Fukushima 137Cs releases dispersion modelling over 1 the Pacific Ocean. Comparisons of models with water, 2 sediment and biota data

Fukushima 137Cs releases dispersion modelling over1 the Pacific Ocean. Comparisons of models with water,2 sediment and biota data3 4 December 14, 20185 Abstract6 A number of marine radionuclide dispersion models (both Eulerian and La-7 grangian) were applied to simulate 137Cs releases from Fukushima Daiichi nuclear8 power plant accident in 2011 over the Pacific at oceanic scale. Simulations extended9 over two years and both direct releases into the ocean and deposition of atmospheric10 releases on the ocean surface were considered. Dispersion models included an embed-11 ded biological uptake model (BUM). Three types of BUMs were used: equilibrium,12 dynamic and allometric. Model results were compared with 137Cs measurements13 in water (surface, intermediate and deep layers), sediment and biota (zooplankton,14 non-piscivorous and piscivorous fish). A reasonable agreement in model/model and15 model/data comparisons was obtained.16 Keywords: Fukushima-Daiichi accident; dispersion model; ocean; sediment; biological17 uptake model; caesium18 1 Introduction19 After the 9.0 magnitude earthquake and resulting tsunami occurred on March 11th, 2011,20 in Japan, significant amounts of radioactive material were released to the environment21 from Fukushima Dai-ichi nuclear power plant (FDNPP). Radionuclides released to the22 atmosphere were transported eastward by a strong jet stream and reached the coast of23 North America in four days (Takemura et al., 2011). A portion of these radionuclides was24 deposited on the Pacific Ocean surface by wet and dry deposition processes. In addition,25 1 water used to cool a damaged nuclear reactor leaked into the ocean (Kobayashi et al.,26 2013).27 Some exercises comparing numerical model performances when applied to simulate the28 137Cs releases from FDNPP in the Pacific Ocean have been carried out, as for instance in29 Masumoto et al. (2012). These authors found discrepancies between the five participating30 models and concluded that they were due to the different calculated current fields in the31 coastal waters of Japan, off Fukushima, which lead to different radionuclide distributions.32 Differences in circulation fields were caused by the different ocean models and dispersion33 model settings used by the research groups. However, a systematic assessment aimed at34 investigating the reasons of differences was not carried out.35 The Science Council of Japan (SCJ, 2014) carried out a similar intercomparison study36 for 137Cs, with eleven models involved. Again, significant differences between models37 were found. Models were different in concept (Eulerian vs. Lagrangian), with different38 setting and even different source terms. It was concluded that a simple comparison was39 not straightforward and consequently detailed systematic comparison studies, such as40 ones that use the same radionuclide forcing with different models and/or the same model41 with different forcing scenarios, were required. This kind of intercomparison exercise was42 carried out in the frame of IAEA (International Atomic Energy Agency) MODARIA1pro-43 gram (Peri´a˜nez et al., 2015a; 2016a). The MODARIA project was running from 2012 to44 2015 to make progress in the assessment of radioactive substances in the environment and45 its impact to man and biota. Different dispersion models were applied to simulate FDNPP46 releases in the Pacific, using different and also the same water circulation fields. Simu-47 lations with the same set of parameters (like diffusion coefficients for instance) were also48 carried out. It was found that the main source of discrepancy between different dispersion49 models was due to the different circulation fields. Model/model and model/measurements50 1Modelling and Data for Radiological Impact Assessments. Further information can be found here: http://www-ns.iaea.org/projects/modaria/default.asp?l=116 2 comparisons for both the dissolved phase and bed sediments (not included in earlier model51 comparison exercises) were carried out in such study.52 Alternatively, the same dispersion model forced with different circulation fields was53 tested as well, although water/sediment interactions were not included in this study54 (Kawamura et al., 2017).55 An interesting exercise was described in Maderich et al. (2018). In this case the56 same dispersion model, running with generic parameters, was applied to describe 137Cs57 dispersion from Chernobyl NPP accident in the Baltic and Black seas, and FDNPP acci-58 dent in the Pacific Ocean. The applied box model (POSEIDON-R; Lepicard et al., 2004;59 Maderich et al., 2014a; 2014b; Bezhenar et al., 2016) contained an embedded food web60 model. Comparisons of model results with measurements in the three scenarios indicated61 that, with some restrictions, the model could be used with generic parameter values in62 radiation emergency situations in areas where limited information is available.63 MODARIA-II program2was launched by the IAEA in 2016 as a follow-up of MODARIA.64 The work in comparing numerical model performances when applied to simulate FDNPP65 releases in the ocean was continued in the frame of this project. Nevertheless, spatial range66 and temporal frame of simulations were extended: two year long simulations over almost67 the whole North Pacific Ocean were carried out. In addition, marine dispersion models68 contain an integrated biological uptake model (BUM) with four components (phytoplank-69 ton, zooplankton, non-piscivorous and piscivorous fish). Model/model and model/data70 comparisons were carried out for water, bed sediments and biological components of the71 models, which has not been done before.72 Six institutes have participated in the model comparisons. These are the Institute73 of Mathematical Machines and System Problem (IMMSP, Ukraine), Korea Institute of74 Ocean Science and Technology (KIOST, Rep. of Korea), ABmerit (Slovakia), Univer-75 sity of Seville (USEV, Spain), Japan Atomic Energy Agency (JAEA, Japan), and Korea76 2http://www-ns.iaea.org/projects/modaria/modaria2.asp?s=8&l=129 3 Atomic Energy Research Institute (KAERI, Rep. of Korea).77 The methodology is presented in section 2, where water circulation used by models,78 source terms, and the origin of experimental data on 137Cs concentrations are described.79 Results are presented in section 3. Some general discussion on model uncertainty and80 complexity is finally included in section 4.81 2 Methods82 2.1 Hydrodynamics83 Water circulation provided by FORA3model was used for calculations. This model,84 Four-dimensional Variational Ocean ReAnalysis for the Western North Pacific (FORA-85 WNP30), is the first-ever dataset covering the western North Pacific over the last three86 decades (1982-2014) at eddy-resolving resolution. It is a cooperative work of the Japan87 Agency for Marine-Earth Science and Technology (JAMSTEC) and the Meteorological88 Research Institude, Japan Meteorological Agency (JMA/MRI) using the Earth Simulator89 (Usui et al., 2017; Tsujino et al., 2010).90 The domain used in the present calculations extends 117◦E-160◦Wand15 ◦N-65◦Nin91 longitude and latitude, respectively. Horizontal resolution is 0.1oand there are 54 vertical92 levels (0-6300 m) with increasing thickness from the surface to the sea bottom. Monthly93 climatological data from 2011 to 2014 were used. Two year long (March 11, 2011 to March94 11, 2013) simulations were made.95 The model domain showing water depths and an example of surface water circulation96 (averaged value for March 2011) can be seen in Fig. 1. The general large scale circulation97 in the western Pacific Ocean is dominated by the interaction between the Kuroshio and98 Oyashio currents. The Kuroshio Current is the western boundary current in the north99 Pacific, which flows along the coast of Japan towards the north and curves to the central100 3http://synthesis.jamstec.go.jp/FORA/e/index.html 4 Pacific Ocean, then forming the so-called Kuroshio Extension. The Oyashio Current is101 a cold current which flows from the north. These two current systems converge in the102 coastal waters off Fukushima coast. Such convergence leads to the generation of unsteady103 eddies in the area. These features may be seen in Fig. 1.104 2.2 Radionuclide sources105 Radionuclides were directly introduced into the Pacific Ocean from FDNPP. They were106 also released to the atmosphere; radionuclides which were later deposited on the sea107 surface. Both sources were considered in calculations.108 Direct releases of 137Cs are given for the period March 25th, 2011, to December 31th,109 2011, and presented in Fig. 2. They were reconstructed by JAEA as explained in de-110 tail in Kobayashi et al. (2013). Monitoring data from the web site of Tokyo Electric111 Power Company (TEPCO), regarding the area near the northern and southern discharge112 channels of the Fukushima Daiichi NPP (TEPCO, 2011), were used for this purpose.113 Atmospheric deposition in the North Pacific Ocean was obtained from the averaged114 values from WSPEEDI-II (JAEA: Terada et al., 2012) and LADAS (KAERI: Suh et al.,115 2006; Suh et al., 2009) atmospheric dispersion models for the period March 12th, 2011,116 to June 1st, 2011. Even though simulations are 2 year long, most deposition occurred117 within the first months after the accident. Daily integrated values were provided. As an118 example, the integrated deposition for March 15th, 2011, averaged from both models, is119 presented in Fig. 2.120 In addition, a pre-FDNPP accident 137Cs uniform background of 1.5 Bq/m3was con-121 sidered over the Pacific Ocean waters, in order to carry out comparisons of model results122 with field measurements.123 5 2.3 Dispersion models124 Some of the main characteristics of the dispersion models which were applied are summa-125 rized in Table 1. Both Eulerian and Lagrangian models were used with different param-126 eterizations of horizontal and vertical diffusivities. The general characteristics and basic127 equations describing the two types of dispersion models which were applied are presented128 in appendix A.1 and A.2.129 A kinetic (dynamic) approach was applied to describe water/sediment interactions in130 both Eulerian and Lagrangian models, which is based on a desorption coefficient and the131 distribution coefficient, kd, of the corresponding radionuclide (Peri´a˜nez, 2005).132 All models used an equilibrium distribution coefficient of 2.0 m3/kg. This is the133 mean value recognized by IAEA (2004) for open ocean waters and is also in agreement134 with measurements off Fukushima (Honda et al., 2012). The kinetic rate describing135 release from sediments, k2=1,16 ×10−5s−1, was determined for Cs from experiments136 (Nyffeler et al., 1984). The kinetic rate describing uptake (k1) is derived from k2and the137 distribution coefficient, as usually done (Peri´a˜nez, 2005). A stochastic method is used to138 solve uptake/release processes in Lagrangian models (Peri´a˜nez and Elliott, 2002).139 Most models include a biological uptake model (BUM). Four species were considered:140 phytoplankton, zooplankton, non-piscivorous and piscivorous fish. Three types of BUM141 were used in the models: an equilibrium model based upon a concentration factor CR142 (appendix B.1), a dynamic model (B.2) and an allometric method (B.3). The BUM143 incorporated within each physical dispersion model is indicated in Table 1 as a reference144 to the appendix where the corresponding BUM characteristics are commented.145 2.4 Experimental data146 Model results were compared with available 137Cs measurements in water at three different147 layers, bed sediments and biological compartments (zooplankton, non-piscivorous and148 6 I/K THREETOX I/K Lagrangian ESTE USEV SEA-GEARN LORAS Model (IMMSP/KIOST) (IMMSP/KIOST) (ABmerit) (Univ. Seville) (JAEA) (KAERI) Model type Eulerian Lagrangian Lagrangian Lagrangian Lagrangian Lagrangian Maderich Brovchenko Peri´a˜nez et Kobayashi Min et al. Reference et al. (2016) et al. (2018) www.abmerit.sk al. (2016b) et al. (2007) (2013) Horizontal SmagorinskyaSmagorinsky Smagorinsky diffusion formula 10 m2/s formula formula 10 m2/s 10 m2/s 10−3m2/s for d<60 m Vertical 10−5m2/s for d>120 m diffusion 10−4m2/s linear function for 10−4m2/s 10−4m2/s 10−4m2/s 10−3m2/s 60 <d<120 m Bed porosity 0.6 0.6 0.6 0.6 0.6 0.7 Sediment thickness 0.05 m 0.05 m 0.05 m 0.05 m 0.05 m 0.1 m Particle density 2600 kg/m32600 kg/m32600 kg/m32600 kg/m32600 kg/m3 137Cs kd2m 3/kg 2 m3/kg 2 m3/kg 2 m3/kg 2 m3/kg 2 m3/kg k2(s−1)3.17 ×10−8Maderich et al. (2017) 1.16 ×10−61.16 ×10−61.16 ×10−61.16 ×10−6 BUM B.2 B.2 B.1 B.2 no B.3 Table 1: Model main characteristics. dis water depth and k2is the 137Cs desorption coefficient. aSee for instance CushmanRoisin and Beckers (2011). A selected reference is given for each model. The BUM row indicates the appendix where some details of the uptake model are given: B.1 is an equilibrium model, B.2 is a dynamic model and B.3 is the allometric method. 7 piscivorous fish) in the surface layer (to 20 m depth). The other two considered water149 layers are 20-460 m and 460 m to the seabed.150 Measurements were compiled from the following references: Honda et al. (2012),151 Charette et al. (2013), Kaeriyama et al. (2013) for water; the “Database for Radioac-152 tive Substance Monitoring Data”4for sediments; Honda et al. (2012), Kitamura et al.153 (2013) for zooplankton; Wada et al. (2016); Men et al. (2017); Johansen et al. (2014)154 for fish (piscivorous and non-piscivorous). Only data for pelagic fish were used. Sam-155 pled pelagic non-piscivorous fish are Engraulis japonicus,Etrumeus teres,Clupea pallasii156 and Hyporhamphus sajori. Sampled pelagic piscivorous fish are Hexagrammos sebastes,157 Todarodes pacificus,Snake mackerel,Oncorhynchus keta,Ammodytes japonicus,Seriola158 quinqueradiata,Seriola quinqueradiata,Trachurus japonicus and Scomber japonicus.Wa-159 ter samples collected in the direct release area have been filtered out since the models are160 giving average value of radionuclide concentrations over boxes, as explained below.161 Locations where samples were collected during the simulation period are indicated as162 dots in Fig. 3. The Pacific Ocean was divided into a number of boxes, presented in Fig. 4,163 according to general circulation and the location of the release point. Model results were164 averaged for each box and then these averaged values were compared with measurements.165 Boxes in the release area may be too large for a detailed study of radionuclide be-166 haviour in such region close to FDNPP. However, it should be taken into account that167 the dispersion of FDNPP 137Cs releases was studied at a smaller spatio-temporal scale in168 a previous paper of the group (Peri´a˜nez et al., 2015a); and model predictions and mea-169 surements were compared in the area close to FDNPP (less than some 100 km away).170 The present work is complementing such previous paper, going to larger spatial and tem-171 poral scales. Thus, large boxes are used. In addition, it should be considered that a172 model/data comparison for specific points in such a large domain is not feasible with173 Lagrangian models which release individual particles, and it is better to use averages over174 4http://emdb.jaea.go.jp/emdb/en/ 8 given areas, which are defined in view of the physical oceanography of the region (Peri´a˜nez175 et al., 2015a; 2015b; 2016a). However, it should be noted that measurements were not176 distributed homogeneously in the relatively large considered boxes.177 3 Results178 As explained before, two year long simulations were carried out; from March 2011 to179 March 2014. Monthly mean values of 137Cs concentrations in each box in Fig 4 were180 provided by the models for the three water layers, seabed sediments and the four biological181 compartments (surface layer only).182 Model results and 137Cs measurements are presented in Fig. 5 to Fig. 12. Results are183 presented only for such boxes where measurements are available. Results for the abiotic184 and biotic components of the models are discussed separately in the following subsections.185 3.1 Water and sediments186 Results for surface water may be seen in Fig. 5 and Fig. 6, for boxes which are far187 from Japan and boxes located closer, around FDNPP, respectively. In boxes 1, 3, 5 and188 20 (Fig. 5) there is a slight increase in 137Cs concentrations with respect to background189 immediately after the accident, which must be attributed to atmospheric deposition. In190 general, models produce this initial increase, which is about one order of magnitude above191 background. In other boxes (like 15 and 16), both models and measurements indicate pre-192 FDNPP accident background. Thus, releases did not affect these areas in the considered193 temporal frame.194 In contrast, high concentrations are found closer to FDNPP (Fig. 6). For some of the195 boxes (6, 7, 12) models and measurements show a trend towards achieving background196 concentrations after approximately one year. The initial concentration increase above197 background is about two orders of magnitude. Other regions south from Japan (boxes 13198 9 concentrations over such boxes; as it was done for the Baltic Sea model intercomparison355 in Peri´a˜nez et al. (2015b). However, it should be taken in account that in the vicinity of356 FDNPP measurements were not distributed homogeneously in space. This can result in357 overestimation of experimental box-averaged values.358 Models agree in predicting areas in the Pacific Ocean which were affected by FDNPP359 releases (direct and/or atmospheric deposition) and regions which were not. In addition,360 predicted concentrations are within the same order of magnitude in most cases.361 With respect to calculated 137Cs temporal trends in biota, dynamic models tend to362 underestimate concentrations. Allometry and the equilibrium approach results are, in363 general, in better agreement with observations. This is explained by the higher 137Cs364 concentrations in water produced by ESTE and LORAS models. Temporal evolutions of365 137Cs concentrations calculated through the different approaches are different, although366 there is not enough experimental data to assess which approach leads to better results.367 However, it is clear that dynamic models provide the known pattern of delayed rise of368 activity concentration in biota.369 Acknowledgement370 Work carried out in the frame of IAEA MODARIA-II (Modelling and Data for Radio-371 logical Impact Assessments) program. This work was partially supported by the National372 Research Foundation of Korea (NRF) and partially funded by the Korean Government373 (MSIP) (MSIP: NRF-2017M2A8A4015253, NRF-2015M2A2B2034282). The authors are374 indebted to all members of MODARIA-II working group 7 for useful discussions held375 during group meetings.376 16 6 References377 Bezhenar, R., Jung, K.T., Maderich, V., Willemsen, S., de With, G., Qiao, F.,378 2016. Transfer of radiocesium from contaminated bottom sediments to marine or-379 ganisms through benthic food chain in post-Fukushima and post-Chernobyl periods.380 Biogeosciences 13, 3021-3034.381 Brovchenko, I., Jung, K.T., Maderich, V., Kim, K.O., Kovalets, S., 2018. Method382 of moments for random walk algorithms. In preparation.383 Brown, J., Borrentzen, P., Dowdall, M., Sazykina, T., Kryshev, I., 2004. The384 Derivation of Transfer Parameters in the Assessment of Radiological Impacts on385 Arctic marine Biota. ARCTIC 57(3), 279-289.386 Carvalho, F. P., 2018. Radionuclide concentration processes in marine organisms:387 A comprehensive review. Journal of Environmental Radioactivity 186, 124-130.388 Charette, A.M. Breier, C.F., Henderson, P.B., Pike, S.M., Rypina,I.I., Jayne, S.R.,389 Buesseler, K.O., 2013. Radium-based estimates of cesium isotope transport and390 total direct ocean discharges from the Fukushima Nuclear Power Plant accident,391 Biogeosciences 10, 2159-2167.392 Cushman-Roisin, B., Beckers, J.M., 2011. Introduction to Geophysical Fluid Dy-393 namics. Elsevier.394 Heling, R., Koziy, L., Bulgakov, V., 2002. On the dynamical uptake model devel-395 oped for the uptake of radionuclides in marine organisms for the POSEIDON-R396 model system. Radioprotection 37(C1), 833-838.397 Higley, K.A., Domotor, S.L., Antonio, E.J., Kocher, D.C., 2003. Derivation of398 a screening methodology for evaluating radiation dose to aquatic and terrestrial399 biota. Journal of Environmental Radioactivity, 66, 41-59.400 17 Honda M., Aono T., Aoyama M., Hamajima Y., Kawakami H., Kitamura M., Ma-401 sumoto Y., Miyazawa Y., Takigawa M., Saino T., 2012. Dispersion of artificial402 caesium-134 and -137 in the western North Pacific one month after the Fukushima403 accident. Geochemical Journal 46, 1-9.404 IAEA, 2004. Sediment distribution coefficients and concentration factors for biota405 in the marine environment. Technical Reports Series 422, Vienna.406 Johansen, M.P., Ruedig, E., Tagami, K., Uchida, S., Higley, K., Beresford, N., 2014.407 Radiological dose rates to marine fish from the Fukushima Daiichi accident: the first408 three years across the North Pacific, Environmental Science and Technology 49(3),409 1277-1285.410 Kaeriyama, H., Ambe, D., Shimizu, Y., Fujimoto, K., Ono, T., Yonezaki, S., Kato,411 Y., Matsunaga, H., Minami, H., Nakatsuka, S., Watanabe, T., 2013. Direct ob-412 servation of 134Cs and 137Cs in surface seawater in the western and central North413 Pacific after the Fukushima Dai-ichi nuclear power plant accident, Biogeosciences414 10, 4287-4295.415 Kawamura, H., Furuno, A., Kobayashi, T., In, T., Nakayama, T., Ishikawa, Y.,416 Miyazawa, Y., Usui, N., 2017. Oceanic dispersion of Fukushima-derived Cs-137417 simulated by multiple oceanic general circulation models. Journal of Environmental418 Radioactivity 180, 36-58.419 Kitamura, M., Kumamoto, Y., Kawakami, H., Cruz, E.C., Fujikura, K., 2013. Hor-420 izontal distribution of Fukushima-derived radiocesium in zooplankton in the north-421 western Pacific Ocean. Biogeosciences 10, 5729-5738.422 Kobayashi, T., Otosaka, S., Togawa, O., Hayashi, K., 2007. Development of a423 nonconservative radionuclides dispersion model in the ocean and its application424 18 to surface cesium-137 dispersion in the Irish Sea. Journal of Nuclear Science and425 Technology 44(2), 238-247.426 Kobayashi, T., Nagai, H., Chino, M., Kawamura, H., 2013. Source term estimation427 of atmospheric release due to the Fukushima Dai-ichi Nuclear Power Plant accident428 by atmospheric and oceanic dispersion simulations. Journal of Nuclear Science and429 Technology 50, 255-264.430 Kusakabe, M., Oikawa, S., Takata, H., Misonoo, J., 2013. Spatiotemporal dis-431 tributions of Fukushima-derived radionuclides in nearby marine surface sediments.432 Biogeosciences 10, 5019-5030.433 Lepicard S., Heling R., Maderich V. 2004. POSEIDON/RODOS model for radio-434 logical assessment of marine environment after accidental releases: application to435 coastal areas of the Baltic, Black and North seas. Journal of Environmental Ra-436 dioactivity 72, 1-2, 153-161437 Lynch D.R., Greenberg D.A., Bilgili A., Mcgillicuddy, D.J., Manning, J.P., Aretx-438 abaleta A.L., 2015. Particles in the Coastal Ocean. Theory and Applications.439 Cambridge University Press, NY.440 Maderich V., Heling R., Bezhenar R., Brovchenko I., Jenner H., Koshebutskyy V.,441 Kuschan A., Terletska K.. 2008. Development and application of 3D numerical442 model THREETOX to the prediction of cooling water transport and mixing in the443 inland and coastal waters. Hydrological Processes 22, 1000-1013.444 Maderich, V., Bezhenar, R., Heling, R., de With, G., Jung, K.T., Myoung, J.G.,445 Cho, Y.K., Qiao, F., Robertson, L., 2014a. Regional long-term model of radioac-446 tivity dispersion and fate in the northwestern Pacific and adjacent seas: application447 to the Fukushima Dai-ichi accident. Journal of Environmental Radioactivity 131,448 4-18.449 19 Maderich V., Jung, K.T., Bezhenar R., de With, G., Qiao F., Casacuberta, N.,450 Masqu´e, P., Kim, Y.H., 2014b. Dispersion and fate of 90Sr in the Northwestern Pa-451 cific and adjacent seas: Global fallout and the Fukushima Dai-ichi accident. Science452 of the total Environment 494-495, 261-271.453 Maderich V., Brovchenko I., Dvorzhak A., Ievdin I., Koshebutsky, V., Peria˜nez, R.,454 2016. Integration of 3D model THREETOX in JRODOS, implementation studies455 and modelling of Fukushima scenarios. Radioprotection 51(HS2), S133-S135.456 Maderich, V., Jung, K.T., Brovchenko, I., Kim, K.O., 2017. Migration of radioac-457 tivity in multi-fraction sediments. Environmental Fluid Mechanics 17, 1207-1231.458 Maderich, V., Bezhenar, R., Tateda, Y., Aoyama, M., Tsumune, D., 2018. Sim-459 ilarities and differences of 137Cs distributions in the marine environments of the460 Baltic and Black seas and off the Fukushima Dai-ichi nuclear power plant in model461 assessments. Marine Pollution Bulletin 135, 895-906.462 Masumoto, Y., Miyazawa, Y., Tsumune, D., Kobayashi, T., Estournel, C., Marsaleix,463 P., Lanerolle, L., Mehra, A., and Garraffo, Z.D., 2012. Oceanic dispersion simulation464 of Cesium-137 from Fukushima Dai-ichi nuclear power plant, Elements, 8, 207-212.465 Men,W.,Deng,F.,He,J.,Yu,W.,Wang,F.,Li,Y.,Lin,F.,Lin,J.,Lin,L.,Zhang,466 Y., Yu, X., 2017. Radioactive impacts on nekton species in the Northwest Pacific467 and humans more than one year after the Fukushima nuclear accident, Ecotoxicology468 and Environmental Safety 144, 601-610.469 Min, B.I., Peri´a˜nez, R., In-Gyu Kim, Kyung-Suk Suh, 2013. Marine dispersion470 assessment of 137Cs released from the Fukushima nuclear accident. Marine Pollution471 Bulletin 72, 22-33.472 Monte, L., Hakason, L., Peri´a˜nez,R., Laptev, G., Zheleznyak, M., Maderich, V.,473 20 Koshebutsky, V., 2006. Experiences from a case study of multi-model application to474 assess the behaviour of pollutants in the Dnieper-Bug estuary. Ecological Modelling475 195, 247-263.476 Nyffeler, U.P., Li, Y.H., Santschi, P.H., 1984. A kinetic approach to describe477 trace element distribution between particles and solution in natural aquatic sys-478 tems. Geochimica Cosmochimica Acta 48, 1513-1522.479 Peri´a˜nez, R., Elliott, A. J., 2002. A particle-tracking method for simulating the480 dispersion of non-conservative radionuclides in coastal waters. Journal of Environ-481 mental Radioactivity 58(1), 13-33.482 Peri´a˜nez, R., 2005. Modelling the dispersion of radionuclides in the marine environ-483 ment: an introduction. Springer, Berlin.484 Peri´a˜nez, R., Bezhenar, R., Byung-Il, Min, Duffa, C., Jung, K., Kobayashi, T.,485 Kyung-Suk, Suh, Lamego, F., Maderich,V., Nies,H., Osvath, I., Psaltaki, M., 2015a.486 A new comparison of marine dispersion model performances for Fukushima releases487 in the frame of IAEA MODARIA program. Journal of Environmental Radioactivity488 150, 247-269.489 Peri´a˜nez, R. Bezhenar, R., Iosjpe, M., Maderich, V., Nies, H., Osvath, I., Outola, I.,490 de With, G., 2015b. A comparison of marine radionuclide dispersion models for the491 Baltic Sea in the frame of IAEA MODARIA program. Journal of Environmental492 Radioactivity 139, 66-77.493 Peri´a˜nez, R., R. Bezhenar, I. Brovchenko, C. Duffa, K.T. Jung, T. Kobayashi, F.494 Lamego, V. Maderich, B.I. Min, H. Nies, I. Osvath, I. Outola, M. Psaltaki, K.S.495 Suh, G. de With, 2016a. Modelling of marine radionuclide dispersion in IAEA496 MODARIA program: lessons learnt from the Baltic Sea and Fukushima scenarios.497 Science of the Total Environment 569/570, 594-602.498 21 Peri´a˜nez, R., Kyung-Suk Suh, Byung-Il Min, 2016b. The behaviour of 137Cs in the499 North Atlantic Ocean assessed from numerical modelling: releases from nuclear fuel500 reprocessing factories, redissolution from contaminated sediments and leakage from501 dumped nuclear wastes. Marine Pollution Bulletin 113, 343-361.502 Proehl, J.A., Lynch, D.R., McGuillicuddy, D.J., Ledwell, J.R., 2005. Modeling503 turbulent dispersion on the North Flank of Georges Bank using Lagrangian particle504 methods. Continental Shelf Research 25, 875-900.505 Protter, P.E., 2004. Stochastic Integration and Differential Equations (2nd ed.).506 Springer, ISBN 3-540-00313-4.507 Schmidt-Nielsen, K., 1977. Problems of scaling: locomotion and physiological cor-508 relates. In: Scale Effects in Animal Locomotion, T.J. Pedley (editor). Academic509 Press, New York. pp. 1-21.510 SCJ, 2014. A review of the model comparison of transportation and deposition of511 radioactive materials released to the environment as a result of the Tokyo Electric512 Power Compny’s Fukushima Daiichi Nuclear Power Plant accident. Report of the513 Sectional Committee on Nuclear Accident Committee on Comprehensive Synthetic514 Engineering, Science Council of Japan.515 Suh, K.S., Jeong, H.J., Kim, E.H., Hwang, W.T., Han, M.H., 2006. Verification516 of the Lagrangian particle model using the ETEX experiment. Annals of Nuclear517 Energy 33, 1159-1163.518 Suh, K.S., Han, M.H., Jung, S.H., Lee, C.W., 2009. Numerical simulation for a long-519 range dispersion of a pollutant using Chernobyl data. Mathematical and Computer520 Modeling 49, 337-343.521 Takemura, T., Nakamura, H., Takigawa, M., Kondo, H., Satomura. T., Miyasaka,522 22 T., Nakajima, T., 2011. A numerical simulation of global transport of atmospheric523 particles emitted from the Fukushima Daiichi Nuclear Power Plant. SOLA 7, 101-524 104.525 TEPCO, 2011. http://www.tepco.co.jp/en/nu/fukushima-np/f1/index2-e.html.526 Terada, H., Katata, G., Chino, M., Nagai, H., 2012. Atmospheric discharge and527 dispersion of radionuclides during the Fukushima Dai-ichi Nuclear Power Plant ac-528 cident. Part II: verification of the source term and analysis of regional-scale atmo-529 spheric dispersion. Journal of Environmental Radioactivity 112, 141-154.530 Tsujino, H., Motoi, T., Ishikawa, I., Hirabata, M., Nakano, H., Yamanaka, G.,531 Yasuda, T., Ishizaki, H., 2010. Reference Manual for the Meteorological Research532 Institute Community Ocean Model (MRI.COM) Version 3. Tsukuba (Japan): Me-533 teorological Research Institute. Technical Reports of the MRI, 59.534 Usui, N., Wakamatsu, T., Tanaka, Y., Hirose, N., Toyoda, T., Nishikawa, S., Fujii,535 Y., Takatsuki, Y., Igarashi, H., Ishikawa, Y., Kuragano, T., Kamachi, M., 2017.536 Four-dimensional Variational Ocean Reanalysis: A 30-year high-resolution dataset537 in the western North Pacific (FORA-WNP30). Journal of Oceanography 73(2),538 205-233.539 Vives i Batlle, J., Wilson, R.C., Watts, S.J., Jones, S.R., McDonald, P., Vives-540 Lynch, S., 2008. Dynamic model for the assessment of radiological exposure to541 marine biota, Journal of Environmental Radioactivity 99, 1711-1730.542 Vives i Batlle, J., Beresford, N., Beaugelin-Seiller, K., Bezhenar, R., Brown, J., Jing-543 Jy Cheng, Cujic, M., Dragovic, S.S., Duffa, C., Fievet, B., Hosseini, A., Jung, K.T.,544 Kamboj, S., Dong-Kwon Keum, Kobayashi, T., Kryshev, A., Le Poire, D., Maderich,545 V., Byung Il Min, Peri´a˜nez, R., Sazykina, T., Suh, K.S., Yu, C., Wang, C., Heling,546 23 R., 2016. Inter-comparison of dynamic models for radionuclide transfer to marine547 biota in a Fukushima accident scenario. Journal of Environmental Radioactivity548 153, 31-50.549 Wada, T., Fujita, T., Nemoto, Y., Shimamura, S., Mizuno, T., Sohtome, T., Kami-550 yama, K., Narita, K., Watanabe, M., Hatta, N., Ogata, Y., Morita, T., Igarashi, S.,551 2016. Effects of the nuclear disaster on marine products in Fukushima: An update552 after five years. Journal of Environmental Radioactivity 164, 312-324.553 West G.B,, Brown, J.H., Enquist, B.J., 1997. A general model for the origin of554 allometric scaling laws in biology. Science 276, 122-126.555 A Physical dispersion models556 Physical dispersion models (both Eulerian and Lagrangian) are based on the same general557 principles and equations; and then particularized as presented in Table 1. Thus, those558 common general descriptions are given below.559 A.1 Eulerian models560 In Eulerian models the differential equations giving temporal and spatial evolution of the561 radionuclide concentrations in different states (e.g. dissolved in water column and pore562 water in sediments, fixed on the suspended and bottom sediment etc) are solved. The563 general compact form of these equations for concentration of radioactivity Cαin state564 αper unit of volume (Bq m−3)orperunitofmass(Bqkg −1) are written in Cartesian565 coordinates as:566 ∂Cα ∂t +∂(uαCα) ∂x +∂(vαCα) ∂y +∂(wαCα) ∂z =∂ ∂x Kh ∂Cα ∂x +∂ ∂y Kh ∂Cα ∂y + 24 +∂ ∂z Kv ∂Cα ∂z + n  β=1 kβαCβ+Sα−λCα(1) where (x, y, z) are Cartesian coordinates, uα,vαand wαare components of flow field for the567 radionuclide in the state α. In general, velocity can differ for different states (e.g. due the568 presence of settling velocity for suspended sediment or to be zero in the bottom deposit).569 Khand Kvare turbulent or molecular diffusivities in the horizontal and vertical directions570 respectively, and/or biodiffusivity in the bottom deposit, which are variable in time and571 space. The term n β=1 kβαCβdescribes first order reactions between the radionuclides in572 different states, kβα are kinetic transfer coefficients and kαα =−n β=1 kαβ for α=β;Sα 573 is the radionuclide source term and λis the radionuclide decay constant. Equations for574 the water column and bottom sediment layer are linked by fluxes of activity.575 A.2 Lagrangian models576 In Lagrangian models the released activity is represented by a number of particles, each577 one equivalent to a given amount of activity (Bq). The path followed by each particle is578 calculated and radionuclide concentrations are obtained from the number of particles per579 volume or mass unit. The equations describing variations of particle (in state α) position580 over each time increment dt aregivenbytheItˆo (Protter, 2004) stochastic differential581 equations:582 dx =uαdt +∂Kh ∂x dt +2KhdWx,(2) dy =vαdt +∂Kh ∂y dt +2KhdWy,(3) dz =wαdt +∂Kv ∂z dt +2KvdWz,(4) where uα,vαand wαare velocity components on coordinate axis (x, y, z) for state α;583 Wx,W y,W zare independent components of the stochastic motion (the Wiener process).584 25 120 140 160 180 200 20 30 40 50 60 Longitude Latitude Surface water 135 140 14 5 35 36 37 38 39 40 41 42 Longitude Latitude Intermediate water 135 140 145 35 36 37 38 39 40 41 42 Longitude Latitude Deep water 135 140 14 5 35 36 37 38 39 40 41 42 Longitude Latitude Sediment 120 140 160 180 200 20 30 40 50 60 Longitude Latitude Zooplankton 135 140 145 35 36 37 38 39 40 41 42 Longitude Latitude Non−piscivorous fish 120 140 160 180 200 20 30 40 50 60 Longitude Latitude Piscivorous fish Figure 3: Locations of sampling points for all considered environmental compartments. 32 120 130 140 150 160 170 180 190 200 15 20 25 30 35 40 45 50 55 60 65 Longitude Latitude 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Figure 4: Box division of the Pacific Ocean for model/data comparisons. 33 200 400 600 10−1 100 101 102 C (Bq/m3) BOX 1 USEV ESTE SEA−GEARN I/K Lagr. THREETOX LORAS 200 400 600 10−1 100 101 102 BOX 2 200 400 600 10−1 100 101 102 C (Bq/m3) BOX 3 200 400 600 10−1 100 101 102 BOX 4 200 400 600 10−1 100 101 102 C (Bq/m3) BOX 5 200 400 600 10−1 100 101 BOX 15 200 400 600 10−1 100 101 C (Bq/m3) Day after Jan 1, 2011 BOX 16 200 400 600 10−1 100 101 102 103 Day after Jan 1, 2011 BOX 20 Figure 5: Model predictions and measurements of 137Cs concentrations in surface water for some boxes in the Pacific. 34 200 400 600 10−1 100 101 102 103 C (Bq/m3) BOX 6 USEV ESTE SEA−GEARN I/K Lagr. THREETOX LORAS 200 400 600 10−1 100 101 102 103 BOX 7 200 400 600 10−1 100 101 102 103 C (Bq/m3) BOX 9 200 400 600 100 101 102 103 104 105 BOX 10 200 400 600 100 102 104 106 108 C (Bq/m3) BOX 11 200 400 600 10−1 100 101 102 103 BOX 12 200 400 600 10−1 100 101 C (Bq/m3) Day after Jan 1, 2011 BOX 13 200 400 600 10−1 100 101 Day after Jan 1, 2011 BOX 14 Figure 6: Model predictions and measurements of 137Cs concentrations in surface water for some boxes in the Pacific. 35 200 400 600 100 101 102 103 104 105 C (Bq/m3) BOX 10; Surface water USEV ESTE SEA−GEARN I/K Lagr. THREETOX LORAS 200 400 600 100 102 104 106 BOX 11; Surface water 200 400 600 10−1 100 101 102 103 104 C (Bq/kg) Day after Jan 1, 2011 BOX 10; Sediment 200 400 600 10−1 100 101 102 103 104 Day after Jan 1, 2011 BOX 11; Sediment Figure 7: Model predictions and geometric means of 137Cs concentrations measured for each month in boxes 10 and 11, for surface water and sediments. Geometric standard deviations are not drawn because they are too small compared with the vertical scales. 36 200 400 600 100 101 102 103 C (Bq/m3) BOX 10, Intermediate USEV ESTE SEA−GEARN I/K Lagr. THREETOX LORAS 200 400 600 100 101 102 103BOX 11, Intermediate 200 400 600 10−1 100 101 C (Bq/m3) Day after Jan 1, 2011 BOX 10, Deep 200 400 600 10−1 100 101 Day after Jan 1, 2011 BOX 11, Deep Figure 8: Model predictions and measurements of 137Cs concentrations in intermediate and deep waters for some boxes in the Pacific. 37 200 400 600 10−2 10−1 100 101 C (Bq/kg) BOX 9 USEV ESTE SEA−GEARN I/K Lagr. THREETOX LORAS 200 400 600 10−1 100 101 102 103 104 Day after Jan 1, 2011 BOX 10 200 400 600 10−1 100 101 102 103 104 105 C (Bq/kg) Day after Jan 1, 2011 BOX 11 Figure 9: Model predictions and measurements of 137Cs concentrations in bed sediments for some boxes in the Pacific. 38 200 400 600 10−2 10−1 100 101 102 C (Bq/kg) BOX 1 USEV ESTE THREETOX LORAS 200 400 600 10−2 10−1 100 101 Day after Jan 1, 2011 BOX 7 200 400 600 10−2 10−1 100 101 102 C (Bq/kg) Day after Jan 1, 2011 BOX 12 Figure 10: Model predictions and measurements of 137Cs concentrations in zooplankton for some boxes in the Pacific (Bq/kg wet weight). 39 100 200 300 400 500 600 700 10−2 10−1 100 101 102 C (Bq/kg) BOX 10 USEV ESTE THREETOX LORAS 100 200 300 400 500 600 700 10−1 100 101 102 103 C (Bq/kg) Day after Jan 1, 2011 BOX 11 Figure 11: Model predictions and measurements of 137Cs concentrations in non-piscivorous fish (pelagic) for some boxes in the Pacific (Bq/kg wet weight). 40 200 400 600 10−2 10−1 100 101 C (Bq/kg) BOX 3 USEV ESTE THREETOX LORAS 200 400 600 10−2 10−1 100 101 102 BOX 6 200 400 600 10−2 10−1 100 101 102 C (Bq/kg) BOX 7 200 400 600 10−2 10−1 100 BOX 8 200 400 600 10−1 101 103 C (Bq/kg) BOX 10 200 400 600 10−1 101 103 Day after Jan 1, 2011 BOX 11 200 400 600 10−2 10−1 100 C (Bq/kg) Day after Jan 1, 2011 BOX 15 Figure 12: Model predictions and measurements of 137Cs concentrations in piscivorous fish (pelagic) for some boxes in the Pacific (Bq/kg wet weight). 41