The transport, effective half-lives and age distributions of radioactive releases in the northern Indian Ocean

Marine Pollution Bulletin 169 (2021) 112587 Available online 7 June 2021 0025-326X/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). The transport, effective half-lives and age distributions of radioactive releases in the northern Indian Ocean R. Peri´ a˜ nez a , * , B.I. Min b , K.S. Suh b a Dpt Física Aplicada I, ETSIA, Universidad de Sevilla, Ctra Utrera km 1, 41013 Sevilla, Spain b Korea Atomic Research Institute, Daedeok-Daero 989-111, Yuseong-Gu, Daejeon, Republic of Korea ARTICLE INFO Keywords: Indian Ocean Radionuclide Sediment Transport Model ABSTRACT A Lagrangian model which describes radionuclide transport in the northern Indian Ocean is described. Water circulation is obtained from HYCOM ocean model for year 2017. The model includes advection by currents, turbulent mixing and radionuclide interactions between water and sediments, described in a dynamic way using kinetic transfer coefficients. Hypothetical releases from five coastal nuclear power plants operating in the northern Indian Ocean were simulated. Releases were supposed to start both during the winter and summer monsoons, to study reversing circulation effects. Age distributions of releases were calculated, which adds information about circulation and radionuclide pathways. It was found that, for some of the NPPs, radionuclide distributions resulting from releases starting in both seasons were not as different as could be expected from the opposed circulation schemes during each monsoon. Effective 137 Cs half-lives in the ocean surface were calculated and results were two orders of magnitude below previous estimations. 1. Introduction The interest in improving models which simulate the behaviour and transport of radioactive elements in the marine environment increased since the Fukushima Dai-ichi Nuclear Power Plant (NPP) accident in 2011 in Japan, after which significant amounts of radionuclides were introduced into the Pacific Ocean directly from the NPP as well as deposited from the atmospheric releases resulting from hydrogen explosions in the NPP (Kobayashi et al., 2013). Such interest was embodied, for instance, in modelling programs launched by the International Atomic Energy Agency (IAEA): models were applied and compared in several release scenarios, not only to Fukushima releases into the Pacific (IAEA, 2019; Peri´ a˜ nez et al., 2016, 2019a, 2019b). The northern Indian Ocean is a unique ocean since it is affected by monsoons and thus large scale circulation is seasonal. Essentially, during winter in the northern hemisphere a high pressure system is formed in central Asia, which results in northeast winds over the ocean. This is the winter (or northeast) monsoon. This high pressure system weakens and in summer a low pressure system of thermal origin results, which reverses wind directions. This is the summer (or southwest) monsoon. The winter monsoon extends from November to February and the summer one from May to September. The remaining months are transition ones from one monsoon to the other. Such atmospheric forcing affects circulation in the Indian Ocean, which is analyzed in detail in, for instance, Schott and McCreary (2001) and Shankar et al. (2002). A brief description of circulation in the northern ocean is given later. Only two local-scale pollutant transport models for the northern Indian Ocean could be found in literature. A Lagrangian model was used to simulate tar ball transport in Goa coast in west India (Suneel et al., 2013). Patgaonkar et al. (2012) applied MIKE21 model to estimate the residence time of a small bay in northwest India. The transport of microplastics in the southern Indian Ocean was analyzed by Collins and Hermes (2019) using a Lagrangian model. Several NPPs are operating along the shores (or close to) of the northern Indian Ocean. They are Kanupp NPP in Pakistan and Kakparar, Tarapar, Kaiga, Kudankulam and PFBR NPPs in India (see map in Fig. 1); actually, Kakparar and Kaiga are located some tens of km inland, but by rivers which discharge in the nearby coast. However, models which could be applied to simulate the oceanic dispersion of radionuclides released due to a potential accident occurring in any of the NPPs could not be found in literature. Fukushima accident in Japan showed us the need of having radionuclide marine transport models prepared to operate in areas which could be affected by a potential nuclear accident (IAEA, 2019; Peri´ a˜ nez et al., 2016). The purpose of the present paper * Corresponding author. E-mail address: [email protected] (R. Peri´ a˜ nez). Contents lists available at ScienceDirect Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul https://doi.org/10.1016/j.marpolbul.2021.112587 Received 15 April 2021; Received in revised form 26 May 2021; Accepted 27 May 2021 Marine Pollution Bulletin 169 (2021) 112587 2 consists of filling this gap and describing a radionuclide transport model for the northern Indian Ocean. This model could be used to simulate radionuclide transport from a hypothetical accident, as was done in the case of Fukushima accident, but also can provide insight into dispersion patterns resulting from different release scenarios (different release location and season). Different diffusion schemes were tested and results compared. Age distributions of the released particles were also provided by the model and effective half-lives of radionuclides in the northern ocean were calculated as well and compared with previous estimations. Thus, the present model is not only a tool which may help to assess the aftermath of a nuclear accident, but it can also be used in a more processorientated investigation of the ocean. Essentially, the model is a three-dimensional Lagrangian transport model forced with water circulation obtained from HYCOM (Bleck, 2001) global ocean model. The model includes physical transport (advection due to currents and mixing due to turbulence), uptake/ release reactions between dissolved radionuclides and bed sediments, which are described in a dynamic way, and radioactive decay. It should also be commented that this model could be applied to other chemical pollutants (for instance heavy metals, ammonia etc.) simply setting radioactive decay to zero. The model is briefly described in Section 2. Results of the different numerical experiments are presented and discussed in Section 3. First, a brief description of circulation in the northern Indian Ocean is included in Section 3.1, then radionuclide simulations are presented in Section 3.2. Within this section, the radionuclide releases are described in Section 3.2.1, turbulence models are compared in Section 3.2.2, radionuclide transport simulations from the different NPPs and calculation of half-lives are described in Section 3.2.3 and, finally, age distributions are discussed in Section 3.2.4. Main conclusions are summarized in Section 4. 2. Model description 2.1. Baroclinic circulation HYCOM (Hybrid Coordinate Ocean Model [Bleck, 2001]) model was used to obtain circulation in the Indian Ocean. HYCOM is a primitive equation general circulation model with horizontal resolution equal to Δx =Δy =0.08◦and 40 vertical layers. The model web page (htt ps://www.hycom.org/) provides examples of applications over the world, as well as a detailed description of the model. In particular, HYCOM model is forced by winds, short and long wave radiation and freshwater inflow through the main rivers. The system uses the Navy Coupled Ocean Data Assimilation (NCODA) system (Cummings, 2005; Cummings and Smedstad, 2013). Daily currents of year 2017 were downloaded from HYCOM data server. This year was selected because there are not missing data for it in HYCOM data server and, also, any anomalous circulation feature was not observed in the northern ocean, thus it can be considered as a typical year. Nevertheless, it is worth to comment that an Indian Ocean Dipole (IOD) was observed in 2017 (Zhang et al., 2018). It mainly affects the southern hemisphere and consists of an ocean-atmosphere coupled mode in the tropical Indian Ocean which features a sea surface temperature anomaly gradient. This anomaly is quantified by the dipole mode index, the difference of area-averaged sea surface temperature anomaly between the western tropical ocean (50◦-70◦E, 10◦N-10◦S) and the southeastern tropical Indian Ocean (90◦-110◦E, Equator-10◦S). During a positive IOD index period anomalous easterly winds prevail along the equator and leads to severe flood in East Africa and drought in the Indonesian Archipelago (Zhang et al., 2020). However, after analyzing currents in the northern Indian Ocean (which is the domain of the model) in 2017, any significant effect of this IOD event could not be observed. 2.2. Radionuclide transport model As mentioned before, the model is Lagrangian or particle-tracking. The radionuclide release is thus simulated by means of a number of particles. Each particle is equivalent to a number of units (for instance Bq) whose trajectories are evaluated during the simulated time period. Transport processes are physical advection due to water currents and mixing due to turbulence. In addition, the model includes radioactive decay and interactions of radionuclides with bed sediments (adsorption/ desorption reactions). Radionuclide concentrations are evaluated from the number of particles within each grid cell and compartment (surface water, deep water and sediment). In a Lagrangian model advection is obtained from the following equation for each particle: Lx=uΔt+ ∂ Kh ∂ xΔt(1) Ly=vΔt+ ∂ Kh ∂ yΔt(2) where L x and L y are the changes in particle position; u and v are water velocity components at the particle position and depth and for the corresponding calculation time step, since currents change in time, downloaded from HYCOM model. Note that daily HYCOM currents are used. Derivatives of the horizontal diffusion coefficient (K h ) prevent the artificial accumulation of particles in regions were diffusion coefficients are smaller (Proehl et al., 2005). A first order accuracy scheme was applied to solve advection. However, Elliott and Clarke (1998) did not find improvements in results if a second order accuracy scheme were used. Moreover, turbulence masks small errors in the advection scheme (Elliott and Clarke, 1998). The maximum size of the horizontal step given by the particle due to turbulent mixing, D h , is (Proctor et al., 1994; Peri´ a˜ nez and Elliott, 2002): Dh= 12KhΔt √(3) in the direction θ =2 π RAN, where RAN is a uniform random number between 0 and 1 and Δt is time step in the Lagrangian model. This equation gives the maximum size of the step. The real size at a given time and for a given particle is obtained multiplying the equation by another independent random number. This procedure is required to ensure that a Fickian diffusion process (Proctor et al., 1994) is simulated. Time step used to integrate the Lagrangian model was set as Δt = 600 s. Similarly, the size of the vertical step is (Proctor et al., 1994; Peri´ a˜ nez and Elliott, 2002): Dv= 2KvΔt √(4) 50 60 70 80 90 100 5 10 15 20 25 30 Longitude Latitude • Kanupp • Kakparar • Tarapur • Kaiga • Kudankulam • PFBR Fig. 1. Map of the Indian Ocean, corresponding to the model domain, showing the locations of coastal NPPs operating in the area. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 3 given either upward or downward. In this equation K v is the vertical diffusion coefficient. Three different schemes were tested to describe turbulent mixing. The simplest model, denoted as UD (uniform diffusivities) uses uniform and constant values for K h and K v . Model denoted as SD uses a Smagorinsky scheme (Cushman-Roisin and Beckers, 2011) for K h and constant and uniform K v . Finally model SDV uses the Smagorinsky scheme for K h and K v is described as a function of depth, although remaining constant in time. In case of the UD model, horizontal and vertical diffusivities are respectively set as 10 m 2 /s (Peri´ a˜ nez, 2020) and 1.0 ×10 −5 m 2 /s (Elliott et al., 2001; Brovchenko and Maderich, 2021). This same value is used for K v in the SD model. The Smagorinsky scheme (Cushman-Roisin and Beckers, 2011) used in SD and SDV models describes the horizontal diffusion coefficient as a function of the grid spacing, Δx and Δy, and the horizontal shear of currents: Kh=ΔxΔy ( ∂ u ∂ x)2 +( ∂ v ∂ y)2 +1 2( ∂ u ∂ y+ ∂ v ∂ x)2 √(5) Finally, vertical diffusion in the SDV model is written following the description in Peri´ a˜ nez et al. (2019b) used to describe radionuclide transport in the Pacific Ocean after Fukushima accident: Kv=⎧ ⎨ ⎩ 10−3m2/sz<60 m 10−5m2/sz>120 m lineal function 60 <z<120 m (6) where z is water depth below the sea surface. This way increased turbulence in the mixed layer of the ocean is described, while it is smaller in the pycnocline and deep ocean. Using a constant mixed layer thickness is just an approximation to speed up calculations; of course this thickness is seasonally variable and its climatological mean ranges from some 30 to 100 m in the Indian Ocean (Schiller and Oke, 2015). The Smagorinsky scheme is used in this model as mentioned before. Of course, these mixing schemes are approximations focused at achieving the short simulation times required in nuclear emergency models. Actually, they were applied in the most recent radionuclide transport models used in the International Atomic Energy Agency (IAEA, 2019) to simulate Fukushima NPP releases in the Pacific Ocean. A review on more sophisticated models can be seen in Davies et al. (1997a, Table 1 Parameters used in the radionuclide transport model. Parameter Description Value ρ m Mineral particle density 2600 kg/m 3 por Sediment porosity 0.6 ϕ Sediment correction factor 0.1 k 2 Desorption rate 1.16 ×10 −5 s −1 d pic Surface layer thickness 100 m Fig. 2. Map of the Indian Ocean showing its division into FAO major fishing areas. Source: http://www.fao.org/fishery/area/search/en. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 4 1997b). Radioactive decay is solved with a stochastic method (Peri´ a˜ nez and Elliott, 2002). Decay probability is defined as: pd=1−e−λΔt(7) where λ is the radioactive decay constant. A new random number is generated. If RAN ≤p d the particle is considered to decay and it is removed from the computation. The interactions of radionuclides between water and the bed sediments are described in terms of a kinetic adsorption rate k 1 and a desorption rate k 2 ; and a stochastic method is applied as well. The probability that a dissolved particle is adsorbed by the sediment is: pa=1−e−k1Δt(8) If a new generated independent random number is RAN ≤p a , then the particle is fixed to the sediment. The probability that a particle which is attached to the sediment is released to water is: pr=1−e−k2ϕΔt(9) and the same procedure is used. ϕ is a correction factor that takes into account that part of the surface of sediments is hidden by the surrounding sediments. Thus, this hidden part is not interacting with the water. The number of units (in this case Bq) corresponding to each particle, R is deduced from the number of particles used in the simulation (NP = 200, 000) and the magnitude of the release M (total released activity): R=M NP (10) Then radionuclide concentration in the water surface layer of each grid cell C surf (i,j) (in Bq/m 3 ) is: Csurf (i,j) = RNsurf (i,j) ΔxΔydpic (11) where ΔxΔy is the cell surface, N surf (i,j) is the number of particles within the surface layer in cell (i,j) and d pic is the surface layer thickness (thus we count only particles at depth less than d pic ). The radionuclide inventory (Bq/m 2 ) in water below the surface layer (from d pic to the seabed) is given by: Ideep =RNdeep(i,j) ΔxΔy(12) where N deep (i,j) is the number of particles in cell (i,j) at depth larger than d pic . Finally, bed sediment radionuclide concentration (Bq/kg) in cell (i, j) is: Csed(i,j) = RNsed(i,j) ΔxΔyL ρ s (13) where N sed (i,j) is the number of particles in the bed sediment of cell (i,j), L is sediment thickness (set as 0.05 m) and ρ s is sediment bulk density: Fig. 3. Surface currents in the northern Indian Ocean, as downloaded from HYCOM data server, for summer (or SW) and winter (or NE) monsoons. Snapshots of surface currents corresponding to year 2017 for December 31th (winter) and August 1st (summer) are drawn. SC: Somali Current, WICC: West India Coastal Current, EICC: East India Coastal Current, WMC: Winter Monsoon Current, SMC: Summer Monsoon current. Only one of each 100 current vectors is drawn for more clarity. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 5 ρ s= ρ m(1−por)(14) where ρ m is mineral particle density and por is sediment porosity. A number of parameters are defined within the model code, whose values are selected from standard ones or previous works. These parameters are summarized in Table 1. The adsorption rate k 1 is deduced from the desorption rate k 2 and the radionuclide distribution (or partition) coefficient k d for ocean waters, which can be obtained from IAEA (2004), following the procedure described in Peri´ a˜ nez (2008) for instance. The desorption rate k 2 was obtained from the experiments in Nyffeler et al. (1984), value which was used in many other modelling works (see for instance Peria˜ nez et al., 2013 and the review in Peri´ a˜ nez et al., 2019a). In addition, ages of particles in the ocean were calculated. The age is defined as the time elapsed since a given particle was released into the sea, and may provide useful information about circulation in a marine region. It is a Lagrangian concept and is obtained simply attaching a clock to each particle, which is started in the moment when the particle is released. The age-averaging hypothesis, as introduced by Deleersnijder et al. (2001), was used: the mean age of particles which are within each grid cell is defined as the mass-weighted arithmetic average Fig. 4. Calculated water surface 137 Cs concentrations (Bq/m 3 ) in logarithmic scale with the three turbulent mixing models. These simulations correspond to the PBFR NPP summer accident. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 6 of the ages of the particles present there. The mass of particle i is defined as its radioactivity content R i , thus the mean age, 〈age〉, in a given grid cell is: 〈age〉 = ∑N i=1 τ iRi ∑N i=1Ri (15) where N is the number of particles within the cell and τ i is the reading of the clock attached to particle i. If the values of R i were the same for all particles, then the mean age is the arithmetic means of the clock readings: 〈age〉 = 1 N∑ N i=1 τ iif Ri=R∀i(16) The temporal evolution of the average radionuclide concentrations in the surface layer over several areas of the northern Indian Ocean are finally calculated to assess the potential effect of a given release in fishery areas defined by FAO (Food and Agricultural Organization of the United Nations). These areas are shown in Fig. 2. We consider region 1 (Bay of Bengal) in the eastern ocean, and the western ocean is divided into regions 3 (western Arabian Sea), 4 (eastern Arabian Sea) and the part of region 5 (Somalia) which is within the model domain (northern portion of this region). The marginal Red Sea and Arabian/Persian Gulf are not considered. Of course these averages over such large regions cannot be used to evaluate the impact of NPP accidents on fisheries during the acute phase of the accident, since very high concentrations are locally produced in the region close to the NPP and negligible values far from it. However, they can provide information on the general trends of radionuclide distributions at longer temporal scales, when a relatively high homogenization of the radionuclide patch can be expected. Actually, this methodology consisting of dividing the ocean space into relatively large boxes was recently used in Peri´ a˜ nez et al. (2019b) to study the impact of Fukushima NPP accident in the Pacific Ocean and to compare model predictions with measurements. In addition, these calculations allow the calculation of 137 Cs effective half-lives for such areas of the Indian Ocean. Some estimations of this figure were already provided by Povinec et al. (2005) and Sartandel et al. (2015). These authors consider that the average 137 Cs concentration in the ocean surface, C, appears to decrease exponentially with time and thus is described by the following equation: Fig. 5. Calculated 137 Cs concentrations in the surface water layer (Bq/m 3 ), inventory in the deep layer (Bq/m 2 ) and concentrations in bed sediments (Bq/kg) for releases occurring during the summer (left column) and winter (right column) monsoons from PFBR NPP. Logarithmic scales are used. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 7 dC dt = − FC,(17) where F is a flux coefficient. The effective half-life is then: T1/2=ln2 F,(18) and is calculated fitting the temporal trend of the radionuclide concentration in the ocean surface to an exponential decay function. 3. Results and discussion A brief description of water circulation in the northern Indian Ocean, based upon currents calculated by HYCOM model is given first. Next results of radionuclide transport simulations are presented. 3.1. Circulation in the northern Indian Ocean Typical surface water circulation schemes for both monsoons are presented in Fig. 3, as downloaded from HYCOM data server. The summer monsoon map corresponds to August 1st and the winter monsoon map is obtained in December 31th. Note that daily currents are used in the radionuclide transport model, as explained before, and these maps are snapshots of surface circulation at the mentioned days, as examples. The names of the main currents are included in the maps. Essentially, it is well known that currents are seasonally reversing. Thus, in winter the overall current direction in the northern ocean is to the west: the East India Coastal Current (see Fig. 3) flows to the south, the Winter Monsoon Current to the west in the south of India and the West India Coastal Current to the north along the western India coast. Finally, the Somali Current flows to the south along the eastern Africa coast. The pattern reverses in summer, with the Somali Current flowing towards the north and a general current direction to the east. Circulation in the Arabian Sea is anticyclonic in summer and cyclonic in winter (see for instance Schott and McCreary, 2001). This circulation scheme is reproduced by HYCOM ocean model, as can be seen in Fig. 3. It is interesting to point out that conditions for coastal upwelling in the Indian Ocean are different than those of the Atlantic and Pacific ones, where this process occurs along the eastern coasts. The strongest upwelling takes place in the Indian Ocean along Somalia and Arabia coasts, when the SW monsoon forcing produces Ekman transport away from such coasts (Tomczak and Godfrey, 2003). A detailed description of physical oceanography of the Indian Ocean, which is not the purpose of the present paper, may be seen for instance Fig. 6. Same as Fig. 5 but for releases occurring from Kanupp NPP. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 8 in Tomczak and Godfrey (2003), Shankar et al. (2002) and Schott and McCreary (2001). 3.2. Radionuclide transport 3.2.1. Radionuclide releases A number of numerical experiments were carried out to assess the transport of radionuclides from the four coastal NPPs (see Fig. 1) and under winter and summer monsoon conditions. A 137 Cs (radioactive decay constant 7.31 ×10 −10 s −1 ) release of 1 PBq lasting 90 days (at a constant rate) was supposed to occur at the sea surface. This amount is an example of the same order of magnitude as the 137 Cs direct release from Fukushima into the Pacific Ocean in the first three months after the accident (Kobayashi et al., 2013). Thus, this figure is just an example to simulate a realistic accident. The 137 Cs distribution coefficient k d was taken as 4.0 m 3 /kg, which is the recommended value for coastal waters by IAEA (2004) for this radionuclide. The open ocean k d (2.0 m 3 /kg) was not used since water/sediment interactions occur in the shallow coastal areas. Simulations were 365 days long; the winter monsoon accidental release was supposed to start on December 15th and the summer monsoon release on May 15th. 3.2.2. Comparison of mixing models First, results from the three mixing models were compared. A summer monsoon accident occurring in PFBR NPP was simulated using the UD, SD and SDV models. Resulting 137 Cs concentrations in the surface water layer after 90 days are presented in Fig. 4. The Smagorinsky scheme produces horizontal diffusivities larger than the 10 m 2 /s defined in the UD model. Thus, in the SD and SDV models there is a higher offshore transport of radionuclides than in the UD model. This implies a higher fixation of radionuclides on the coastal seabed sediments in this model (result not shown) than with SD and SDV models. The SDV model uses a higher vertical diffusivity in the surface layer than the other two models. Thus, in this model transport of radionuclides to the deep water layer is higher than in UD and SD models. Consequently, surface water radionuclide concentrations in the central ocean are higher with SD than with SDV model (Fig. 4). In contrast, radionuclide content in the deep water layer is higher with the SDV model (result not shown). Another significant difference between models is running time. For the 365 days long simulation, computation times on a desktop PC were respectively 32 min, 381 min and 397 min with the UD, SD and SDV models. The calculation of Smagorinsky diffusivities over the large computational domain and the 40 vertical levels is quite demanding. If the purpose of a simulation is the rapid assessment of the effects of Fig. 7. Same as Fig. 5 but for releases occurring from Kudankulam NPP. R. Peri´ a˜ nez et al. Marine Pollution Bulletin 169 (2021) 112587 9 an accident in the local area around the NPP, the UD model may be adequate: it provides a fast response and, due to the smaller diffusion coefficients, radionuclide concentrations in water and sediments over such local area tend to be overestimated, thus providing a conservative calculation (i.e., maximum possible values) which is a useful concept in radiological protection (IAEA, 2001). For a more refined evaluation, and also to assess transport over larger spatial scales, the SDV model is recommended. Note that a fast simulation would not be a requirement for such a large scale evaluation. 3.2.3. Radionuclide transport from the NPPs As mentioned in Section 3.2.1 two simulations were carried out for each coastal NPP, for summer and winter monsoons, lasting 365 days. Calculated 137 Cs concentrations in the water surface layer, inventory in the deep layer and concentrations in the bed sediments can be seen in Figs. 5 to 8. These maps are obtained after 90 days, i.e., just at the moment when releases stop. It can be first observed that, as a general result, contamination of the sediment is essentially limited to the coastal area close to the release point. Only in the case of Tarapur NPP releases (Fig. 8) a larger portion of seabed is affected by the release. Actually, this contamination extends to the border of the continental shelf, which is of the order of 100 km wide here, the widest in all the coast around Pakistan and India (it is similar only in the northern Bay of Bengal). The narrow continental shelf prevents the sediment contamination offshore. The second result is that radionuclide distributions in water resulting from releases starting in summer and winter monsoons are in general not so different as could be initially expected from the different circulation schemes (shown in Fig. 3). They are particularly similar in the case of Kannup releases (Fig. 6), thus this region of the northern Arabian Sea seems to be little affected by the seasonally changing monsoon circulation. They are also not very different in the case of Tarapur NPP releases (Fig. 8), at least in the coastal region. The most significant differences appear in the case of Kudankulam NPP releases (Fig. 7), in the most southern point of India. Currents are stronger in this region (Fig. 3) and radionuclides, once they reach the nearly flowing WMC or SMC, are efficiently transported to the west or east respectively. In the case of releases starting in summer, they are mainly transported to the central Bay of Bengal, while winter releases tend to flow northwards along the western India coast. Releases from PFBR tend to travel to the central Bay of Bengal, however the east (west) coast of Sri Lanka is contaminated in case of summer (winter) monsoon releases (Fig. 5). Generally speaking, the radionuclide distributions below the water surface layer are reproducing the situations in the surface. Thus, there is not a significant difference between the surface and deep circulation in Fig. 8. Same as Fig. 5 but for releases occurring from Tarapur NPP. R. Peri´ a˜ nez et al.