above,flowing toward North America and extending down to 400 m and500 m respectively. In early 2014 at 301N (Fig. 4b), the coastalupwelling along the Californian coast is clear and prevents thesurface plume from penetrating onto the shelf. By early 2021,relatively high Cs-137 concentrations (∼10 Bq/m3) are foundbetween 400 and 600 m in the east, spreading south of 401N(20–401N/1801E–1201W; Fig. 5a). These mode waters were sub-ducted in the western North Pacific along the main pathway of theplume about 6–9 years earlier and have gradually been upwelledalong isopycnals while moving eastward. The mean depth of the26 kg/m3 isopycnal (black contours in Figs. 4 and 5) reveals thezonal tilting of the stratification in the eastern half of the NorthPacific basin. Lighter Central Mode Waters, and Subtropical ModeWaters in moderate proportion, were formed between 300 and600 m in the west (Fig. 4a–c) and are then uplifted to 150–250 mto the southeast of the basin (Fig. 5a). When these mode watersapproach the continental shelf, it leads to higher subsurface (150–250 m) concentrations off the south California upwelling region in2021 (south of 351N; Fig. 5b and c).

Although these specific three dimensional pathways below andabove ∼401N might solely explain why the shelf regions of NorthAmerica are affected differently depending on latitude by theradioactive plume (Fig. 2a), local processes might also play a role.When investigating the vertical circulation at two extreme loca-tions, representing the southern and northern shelf regions of theCalifornian Upwelling system, we found a marked difference in themean origin of upwelled waters. Source waters of upwelling alongthe northern shelves are somehow shallower (∼50–150 m) thanalong the southern areas (∼100–250 m; Fig. 6), primarily due to alower offshore Ekman transport (Rossi et al., 2009). Therefore, dueto a combination of remote and local effects, Cs-137 enriched

mode waters upwelled by Ekman pumping intrude onto theSouthern US shelves over a longer period of time (∼9 years,Fig. 2a and b) than over the Northern US shelves.

4. Discussion

4.1. Long-term evolution of the Cs-137 plume

Using an ensemble of Lagrangian particle simulations, thisstudy has examined the future pathways and concentration levelsof Cs-137 in the North Pacific resulting from the 2011 Fukushimanuclear disaster. Given the considerable uncertainties in all thesource terms, our approach considers only the direct oceanicsource but remains adaptable to any future improved estimatesof the source function. In addition, the OFES velocity field usedhere is currently the most realistic eddy-resolving resolution of theNorth Pacific circulation, thus providing reasonable confidence inour predictions of the long-term fate of the Fukushima Cs-137. TheKuroshio Current, Kuroshio Extension and their associated mesos-cale activity causes a rapid dilution and offshore transport ofcontaminated waters, with maximum concentrations droppingby an order of magnitude or more within the first 4 months, ingood agreement with Buesseler et al. (2012). Our model suggeststhat the first Fukushima derived Cs-137 contaminated waters(41 Bq/m3) are expected to reach Hawaii and the North Americanshelf in early 2014. A recent study by Kameník et al. (2013)indicated that the radiation plume was not detectable at the mainHawaiian Islands in September 2012, in good agreement with oursimulated results which project the arrival for 2014–2015 (Fig. 2b),providing additional confidence in the “ensemble” approach andthe model velocity fields used here. On the eastern boundary ofthe North Pacific, Cs-137 levels ranging between 5 and 30 Bq/m3

are found along the coastline and remain elevated for 5–10 yearsdepending on latitude. The highest Cs-137 concentrations (10–30 Bq/m3) are simulated along the northwest American continen-tal shelf (north of 401N) from early 2014, persisting for about

Fig. 5. Subsurface concentrations of Cs-137 (Bq/m3) in April 2021 (a) at 400–600 m and (b) at 150–250 m and (c) along a vertical section at 301N (white dotted rectanglesshow the horizontal sections in (a and b). Black contours in (b) represent the annual mean depth (in m) of the s0¼26 kg/m3, isopycnal, as derived from the OFES ensemble.The white dotted lines indicate the position of the respective sections (for interpretation of the colors in this figure, the reader is referred to the web version of this article).

Fig. 6. Original depth (m) of upwelled particles 1 year before reaching the surface shelf waters within the two black boxes at the coast represented in panel 5(b) (28–301Nand 48–501N). Bars are the ensemble mean distributions, error lines indicate the standard deviation of the distributions as obtained from the 27 ensemble members (forinterpretation of the colors in this figure, the reader is referred to the web version of this article).

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–46 43

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–4644

6 years. In contrast, maximum concentrations are projected to belower (10–20 Bq/m3) from 2016 along the Californian coast, butthese can be expected to prevail for a longer period of time (∼9years). This seems to be due to the subduction of Cs-137 enrichedmode waters along the path of the plume in the western NorthPacific, which is subsequently upwelled from below the thermo-cline along the south Californian upwelling. These transit-timesacross the Pacific (3–5 years after release) are consistent with theresults of Nakano and Povinec (2012) who suggest that the plumewillreach the US west coast about 4–5 years after the accident. Thisestimate is, however, 1–2 years earlier than that estimated by Behrenset al. (2012), likely due to the different velocity fields and sourcefunctions used in each study. Our simulations suggest that Cs-137 willpenetrate mode and intermediate waters over the next 30 years, inline with the results of Behrens et al. (2012) and Nakano and Povinec(2012). Overall, our modeled pathways in the first 10 years are in goodagreement with the findings of Behrens et al. (2012), although theyused a different source function. Note that some differences remain inthe simulated concentrations and timing. For example, Moreover, thetime taken for the plume to reach the Hawaiian Archipelago (∼3–5years) is very similar in both studies, while Cs-137 concentrationssimulated by Behrens et al. (2012) are about half that in ourprojections, a result of the lower source function used in their study(10 PBq). Our simulated long-term pathways reveal some differenceswith the results of Nakano and Povinec (2012), such as the strength ofthe southern recirculation and the absence of small-scale jets in theirsimulations. However, their approach is based around a coarseresolution (21) ocean model (such models tend to have a moresluggish circulation compared to high resolution models such as theone used here) and climatological velocity fields, which average outmany of the highly energetic features that can effectively transporttracers. It is further noted that while Behrens et al. (2012) focusmainly on the vertically integrated tracer inventory projected over thenext 10 years, here we have assessed both surface and interiorpathways for the next three decades using high resolution modeloutputs.

Finally, our simulations suggest that after 30 years about 25% ofthe initial Cs-137 release is likely to exit the North Pacific towardother oceanic basins, with most exiting via the IndonesianThroughflow and/or crossing the Equator to join the South Pacific;only a small proportion leaves via the Bering Strait into the Arctic,a pathway also simulated by Behrens et al. (2012). This finding isconsistent with the analysis of interbasin transport of the pre-viously released Cs-137 undertaken by Tsumune et al. (2011), whofind that the North Pacific has been a source of Cs-137 to otherocean basins, in particular via the Indonesian Archipelago towardthe Indian Ocean. This is also in line with the global simulationsperformed by Nakano and Povinec (2012), which indicate that theFukushima-derived Cs-137 will be transported to the South Pacificand the Indian Ocean after 20 years.

4.2. Neglected processes and uncertainties

A number of assumptions are made in the present analysis thatcould affect the evolution of oceanic Cs-137. In these multi-decadalbasin-scale Lagrangian experiments using an eddy-resolvingmodel, we have assumed that there is no need for an additionalrandom walk term to parameterize the effects of mesoscale eddyactivity via an effective horizontal and vertical diffusion of theparticles. However, a sensitivity test using additional vertical andhorizontal diffusion terms (via a randomwalk, with the horizontaldiffusion coefficient set to 100 m2 s−1 and the vertical diffusion setto 10−5 m2 s−1) confirms that the effect on Cs-137 concentrations inthe core of the plume is relatively small (∼18%, not shown). Incontrast, a larger percentage impact is found on the southern flankof the plume, suggesting an enhancement of the southern

recirculation loop after the first year, while concerning only smallCs-137 concentrations (o10 Bq m−3).

Another potential simplification is the assumption that Cs-137 actsas a passive conservative tracer, apart from its radioactive decay.Alternative modeling approaches which consider the interactions ofthe dissolved nucleotide with the solid phase do exist and imply thatthe suspended matter, the settling of particles and the deposition/erosion of the bottom sediment may be important (e.g. Periañez,2000; Periañez and Elliott, 2002; Kobayashi et al., 2007; Choi et al.,2013). By neglecting scavenging and associated settling/depositionprocesses, our surface estimates are potentially overestimated, whilethe deep penetration of Cs-137 might be underestimated. Theproportion of Cs-137 released to the open ocean versus the amountof radioactive material deposited into the sediment near the coast (upto 10% or less of the total release) is also relatively sensitive to localconditions such as vertical mixing and dispersion over the shelf (Choiet al., 2013). Note, however, that scavenging and sediment deposition/re-suspension processes are unlikely to be important away fromcoastal regions (e.g. in the oligotrophic gyre of the North Pacific), dueto the low concentrations of particulate matter and the distance to theseabed in such areas.

We have also neglected land/ocean interactions, in particularinputs of contaminated water from rivers and groundwater. Althoughrivers transport additional Cs-137 into the ocean, especially duringheavy rainfall events (Nagao et al., 2013), the total estimate of thesesource terms remains unknown. In addition, rivers are unlikely tomake up more than a small contribution to dissolved Cs-137 in theopen ocean as the isotopes bind to soil particles (Tsumune et al.,2011). A recent study by Kusakabe et al. (2013) showed that theCs-137 distribution in coastal sediments is not directly linked to theproximity between river mouth locations and the site of the accident,but is instead driven by the concentration of the overlying waterduring the first few months after the accident, and by the physicalcharacteristics of the sediment.

A larger source of uncertainty relates to the atmospheric falloutonto the surface ocean. Most of the airborne Cs-137 released fromFukushima is thought to have been deposited over the North PacificOcean due to the prevailing westerly winds during March–April 2011(Stohl et al., 2011; Yasunari et al., 2011), which would lead to higherlevels of Cs-137 in surface waters than those calculated here.Buesseler et al. (2012) found that the most consistent atmosphericfallout over the Pacific Ocean matches the highest total atmosphericemission available (estimates range from about 9.9 to 35.8 PBq, seeTable S2 of Buesseler et al. (2012)). Indeed, Stohl et al. (2011) showedthat about 80% of the total 35.8 PBq (23.3–50.1 PBq) were depositedover the Pacific Ocean, although the timing and the detailed extensionof the deposition remains unclear. Another estimate by Kawamuraet al. (2011), also used by Nakano and Povinec (2012), suggests thatonly a small fraction (5 PBq) of their total estimate of atmosphericrelease (13 PBq) was deposited over the North–West Pacific Ocean,while the rest was deposited over land. If a significant proportion ofthe atmospheric deposition occurred in the area of mode waterformation (30–451N and 140–1701E), this would lead to an increaseof the Cs-137 penetration into the deep ocean, due to subsequentsubduction during late winter/early spring (Oka and Qiu, 2012). Inaddition, the choice of a non-local input area (using a circle ofdiameter 60 km) might somehow partly capture the effect of theatmospheric deposition in the immediate vicinity of Fukushima(Behrens et al., 2012). However, since the estimates derived frommodeling studies are highly variable, and because there is a lack ofdirect in-situ measurements (Tsumune et al., 2013), it is difficult toconstrain the pattern and magnitude of the atmospheric fallout. Inaddition, a recent large-scale study (Nakano and Povinec, 2012)confirmed that due to a large dispersion surface, direct radionuclidedeposition on surface waters of the North Pacific has only a smalleffect (∼10%) on the long-term assessment of the plume.

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–46 45

Concerning the oceanic source, the latest estimates of total Cs-137directly released in the ocean range from only ∼1 PBq up to as high as27 PBq, when estimated from measured radioactivities (JapaneseGovernment, 2011; Kawamura et al., 2011; Bailly du Bois et al., 2012;Normile, 2013; Tsumune et al., 2013). This spread in values is some-what narrower when estimated using numerical simulations andinverse methods, ranging from 3.5 to 5.9 PBq (Tsumune et al., 2012,2013; Estournel et al., 2012). Here we used a release of 22 PBq injectedover 1 month (mid-March to mid-April) based on Bailly du Bois et al.(2012) and Buesseler et al. (2012). An additional simulation was alsoperformed using a modified total release of 27 PBq, with ∼80%(22 PBq) released during the first month and the remaining ∼20%(5 PBq) during the next 3 months, based on the total estimates ofBailly du Bois et al. (2012). Apart from linearly scaling up the Cs-137concentrations reported here by ∼20%, the results for this ensemble ofexperiments are qualitatively identical in the long-term (not shown).For the decadal time-scales considered here, the total release is ofmuch more importance than its precise timing within the first coupleof months after the disaster. Another recent study estimated the Cs-137 released over the first three months directly into the coastal oceanlying between 5.1 and 5.5 PBq, while emphasizing that uncertaintiesremain on the amount of radionuclides released during the first fewdays after the accident (Estournel et al., 2012). If further studiesconfirm that this lower source is the most reliable one, the concentra-tions reported here could be simply scaled downwards to obtainestimates consistent with the lower release rate. Aoyama et al. (2013)reports that the surface Cs-137 plume was transported along 401Nwith concentrations greater than 10 Bqm−3 reaching the InternationalDate Line by March 2012, i.e. 1 year after the accident. Although anexact match between small-scale in-situ observations and our large-scale simulated concentrations is not expected, one way to take intoaccount this observational constraint is to find the scaling factor thatwould make our results match with the available observations. Wehave thus extracted the Cs-137 concentrations along the InternationalDate Line (averaged within a 21 band from 1791E to 1791W) and weinvestigated the time-series at 401N (not shown). To match oursimulated concentrations with the observed localized concentrations,a scaling factor of 0.15 is required, which implies an updated inverseestimate for the total release amount of 3.3 PBq. This is not incon-sistent with the estimates of total Cs-137 release noted above; rangingfrom ∼1 PBq to 27 PBq (Normile, 2013; Tsumune et al. (2013)). Thiswould further suggest a potential overestimation of the sourcefunction by Bailly du Bois et al. (2012), assuming our circulation fieldsand flow rates match up well with observed. If a total release of3.3 PBq is confirmed, the coastal areas north of 451N will be the onlyshelf regions affected by Cs-137 concentrations of ∼10 Bqm−3 by late2014—early 2015, while other southern shelf regions and the Hawai-ian archipelago will only see low concentrations (o5 Bqm−3) from2017 onward. Note that important caveats exist in this inverse method.For example, the method assumes that the model flow fieldscorrespond perfectly to observations, and that the source functionwe applied matches the observations in terms of distribution andduration; only differing in magnitude. Yet the circulation patterns andsource function remain uncertain; for example, two studies recentlysuggest that the release of nucleotides into the ocean might havelasted longer than just 1–2 months after the Fukushima accident(Kanda, 2013; Normile, 2013). While these uncertainties persist, ourchoice of a “scalable” approach is a useful method to reconcile thesimulations presented here and the observations of Cs-137 dispersion,as they become available in time.

Given that a large proportion of the multi-decadal variability ofthe North Pacific circulation is associated with the Pacific DecadalOscillation (PDO), we have performed additional simulations toartificially test the effect of extreme PDO states on the Cs-137plume. This was tested by looping OFES velocities over 30 yearsduring a period dominated by the positive phase of the PDO

(1983–1987), and comparing to an experiment looping velocitiesover a period dominated by the negative phase (1998–2002).Overall differences between these two experiments are small after10-year of simulation time, with the tracer concentration differ-ences remaining at less than 5 Bq m−3, while the major tracerpathways previously described remain robust in both PDO experi-ments (not shown). The main differences in the first 10-year of thesimulation relate to (i) the north/south orientation of the axisof the Kuroshio Current (KC) and the Kuroshio Extension (KE),(ii) the zonal transport of the KC/KE and associated jets and(iii) the strength of the southern recirculation. This is consistentwith the work of Qiu (2000, 2003), and Qiu and Chen (2005), whodescribed the interannual variability of the Kuroshio systemidentifying two different states: one “elongated” state, which hasbeen associated with a negative PDO, characterized by a strongzonal mean transport and increased jet-like structures of the KC/KE, accompanied by a more intense southern recirculation; andthe second “contracted” state, typified by a weaker zonal advec-tion, a more southward axis of the KC and KE, and a weakerrecirculation, associated with positive PDO phases. This inter-decadal variability, which also influences eddy-activity (Qiu andChen, 2005) and mode water formation in the NW Pacific (Okaet al., 2012), will likely affect the future three-dimensional path-ways of the plume, but this is impossible to estimate directly.However, as noted above, the overall pattern of tracer spread andcirculation pathways remains consistent across the control, PDOpositive, and PDO negative experiments, adding further confi-dence in our large-scale multi-decadal projections.

5. Conclusions

Based on output from an eddy-resolving ocean model we find thatthe Fukushima Cs-137 plume is rapidly diluted within the Kuroshiosystem over a time-scale of 4 months. Over the subsequent fewdecades the model projects that a significant amount of Fukushima-derived nucleotides will spread across the North-Pacific basin, drivenprimarily by advection within the subtropical gyre. The modelestimates that a component of Fukushima Cs-137 will be injectedinto the interior ocean via subduction, before eventually returning tothe surface by coastal upwelling along the west coast of NorthAmerica. Ultimately, the signature will spread into other ocean basins,particularly the Indian and South Pacific Oceans. The long-termevolution of the modeled plume, eventually scaled up to a reliablesource function, could be compared to observations from futuremonitoring programs to evaluate water mass pathways and transittimes in the North Pacific.

Acknowledgments

V.R. acknowledges support from MICINN and FEDER throughthe ESCOLA project (CTM2012-39025-C02-01) while finishing thispaper. The OFES simulation was conducted on the Earth Simulatorunder the support of JAMSTEC. E.V.S. and M.H.E. are supported bythe Australian Research Council (ARC), including the ARC Centre ofExcellence for Climate System Science. The authors thank thethree anonymous reviewers for their constructive comments thathelped improve the original manuscript.

References

Aoki, S., Hariyama, M., Mitsudera, H., Sasaki, H., Sasai, Y., 2007. Formation regions ofSubantarctic Mode Water detected by OFES and Argo profiling floats. Geophys.Res. Lett. 34 (10), L10606, http://dx.doi.org/10.1029/2007GL029828.

V. Rossi et al. / Deep-Sea Research I 80 (2013) 37–4646

Aoyama, M., Hirose, K., Nemoto, K., Takatsuki, Y., Tsumune, D., 2008. Water masseslabeled with global fallout 137Cs formed by subduction in the North Pacific.Geophys. Res. Lett. 35, L01604.

Aoyama, M., Povinec, P.P., Sanchez-Cabeza, J.-A., 2011. The Southern HemisphereOcean Tracer Studies (SHOTS) project. Prog. Oceanogr. 89, 1–6.

Aoyama, M., Uematsu, M., Tsumune, D., Hamajima, Y., 2013. Surface pathway ofradioactive plume of TEPCO Fukushima NPP1 released 134-Cs and 137-Cs.Biogeosci. Discuss 10, 265–283.

Bailly du Bois, P., Laguionie, P., Boust, D., Korsakissok, I., Didier, D., Fiévet, B., 2012.Estimation of marine source-term following Fukushima Dai-ichi accident. J.Environ. Radioact. 114, 2–9.

Behrens, E., Schwarzkopf, F.U., Lübbecke, J.F., Böning, C.W., 2012. Model simulationson the long-term dispersal of 137Cs released into the Pacific Ocean offFukushima. Environ. Res. Lett. 7, 034004.

Buesseler, K., Aoyama, M., Fukasawa, M., 2011. Impacts of the Fukushima NuclearPower Plants on Marine Radioactivity. Environ. Sci. Technol. 45, 9931–9935.

Buesseler, K.O., Jayne, S.R., Fisher, N.S., Rypina, I.I., Baumann, H., Baumann, Z.,Breiera, C.F., Douglass, E.M., George, J., Macdonald, A.M., Miyamoto, H.,Nishikawa, J., Pike, S.M., Yoshida, S., 2012. Fukushima-derived radionuclidesin the ocean and biota off Japan. Proc. Natl. Acad. Sci. USA 109 (16), 5984–5988,http://dx.doi.org/10.1073/pnas.1120794109, in press.

Chino, M., Nakayama, H., Nagai, H., Terada, H., Katata, G., Yamazawa, H., 2011.Preliminary estimation of release amounts of 131I and 137Cs accidentallydischarged from the Fukushima Daiichi nuclear power plant into the atmo-sphere. J. Nucl. Sci. Technol. 48, 1129–1134.

Choi, Y., Kida, S., Takahasi, K., 2013. The impact of oceanic circulation and phasetransfer on the dispersion of radionuclides released from the Fukushima Dai-ichi Nuclear Power Plant. Biogeosci. Discuss 10, 3677–3705.

Dietze, H., Kriest, I., 2011. Tracer distribution in the Pacific Ocean following a releaseoff Japan—what does an oceanic general circulation model tell us? Ocean Sci.Discuss 8, 1441–1466.

Estournel, C., Bosc, E., Bocquet, M., Ulses, C., Marsaleix, P., Winiarek, V., Osvath, I.,Nguyen, C., Duhaut, T., Lyard, F., Michaud, H., Auclair, F., 2012. Assessment of theamount of cesium-137 released into the Pacific Ocean after the Fukushimaaccident and analysis of its dispersion in Japanese coastal waters. J. Geophys.Res. 117, C11014, http://dx.doi.org/10.1029/2012JC007933.

Garnier-Laplace, J., Beaugelin-Seiller, K., Hinton, T.G., 2011. Fukushima wildlife dosereconstruction signals ecological consequences. Environ. Sci. Technol. 45,5077–5078.

Hazell, D.R., England, M.H., 2003. Prediction of the fate of radioactive material inthe South Pacific Ocean using a global high resolution ocean model. J. Environ.Radioact. 65, 329–355.

Hawkins, E., Sutton, R., 2009. The potential to narrow uncertainty in regionalclimate predictions. Bull. Am. Meteorol. Soc. 90 (8), 1095–1107.

Heldal, H.E., Stupakoff, I., Fisher, N.S., 2001. Bioaccumulation of 137-Cs and 57-Coby five marine phytoplankton species. J. Environ. Radioact. 57, 231–236.

IAEA (International Atomic Energy Agency), 2004. Sediment Distribution Coeffi-cients and Concentration Factors for Biota in the Marine Environment. IAEA,Vienna.

Japanese Government, 2011. Report of Japanese Government to the IAEA MinisterialConference on Nuclear Safety—The Accident at TEPCO's Fukushima NuclearPower Stations. Available from: ⟨http://www.kantei.go.jp/foreign/kan/topics/201106/pdf/chapter_vi.pdf⟩ (accessed 25.04.12.).

Kanda, J., 2013. Continuing 137-Cs release to the sea from the Fukushima Dai-ichiNuclear Power Plant through 2012. Biogeosci. Discuss 10, 3577–3595.

Kameník, J., Dulaiova, H., Buesseler, K.O., Pike, S.M., Štastná, K., 2013. Cesium-134and 137 activities in the central North Pacific Ocean after the Fukushima Dai-ichi nuclear power plant accident. Biogeosci. Discuss 10, 5223–5244.

Kawamura, H., Kobayashi, T., Furuno, A., In, T., Ishikawa, Y., Nakayama, T., Shima, S.,Awaji, T., 2011. Preliminary numerical experiments on oceanic dispersion of131I and 137Cs discharged into the ocean because of Fukushima Daiichi nuclearpower plant disaster. J. Nucl. Sci. Technol. 48, 1349–1356.

Kobayashi, T., Otosaka, S., Togawa, O., Hayashi, K., 2007. Development of a non-conservative radionuclides dispersion model in the ocean and its application tosurface cesium-137 dispersion in the Irish Sea. J. Nucl. Sci. Technol. 44 (2),238–247.

Kusakabe, M., Oikawa, S., Takata, H., Misonoo, J., 2013. Spatio-temporal distribu-tions of Fukushima-derived radionuclides in surface sediments in the waters offMiyagi, Fukushima, and Ibaraki Prefectures, Japan. Biogeosci. Discuss 10,4819–4850.

Livingston, H.D., Povinec, P.P., 2000. Anthropogenic marine radioactivity. OceanCoast. Manage. 43, 689–712.

Masumoto, et al., 2004. A fifty-year eddy-resolving simulation of the world ocean—preliminary outcomes of OFES (OGCM for the earth simulator). J. EarthSimulator 1, 35–56.

Monte, L., Periañez, R., Boyer, P., Smith, J.T., Brittain, J.E., 2009. The role of physicalprocesses controlling the behaviour of radionuclide contaminants in theaquatic environment: a review of state-of-the-art modelling approaches. J.Environ. Radioact. 100, 779–784.

Nagao, S., Kanamori, M., Ochiai, S., Tomihara, S., Fukushi, K., Yamamoto, M., 2013.Export of 134-Cs and 137-Cs in the Fukushima river systems at heavy rains byTyphoon Roke in September 2011. Biogeosci. Discuss 10, 2767–2790.

Nakano, H., Motoi, T., Hirose, K., Aoyama, M., 2010. Analysis of 137Cs concentrationin the Pacific using a Lagrangian approach. J. Geophys. Res. 115, C06015.

Nakano, M., Povinec, P.P., 2012. Long-term simulations of the 137-Cs dispersionfrom the Fukushima accident in the world ocean. J. Environ. Radioact. 111,109–115.

Niiler, P.P., Maximenko, N.A., Panteleev, G.G., Yamagata, T., Olson, D.B., 2003. Near-surface dynamical structure of the Kuroshio Extension. J. Geophys. Res. 108,C63193.

Nonaka, M., Xie, S-P, Sasaki, H., 2012. Interannual variations in low potentialvorticity water and the subtropical countercurrent in an eddy-resolving OGCM.J. Oceanogr. 68, 139–150.

Normile, D., 2013. The Pacific Swallows Fukushima's Fallout. Science 340 (6132),547, http://dx.doi.org/10.1126/science.340.6132.547.

Oka, E., Qiu, B., 2012. Progress of North Pacific mode water research in the pastdecade. J. Oceanogr. 68, 5–20, http://dx.doi.org/10.1007/s10872-011-0032-5.

Oka, E., Qiu, B., Kouketsu, S., Uehara, K., Suga, T., 2012. Decadal seesaw of theCentral and Subtropical Mode Water formation associated with the KuroshioExtension variability. J. Oceanogr. 68, 355–360.

Paris, C.B., Helgers, J., Van Sebille, E., Srinivasan, A., 2013. The ConnectivityModeling System: a probabilistic modelling tool for the multi-scale trackingof biotic and abiotic variability in the ocean. Environ. Modell. Softw. 42, 47–54,http://dx.doi.org/j.envsoft.2012.12.006.

Periañez, R., 2000. Modelling the tidal dispersion of 137-Cs and 239,240-Pu in theEnglish Channel. J. Environ. Radioact. 49, 259–277.

Periañez, R., Elliott, A.J., 2002. A particle-tracking method for simulating thedispersion of non-conservative radionuclides in coastal waters. J. Environ.Radioact. 58, 13–33.

Qiu, B., 2000. Interannual variability of the Kuroshio Extension and its impact onthe wintertime SST field. J. Phys. Oceanogr. 30, 1486–1502.

Qiu, B., 2003. Kuroshio Extension variability and forcing of the Pacific decadaloscillations: responses and potential feedback. J. Phys. Oceanogr. 33,2465–2482.

Qiu, B., Chen, S., 2005. Variability of the Kuroshio Extension jet, recirculation gyreand mesoscale eddies on decadal timescales. J. Phys. Oceanogr. 35, 2090–2103.

Qu, T., Chen, J., 2009. A North Pacific decadal variability in subduction rate.Geophys. Res. Lett. 36 (22), L22602, http://dx.doi.org/10.1029/2009GL040914.

Rossi, V., Lopez, C., Hernandez-Garcia, E., Sudre, J., Garcon, V., Morel, Y., 2009.Surface mixing and biological activity in the four Eastern Boundary UpwellingSystems, Nonlin. Processes Geophys. 16, 557–568, http://dx.doi.org/10.5194/2009.

Sasaki, H., Nonaka, M., Masumoto, Y., Sasai, Y., Uehara, H., Sakuma, H., 2008. Aneddy-resolving hindcast simulation of the quasiglobal ocean from 1950 to 2003on the Earth Simulator. In: Hamilton, K., Ohfuchi, W. (Eds.), High ResolutionNumerical Modelling of the Atmosphere and Ocean. Springer, New York,pp. 157–185.

Sasaki, H., Xie, S-P, Taguchi, B., Nonaka, M., Hosoda, S., Masumoto, Y., 2012.Interannual variations of the Hawaiian Lee Countercurrent induced bypotential vorticity variability in the subsurface. J. Oceanogr. 68, 93–111.

Stohl, A., Seibert, P., Wotawa, G., Arnold, D., Burkhart, J.F., Eckhardt, S., Tapia, C.,Vargas, A., Yasunari, T.J., 2011. Xenon-133 and caesium-137 releases into theatmosphere from the Fukushima Dai-ichi nuclear power plant: determinationof the source term, atmospheric dispersion, and deposition. Atmos. Chem. Phys.Discuss. 11, 28319–28394.

Taguchi, B., Xie, S-P., Schneider, N., Nonaka, M., Sasaki, H., Sasai, Y., 2007. Decadalvariability of the kuroshio Extension: observations and an eddy-resolvingmodel hindcast. J. Clim. 20 (11), 2357–2377.

Taguchi, B., Qiu, B., Nonaka, M., Sasaki, H., Xie, S-P., Schneider, N., 2010. Decadalvariability of the Kuroshio Extension: mesoscale eddies and recirculations.Ocean Dynam. 60, 673–691.

Tsumune, D., Aoyama, M., Hirose, K., Bryan, F.O., Lindsay, K., Danabasoglu, G., 2011.Transport of 137Cs to the Southern Hemisphere in an ocean general circulationmodel. Prog. Oceanogr. 89, 38–48.

Tsumune, D., Tsubono, T., Aoyama, M., Hirose, K., 2012. Distribution of oceanic 137-Cs from the Fukushima Dai-ichi nuclear power plant simulated numerically bya regional ocean model. J. Environ. Radioact. 111, 100–108 10.1016/j.jenvrad.2011.10.007.

Tsumune, D., Tsubono, T., Aoyama, M., Uematsu, M., Misumi, K., Maeda, Y., Yoshida,Y., Hayami, H., 2013. One-year, regional-scale simulation of 137Cs radioactivityin the ocean following the Fukushima Daiichi Nuclear Power Plant accident.Biogeosci. Discuss 10, 6259–6314.

Van Sebille, E., 2009. Assessing Agulhas Leakage. Utrecht University, Utrecht (Ph.D.dissertation).

Van Sebille, E., Baringer, M.O., Johns, W.E., Meinen, C.S., Beal, L.M., de Jong, F., vanAken, H.M., 2011. Propagation pathways of classical Labrador Sea Water fromits source region to 261N. J. Geophys. Res. 116, C12027.

Van Sebille, E., Johns, W.E., Beal, L.M., 2012. Does the vorticity flux from Agulhasrings control the zonal pathway of NADWacross the South Atlantic? J. Geophys.Res. 117, C05037.

Vogel, C., Fisher, N.S., 2010. Metal accumulation by heterotrophic marine bacter-ioplankton. Limnol. Oceanogr. 55 (2), 519–528.

Yasunari, T.J., Stohl, A., Hayano, R.S., Burkhart, J.F., Eckhardt, S., Yasunari, T., 2011.Cesium-137 deposition and contamination of Japanese soils due to theFukushima nuclear accident. Proc. Natl. Acad. Sci. USA 108 (49), 19530–19534.