Inter-comparison of dynamic models for radionuclide transfer to marine biota in a Fukushima accident scenario

Highlights

• Comparison of 7 dynamic models for radionuclide transfer in marine biota with the ERICA Tool.

• ⁹⁰Sr, ¹³¹I, ¹³⁷Cs in fish, crustaceans, algae and molluscs in a Fukushima scenario.

• Consistent pattern of delayed uptake and slow turnover by the dynamic models.

• Differences between ERICA and dynamic models increase with biological half-life.

• Significant variability between models linked to parameter and methodology differences.

Abstract

We report an inter-comparison of eight models designed to predict the radiological exposure of radionuclides in marine biota. The models were required to simulate dynamically the uptake and turnover of radionuclides by marine organisms.

Model predictions of radionuclide uptake and turnover using kinetic calculations based on biological half-life (T_B1/2) and/or more complex metabolic modelling approaches were used to predict activity concentrations and, consequently, dose rates of ⁹⁰Sr, ¹³¹I and ¹³⁷Cs to fish, crustaceans, macroalgae and molluscs under circumstances where the water concentrations are changing with time. For comparison, the ERICA Tool, a model commonly used in environmental assessment, and which uses equilibrium concentration ratios, was also used. As input to the models we used hydrodynamic forecasts of water and sediment activity concentrations using a simulated scenario reflecting the Fukushima accident releases.

Although model variability is important, the intercomparison gives logical results, in that the dynamic models predict consistently a pattern of delayed rise of activity concentration in biota and slow decline instead of the instantaneous equilibrium with the activity concentration in seawater predicted by the ERICA Tool. The differences between ERICA and the dynamic models increase the shorter the T_B1/2 becomes; however, there is significant variability between models, underpinned by parameter and methodological differences between them.

The need to validate the dynamic models used in this intercomparison has been highlighted, particularly in regards to optimisation of the model biokinetic parameters.

Introduction

Radiological protection of the environment (i.e. wildlife) is still relatively novel and exposure assessment methodologies for non-human biota are being continually improved. It is generally accepted that prediction of the uptake of radionuclides from the surrounding environmental media by organisms is a major source of uncertainty (Beresford et al., 2008).

The development of assessment approaches has focused on chronic exposure scenarios and, for aquatic biota, the majority of radiological assessment models assume that the activity concentration in an organism of mass M (i.e. A_O, in Bq kg⁻¹ expressed on a fresh mass (f.m.) basis) is proportional to the activity concentration (A_W, in Bq L⁻¹) in an adjacent volume V of water via a whole organism concentration ratio, or CR_wo (in L kg⁻¹ f.m.) (IAEA, 2014). The ERICA Tool (Brown et al., 2008) is an example of a model which represents the uptake of radionuclides from environmental media by these simple CR_wos. These methodologies are unlikely to assess reliably situations outside of equilibrium.

The truth is that, in reality, instantaneous equilibrium between biota and the medium does not exist. This is because biota accumulates radionuclides with a ‘time delay’ relative to variations of activity concentration in seawater. In its simplest formulation, the dynamics of the process are determined by a balance between the residence time of the radionuclide in the water in the presence of efficient hydraulic dilution, and the biological half-life (T_B1/2) of an organism. For a single component biological half-life, the activity concentrations in biota (A_O, Bq kg⁻¹) and water (A_W, Bq m⁻³) can be represented by a simple model with two rate constants; k_W for uptake and k_O for elimination: $\frac{d{A}_O}{dt}={\text{k}}_W{A}_W\frac{V}{M}-\left({\text{k}}_O+\lambda \right)A_O;\frac{d{A}_W}{dt}=-\left({\text{k}}_W+\lambda \right)A_W+{\text{k}}_O\frac{M}{V}{A}_O$.

Where $k_O=\frac{\ln 2}{T_{B1/2}}$, k_W = ((k_O + λ)M/V)CR_wo and λ is the radionuclide decay constant (Vives i Batlle, 2012). This type of model can be simplified by assuming that the water concentration does not depend on the exchange from an aquatic organism (because the amount of radioactivity in the organism is much smaller than in the surrounding volume of water, V) – hence dA_W/dt ≈ 0, and that the organism uptake rate does not change with time (i.e. ignoring the effect of organism growth).

Other dynamic models exist that are more complex and can, for example, model uptake by higher organisms via food (Brown et al., 2004, Keum et al., 2015, Maderich et al., 2014), requiring two additional parameters: assimilation efficiency and ingestion rate. Furthermore, some models consider organism growth processes requiring information on metabolism (Sazykina, 2000) and other models include more complex food web modelling (Heling et al., 2002).

The Fukushima nuclear accident has refocused strongly the vision for marine radioecology and highlighted the limited knowledge that we have in this area (Vives i Batlle, 2011). This disaster has brought some evidence that a dynamic modelling approach is advantageous compared with traditional equilibrium-based transfer approaches (Psaltaki et al., 2013, UNSCEAR, 2014, Vives i Batlle, 2014, Vives i Batlle and Vandenhove, 2014), owing to the relatively slow response of many biota to changing concentrations in seawater. Some models such as BURN-POSEIDON (Maderich et al., 2014), D-DAT (Vives i Batlle et al., 2008) and ECOMOD (Sazykina, 2000) have been applied in a ‘dynamic assessment’ context, including as part of the recent assessments of the impact of the Fukushima nuclear accident on marine biota in the acute phase (Tateda et al., 2013, Vives i Batlle et al., 2014), closely following initial application of equilibrium models to make predictions (Garnier-Laplace et al., 2011).

Notwithstanding the availability of some models for dynamic situations, the availability of parameterisation data is a problem. There are many knowledge gaps, especially concerning elemental biological half-lives, and there are several types of model in use ranging from simple linear first order kinetic approaches to metabolic and foodchain transfer models. To date, there has been no international comparison of dynamic models for estimating biota exposure. For this reason, we decided to perform the first systematic comparison between such models within the International Atomic Energy Agency (IAEA) MODARIA programme (http://www-ns.iaea.org/projects/modaria/default.asp).

The focus of this study was to compare activity concentrations and exposures to biota calculated by dynamic transfer models; the location chosen for this model simulation was close to the point where radionuclides were released from the Fukushima Nuclear Power Plant to the Pacific Ocean during the reactor accident in March/April 2011. We used seven dynamic models: BURN-POSEIDON, the ANL approach, D-DAT, ECOMOD, the IRSN approach, K-BIOTA-DYN-M and the NRPA marine dynamic model; all models are described and referenced in Section 2.1 below. The predictions of these dynamic models were compared with the output from the equilibrium-based ERICA Tool. The input for the intercomparison was a series of hydrodynamic forecasts or monitoring data (activity concentrations in seawater and sediment) for a site close to the Fukushima nuclear complex for the 110 days after the accident, produced by means of marine dispersion models, as referenced below.

The resultant estimates should be considered as illustrative only, and not as a thorough assessment of exposures and effects at this site close to the Fukushima NPP. Such an evaluation using both model prediction and monitoring measurements can be found elsewhere (Vives i Batlle et al., 2014). The present study is based on model comparisons for a single location in close proximity to the release point, and thus the calculated activity concentrations in water and sediments used in the present study represent only a limited area. This area is not representative of the general region inhabited by populations of biota, since the gradients of the activity concentrations for both water and sediments are very pronounced (UNSCEAR, 2014). This is why the discussion of the results is limited to the numerical differences between the models and does not include an evaluation of the levels of exposures and possible effects on biota.