The behaviour of 236-U in the North Atlantic Ocean assessed from numerical modelling: A new evaluation of the input function into the Arctic

The behaviour of 236U in the North Atlantic Ocean assessed from numerical modelling: a new evaluation of the input function into the Arctic R. Peri´a˜nez∗ Dpt F´ısica Aplicada I, ETSIA Universidad de Sevilla Ctra Utrera km 1 41013-Sevilla, Spain Kyung-Suk Suh KAERI Daedeok-Daero 989-111 Yuseong-Gu, Daejeon Republic of Korea Byung-Il Min KAERI Daedeok-Daero 989-111 Yuseong-Gu, Daejeon Republic of Korea M. Villa-Alfageme Dpt F´ısica Aplicada II, ETSIE Avda Reina Mercedes s/n University of Sevilla Spain December 27, 2017 Abstract1 A numerical model, previously validated with other radionuclides, was applied2 to simulate the dispersion of 236U released from European nuclear fuel reprocess-3 ing plants in the North Atlantic and Shelf Seas using a published reconstruction4 of Sellafield and La Hague releases. Model results are in better agreement with5 observations if the lowest estimation of such releases are used. This implies that6 approximately40kgof236U have been discharged from Sellafield. It was found7 that adsorption of 236U on bed sediments of the shallow European Shelf Seas plays8 an essential role in its dispersion patterns. This contrasts strongly with the more9 ∗Corresponding author; e-mail: rp[email protected], phone +34 954486474, fax +34 954486436 1 conservative behavior of 129I in the same area. This has two important implications10 in the use of 236U as oceanographic tracer; i) special care must be taken in coastal11 areas, as sediments might act as sinks and sources of 236U; ii) the annual input12 function of 236U into the Arctic is not directly controlled by the annual discharges13 from Sellafield and La Hague, since sediments from the Irish, Celtic and North Sea14 modulate and smooth the signal. Only 52% of the total releases enter into the Arctic15 Ocean.16 Keywords: Lagrangian model; North Atlantic; 236-uranium; sediments; water tracer;17 Nuclear Fuel Reprocessing Plants18 1 Introduction19 The characterization of the Arctic water masses has become fundamental in the late years,20 e.g., to understand how climate change would affect circulation patterns in the region.21 New data and updated models are of primary importance for tracing transport pathways22 and features of the Arctic water masses. Anthropogenic radionuclides, such as 137Cs, 3H,23 99Tc and 129I, are being used as transient tracers of oceanographic processes.24 236U has been presented in the last years as a powerful new water mass tracer in25 oceanography, especially in the Atlantic Ocean and Arctic Ocean. It is almost entirely,26 but not exclusively, of anthropogenic origin. A total amount of 35 kg of mobile 236Uis27 estimated to come from natural sources, atmospheric nuclear weapons tests contributed28 to the total inventory with about 900 kg (Sakaguchi et al., 2009)). Finally, the Nuclear29 Fuel Reprocessing Plants (NFRP) of Sellafield, Springfields and La Hague have carried30 out significant liquid discharges to the sea. 236U contribution from these sources has been31 estimated to be about 95 kg (Christl et al., 2015a). However, these estimations still need32 to be narrowed down. In order to use 236U as a robust oceanographic tracer, it is essential33 2 to quantify accurately the contribution of the different sources and to better define its34 geochemical behavior. A full three-dimensional Lagrangian model which simulates the35 dispersion of 236U through the North Atlantic (NA) has been used to evaluate the annual36 discharges and the total amount of 236U released in the NA from the NFRP. Further-37 more, using the model it is possible to investigate, for the first time, the water/sediment38 interactions of 236U.39 The model was previously applied to study the dispersion of 129Iand137Cs released40 from the European NFRP (Villa et al., 2015; Peri´a˜nez et al., 2017, respectively) and41 its results were compared with measurements in both water and sediments. These are42 geochemically well-known radionuclides, with well-known discharges from the NFRP. It43 allowed us to evaluate the performances of the model. The next step has been to imple-44 ment the model to improve our knowledge of 236U.45 The model is described in detail in references cited above and, consequently, only46 a brief overview is provided in section 2. Model results are presented in section 3. A47 reconstruction of Sellafield releases was required to run the model. Results show that48 the lowest estimations of such releases should be used to have a better agreement with49 observations. This is described in section 3.1. An investigation on 236U/129Iratiosis50 presented in section 3.2, where it is shown that they must be carefully analyzed when51 used as water tracers, since 129I is significantly more conservative than 236U. Finally, a52 reconstruction of the 236U input function into the Arctic Ocean is presented in section53 3.3.54 2 Model description55 The model is a full three-dimensional Lagrangian dispersion model in which a radionuclide56 release is simulated by a number of particles, each of them equivalent to a number of units57 3 (atoms in the present application). The three-dimensional path followed by each particle58 is computed, turbulent diffusion being modelled as a three-dimensional random walk59 process. A constant typical value of 1.0×10−3m2/s (Elliott et al., 2001) has been used60 for the vertical diffusivity. In a previous work on 129I dispersion, it has been tested that61 model results are little sensitive to this parameter (Villa et al. 2015). The Smagorinsky’s62 scheme (Cushman-Roisin and Beckers, 2011) has been adopted to describe the horizontal63 diffusivity.64 The number of particles per water volume unit is computed to obtain radionuclide65 concentrations over the domain at the desired times and depths.66 Interactions between the dissolved phase and solid phases (suspended matter and bed67 sediments) are described through a dynamic approach. Thus, uptake/release of radionu-68 clides is considered to be described by a reversible reaction. This reaction is described by69 kinetic rates k1and k2as in previous works (Peri´a˜nez, 2008; 2012; Peri´a˜nez et al., 2013,70 among many others). A stochastic method was developed to solve the equations describ-71 ing these processes. Technical details may be consulted elsewhere (Peri´a˜nez and Elliott,72 2002; Peri´a˜nez and Caravaca, 2010; Villa et al., 2015; Peri´a˜nez et al., 2016). This method73 for describing water/sediment interactions has been used by other researches (Kobayashi74 et al., 2007; Min et al., 2013; Zhang and Battista, 2008) and successfully applied in model75 intercomparison exercises (Peri´a˜nez et al., 2015).76 The considered domain in the north Atlantic extends from 50oWto25 oE in longitude77 and from 45.1oN to 75.1oN in latitude (Fig. 1). Water circulation for the period of interest78 is required since currents are the main vector for the transport of tracers through advection79 processes. It has been obtained from JAMSTEC (Japan Agency for Marine-Earth Science80 and Technology) global ocean model. It is OFES (Ocean global circulation model For the81 Earth Simulator)1. A comparison of model performance with data in several regions of82 1http://www.jamstec.go.jp/esc/research/AtmOcn/product/ofes.htmlcite note-1 4 the global ocean (including the North Atlantic) may be seen in Masumoto et al., (2004).83 Horizontal resolution is 0.1oand there are 54 vertical levels, with increasing thickness84 from the surface towards the sea bottom. Monthly mean circulation has been used. As85 an example, surface circulation for January 2008 is shown in Fig. 1.86 The model has already been successfully applied to simulate the dispersion of historical87 radionuclide releases from Sellafield and La Hague nuclear fuel reprocessing plant (Villa88 et al. 2015; Peri´a˜nez et al., 2016). In particular releases of 129I, considering it as a89 perfectly conservative tracer, thus remaining in solution (Villa et al., 2015), and releases90 of 137Cs (Peri´a˜nez et al., 2016). In this last case exchanges of radionuclides between91 water, suspended matter and bed sediments were considered. Computed and measured92 137Cs distributions in water and bed sediments at several areas of the European Shelf93 and at different times were compared. Model results were in reasonable agreement with94 observations (Peri´a˜nez et al., 2016).95 Historical releases of 129Iand236U from Sellafield and La Hague for the period of96 interest (1952-2013, which is the time extent of the simulations) are presented in Fig. 2.97 In the case of 129I, release data is obtained from L´opez-Guti´errez et al. (2004); Sellafield98 (2014); Areva (2014). In the case of 236U information is more limited. La Hague facility99 has provided these releases with a few gaps. The full sequence has been reconstructed by100 Christl et al. (2015a). The authors have also reconstructed 236U releases from Sellafield101 based on total uranium discharges. Three different reconstructions were carried out and102 the mean value and standard error were provided in Christl et al. (2015a). Such mean103 value is the red line in Fig. 2. Upper and lower limits, taken as the uncertainty of the104 mean, are indicated by the dashed black lines. The smaller releases from Springfield105 reprocessing plant (UK) are also given in Christl et al. (2015a) and they are included in106 simulations described in this paper.107 5 3 Results108 The model was initially applied to 129I and compared with measurements (other than109 those already used in Villa et al., 2015). These are mean concentration in North Sea water110 in 2005 (Michel et al., 2012) and 2009 (Christl et al., 2015b); as well as a distribution111 map over this area in Christl et al. (2015b). Then the model was applied to 236Uand112 compared with mean concentration of this radionuclide in the North Sea in 2009 and113 its spatial distribution here (Christl et al., 2015b). Calculated and measured (Christl et114 al., 2015b) 129I/236U ratios were also compared. Subsequently the model was applied to115 evaluate the new 236U input function into the Arctic.116 3.1 236Uand129I dispersion117 Background concentrations of 236Uand129I due to nuclear weapon test fallout should118 be considered in order to compare model results and measurements. According to the119 estimation in Christl et al. (2015a), fallout background in the North Sea in 2009 (year120 and area where measurements are available), is about 107at/kg for 236U and about ten121 times smaller for 129I. These uniform background values have been directly added to the122 calculated concentrations associated to the reprocessing plants.123 Model simulations for 129I, considering it as a perfectly conservative radionuclide,124 were described in Villa et al. (2015). To compare its behavior with the results from 236U,125 additional results from such simulations are presented in this work. Thus, a comparison126 of the calculated and measured (Christl et al., 2015b) distribution in surface water of the127 North Sea in 2009 is presented in Fig. 3. The calculated distribution of 129I is in general128 agreement with the measured one. A plot of measured vs. calculated concentrations can129 be seen in Fig. 4. Due to the Lagrangian nature of the model, not all the grid cells in130 which measurements are available were occupied by particles. Thus, not all observations131 6 in Christl et al. (2015b) can be plotted. It is more adequate, in Lagrangian models, to look132 at general trends and averaged values over relatively larger regions than to measurements133 at given points. This is confirmed in the top panel of Fig. 5 where the temporal evolution134 of the calculated mean 129I concentration over the surface water of the North Sea is shown.135 The calculated mean is obtained as the total 129I content in the North Sea divided by136 the total water volume in the sea. Measured arithmetic mean concentrations, together137 with their 1σstandard deviations, in 2005 (Michel et al., 2012) and 2009 (Christl et al.,138 2015b) are also drawn. The calculated trend indicates an increase in concentrations in the139 beginning of the 1990s, due to the increase in La Hague releases (Fig. 2), and measured140 mean levels are well determined by the model. Although the model tends to overestimate141 measured concentrations when compared point by point (Fig. 4), general distribution142 patterns and mean values are in reasonable agreement with observations.143 The model, that works satisfactorily in the case of 129I, has been applied to 236U. No144 calibration for this radionuclide has been carried out and initially the only change is the145 magnitude of the 236U releases from Sellafield, La Hague and Springfield.146 Calculated and measured mean concentrations in surface water of the North Sea are147 presented in Fig. 5 (central panel). The blue and red lines correspond, respectively, to148 simulations made with the mean value and the lower limit of the Sellafield releases (red149 line and lower black dashed lines in the bottom panel of Fig. 2). It is clear that, even with150 the lower estimation of the discharge, the mean 236U concentration dissolved in the surface151 water of the North Sea in 2009 shown in the figure (75 ×106at kg−1)isoverestimated152 by the present parameterization of the model.153 The equilibrium distribution coefficient (kd) of natural uranium, is of the same order154 of magnitude as that of Cs (IAEA, 2004). And since there are not indications that155 236U and natural uranium behave differently in the ocean (Casacuberta et al., 2016 and156 7 references therein), a fraction of the released 236U might be removed by the sediments157 of the shallow shelf seas, as it actually happens for 137Cs. An additional simulation was158 carried out to test this hypothesis. Model parameters required to describe water/sediment159 interactions are described in several references (Peri´a˜nez, 2005; Peri´a˜nez, 2008; Peri´a˜nez160 et al., 2016). Again any model calibration has not been carried out. The model runs161 with the same parameters as for 137Cs (Peri´a˜nez et al., 2016) except the different releases162 from the reprocessing plants and the value of the radionuclide kd. Although the model163 describes radionuclide exchanges between the dissolved and solid phases in a dynamic164 way, the equilibrium kdis used to derive the adsorption kinetic rate as explained in detail165 in Peri´a˜nez et al. (2016) and references therein. The value recommended by IAEA (2004)166 for coastal waters has been used for the U kd:1.0×103(dimensionless). A summary167 of parameters required by the model to describe water/sediment interactions is given in168 Table 1. These parameters have been considered uniform over the whole model domain.169 This is an approximation, due to the lack of information about their variability over such170 a large region, which obviously will lead to some unavoidable errors in predictions. In171 spite of this, the general behaviour of 236U in sediments should be given by the model,172 since it was successfully applied to describe patterns of 137Cs in sediments with the same173 parameter set (except the kd).174 The result of this new simulation is presented in the central panel of Fig. 5 (black175 line). The “kinetics” simulation corresponds to the lower limit of Sellafield releases and176 includes interactions of the 236U between water and sediments. The mean 236U dissolved177 levels in the North Sea inferred from this new simulation is in better agreement with the178 experimental values.179 Additionally, the calculated inventories of 236U over the model domain in bed sediments180 in 2013 is presented in Fig. 6. It is shown that 236U is mainly absorbed in the sediments of181 8 parameter description value sediment mixing depth L=0.05 m particle density ρ= 2600 kg/m3 correction factor φ=0.1 sediment porosity p=0.6 desorption kinetic coefficient k2=1.16 ×10−5s−1 U distribution coefficient kd=1.0×103 particle radius 1.0×10−6m Table 1: Summary of model parameters involved in water/sediment interaction description. Parameter φis a correction factor which takes into account that not all the sediment particle surface is available to adsorb radionuclides since may be partially hidden by other particles. the shallow areas; this way relatively high 236U concentrations are present in the Celtic Sea182 sediments of the continental shelf, and in the Irish Sea and English Channel (where releases183 from Sellafield and La Hague occur, respectively). Releases from Sellafield are higher than184 releases from La Hague since the 1990s (Fig. 2). Thus, inventories in the Irish Sea reach185 the highest values. Furthermore, releases from both plants are transported by currents to186 the North Sea, thus sediments in this shallow sea also show significant inventories. Values187 in the Celtic Sea are in the order of 1011 −1012 at/m2. The results are in agreement with188 the 236U concentrations measured in two sediment cores collected in the area in 2015,189 which ranged from 10 to 20×1011 at/m2(Villa-Alfageme et al., unpublished). Calculated190 and measured values were higher than 236U concentrations measured in sediments from191 the North Atlantic Porcupine Abyssal Plain (PAP) site -4500 m depth, 49o00’ N, 16o30’192 W- (Villa-Alfageme et al. 2017). The reasons are i) 236U water concentration at the193 PAP site is significantly lower than at the North Sea since the contribution is mostly due194 to fall-out ii) PAP site is a deep-sea area, where particle concentrations and downward195 flux are significantly lower than at the European Shelf Seas. Furthermore, modelled196 and measured inventories at the Celtic Sea are an order of magnitude higher than the197 9 to surface cesium-137 dispersion in the Irish Sea. Journal of Nuclear Science and341 Technology 44(2), 238-247.342 Lepp¨aranta, M., Myberg, K., 2009. Physical Oceanography of the Baltic Sea.343 Springer, Berlin.344 L´opez-Guti´errez, J.M., Garc´ıa-Le´on, M., Schnabel, Ch., Suter, M., Synal, H.A.,345 Szidat, S., Garc´ıa-Tenorio, R., 2004. Relative influence of 129I sources in a sediment346 core from the Kattegat area. Science of the Total Environment 323, 195-210.347 Masumoto, Y., Sasaki, H., Kagimoto, T., Komori, N., Ishida, A., Sasai, Y., Miyama,348 T., Motoi, T., Mitsudera, H., Takahashi, K., Sakuma, H., Yagamata, T., 2004. A349 fifty-year eddy-resolving simulation of the World Ocean - preliminary outcomes of350 OFES (OGCM for the Earth Simulator). Journal of the Earth Simulator 1, 36-56.351 Michel, R., Daraoui, A., Gorny, M., Jakob, D., Sachse, R., Tosch, L., Nies, H.,352 Goroncy, I., Herrmann, J., Synal, H.A., Stocker, M., Alfimov, V., 2012. Iodine-129353 and iodine-127 in European seawaters and in precipitation from Northern Germany.354 Science of the Total Environment 419, 151-169.355 Min, B.I., Peri´a˜nez, R., In-Gyu Kim, Kyung-Suk Suh, 2013. Marine dispersion356 assessment of 137Cs released from the Fukushima nuclear accident. Marine Pollution357 Bulletin 72, 22-33.358 Peri´a˜nez, R., 2005. Modelling the Dispersion of Radionuclides in the Marine Envi-359 ronment. Springer-Verlag, Heidelberg.360 Peri´a˜nez, R., 2008. A modelling study on 137Cs and 239,240Pu behaviour in the361 Albor´an Sea, western Mediterranean. Journal of Environmental Radioactivity 99,362 694-715.363 16 Peri´a˜nez, R., 2012. Modelling the environmental behavior of pollutants in Algeciras364 Bay (south Spain). Marine Pollution Bulletin 64, 221-232.365 Peri´a˜nez, R., Elliott, A.J., 2002. A particle tracking method for simulating the366 dispersion of non conservative radionuclides in coastal waters. Journal of Environ-367 mental Radioactivity 58, 13-33.368 Peri´a˜nez, R., Caravaca, F., 2010. A set of rapid-response models for pollutant369 dispersion assessments in southern Spain coastal waters. Marine Pollution Bulletin370 60, 1412-1422.371 Peri´a˜nez, R., Casas-Ru´ız, M., Bol´ıvar, J.P., 2013. Tidal circulation, sediment and372 pollutant transport in C´adiz Bay (SW Spain): a modelling study. Ocean Engineer-373 ing 69, 60-69.374 Peri´a˜nez, R., Brovchenko, I., Duffa, C., Jung, K.T., Kobayashi, T., Lamego, F.,375 Maderich, V., Min, B.I., Nies, H., Osvath, I., Psaltaki, M., Suh, K.S., 2015. A new376 comparison of marine dispersion model performances for Fukushima Dai-ichi releases377 in the frame of IAEA MODARIA program. Journal of Environmental Radioactivity378 150, 247-269.379 Peri´a˜nez, R., Kyung-Suk Suh, Byung-Il Min, 2016. The behaviour of 137Cs in the380 North Atlantic Ocean assessed from numerical modelling: releases from nuclear fuel381 reprocessing factories, redissolution from contaminated sediments and leakage from382 dumped nuclear wastes. Marine Pollution Bulletin 113, 343-361.383 Sakaguchi, A., Kawai, K., Steier, P., Quinto, F., Mino, M., Tomita, J., Hosh, i, M.,384 Whitehead, N., Yamamoto, M., 2009. First results on 236U levels in global fallout.385 Science of the Total Environment 407,4238-4242.386 17 Sakaguchi, A., Kadokura, A., Steier, P., Takahashi, Y., 2012. Uranium-236 as a387 new oceanic tracer: A first depth profile in the Japan Sea and comparison with388 caesium-137. Earth and Planetary Science Letters 333-334, 165-170.389 Sellafield, 2014. http://sustainability.sellafieldsites.com/environment/environment-390 page/annual-discharge-monitoring-reports/391 Villa, M., L´opez-Guti´errez, J.M., Suh, K.S., Min, B.I.,Peri´a˜nez, R., 2015. The392 behaviour of 129I released from nuclear fuel reprocessing factories in the North At-393 lantic Ocean and transport to the Arctic assessed from numerical modelling. Marine394 Pollution Bulletin 90, 15-24.395 Villa-Alfageme, M., Chamizo, E., Santos-Arevalo, F.J., L´opez-Gutierrez, J.M., G´omez-396 Mart´ınez, I., Hurtado, S., in press. First comprehensive analysis of 236U in a marine397 core in the North Atlantic Ocean. Journal of Environmental Radioactivity.398 Zhang, Y., Battista, A.M., 2008. SELFE: A semi-implicit Eulerian-Lagrangian399 finite-element model for cross-scale ocean circulation. Ocean Modelling 21(3-4),400 71-96.401 18 −50 −40 −30 −20 −10 0 10 20 50 55 60 65 70 75 Longitude Latitude 0.5 m/s EGC NCC NAC North Sea Celtic Sea Norwegian Sea Irminguer Sea E. Channel • LH • SF •SP Figure 1: Top: Model domain, water depths in m. Bottom: Surface currents calculated by JAMSTEC model for January 2008 as an example. The main currents are the East Greeland Current (EGC), North Atlantic Current (NAC) and the Norwegian Coastal Current (NCC). Only one of each 25 vectors is drawn. LH, SF and SP denote La Hague, Sellafield and Springfields reprocessing plants. 19 1960 1970 1980 1990 2000 2010 0 50 100 150 200 250 300 129−I (kg/yr) Sellafield La Hague 1960 1970 1980 1990 2000 2010 0 1 2 3 4 5 236−U (kg/yr) Year Sellafield La Hague Figure 2: Annual releases of 129Iand236U from Sellafield and La Hague reprocessing plants. Dashed black lines indicate the lower and upper limits of the estimation of 236U releases from Sellafield. 20 Figure 3: Calculated, left side, and measured, right side, (Christl et al., 2015b) 129I concentrations in surface water of the North Sea in 2009. White areas in the sea in the calculated map indicate regions with fallout background concentration. 21 1091010 1011 1012 109 1010 1011 1012 Measurements Calculations 1061071081091010 106 107 108 109 1010 Measurements Calculations Figure 4: Calculated vs. measured concentrations (at/kg) in Christl et al. (2015b). Top and bottom panels: 129Iand236U respectively. 22 1950 1960 1970 1980 1990 2000 2010 0 5 10 x 104 129−I (106 at/kg) 1950 1960 1970 1980 1990 2000 2010 0 200 400 600 800 236−U (106 at/kg) lower limit mean value kinetics 1950 1960 1970 1980 1990 2000 2010 0 500 1000 1500 129−I/236−U Year Figure 5: Top: calculated and measured mean 129I concentrations in surface water of the North Sea. Middle panel: the same but for 236U. See the text for explanation of the legend. Bottom: Calculated and measured mean 129I/236U ratios in surface water of the North Sea. 23 Figure 6: Calculated 236U inventory (at/m2, logarithmic scale) in bed sediments of the model domain in 2013. 24 Figure 7: Calculated (left side) and measured (Christl et al., 2015b; right side) 236U concentrations in surface water of the North Sea in 2009 (top row). White areas in the sea in the calculated map indicate regions with fallout background concentration. 129I/236U ratio in surface water (bottom row). 25