SCOPE 50 - Radioecology after Chernobyl, Chapter5, Radionuclide Aquatic Pathways

Aquatic environments occupy a major portion of the Earth's surface and, therefore, an understanding of the radionuclide and radiation exposure pathways within such systems is essential. The environments include rivers, lakes, estuaries, shelf seas, deep oceans, ice sheets and glaciers, groundwater and ground-ice. The aquatic environments all consist of an aqueous phase and a solid phase which is mainly sediment (particulates) in surface environments and the host bedrock in groundwaters. Living organisms, which make up the aquatic biota, can be involved in the geochemical cycle of radionuclides and play a role in their phase distribution. Radionuclides are present in the living and non-living components of each aquatic environment, both natural radionuclides of primordial and cosmogenic origin, and artificial radionuclides from nuclear and non-nuclear industrial wastes, accidental releases and nuclear weapons fallout (see sources in Chapter 1). Solid and liquid inputs of artificial radionuclides to the aquatic environments can be direct or indirect. Direct input mechanisms include fallout deposition onto the water surface, liquid discharges and releases from dumped solid wastes. Indirect inputs from secondary sources are also very important, as a result of the remobilization of contaminated material within an environmental compartment, e.g. erosion of soils contaminated by fallout radionuclides in a river catchment or of marine sediments contaminated by nuclear fuel reprocessing wastes. Indirect inputs also occur by transfer between environments of radionuclides in solution and on sediments.

Each radionuclide will be partitioned between the solid and solution phases. A variety of disparate processes may be involved in this partitioning, including sorption by an inorganic or organic environmental sediment particle, or the bedrock for groundwaters, precipitation

dissolution, colloid aggregation

disaggregation, microbial activity, and uptake into and release from the biota. The solid-solution partitioning of a radionuclide is a very important parameter describing its behaviour and, except for biological uptake, it can be defined by the distribution coefficient (K_d) or the numerically equivalent ratio (R_d) introduced by NEA/OECD (1983):

	Activity concentration of the solid phase (Bq kg^-1)
K_d =		(5.1)
	Activity concentration of the solution phase (Bq l^-1)

The widespread use of the K_d is due to the need for a simple parameter describing solid/solution distribution for the purpose of modelling the biogeochemical distribution of radionuclides. The concept of K_d implies true equilibrium and reversibility, and that the solid/solution ratio has no effect on K_d due to sufficiently low radionuclide concentrations. In practice, the first two requirements are frequently not demonstrated. Variations in K_d are likely to occur not only because of possible biological and colloidal effects but also due to changing solution and sediment chemistries and to the non-attainment of equilibrium.

K_d values quoted in the literature have been determined experimentally or from observation of the partitioning in the environment. Distribution coefficients vary by 9 orders of magnitude between different nuclides and 3 orders of magnitude for any particular one. This variation depends mainly on the solution composition and the nature of the solid substrate. It can be modulated by non-equilibrium conditions, resulting from slow kinetics, and by the effects of colloid or biological processes. Especially important in the aquatic environment are the differences in K_d due to the change in solution chemistry from seawater to freshwater, with K_d generally, but not always, higher in freshwater than seawater (Table 5.1). The fraction of the radionuclide present in each phase is often expressed in terms of the K_d and the solid (sediment) mass concentration (C_P), either as the ratio of the activity in the particulate fraction to that in the solution fraction (F_a):

F_a = K_d C_P

(5.2)

or by the percentage of radionuclide in the particulate fraction (P_P), per unit volume of suspension:

	100
P_P=		(5.3)
	[1+1/(K_d C_p)]

Sediment concentrations vary widely in aquatic environments, both spatially and temporally, e.g. from <1 to >1000 g m^-3. Characteristically, they are highest in high-energy environments such as shelf seas, tidal estuaries and rivers, especially during events which have low frequency of occurrence, such as storms. This variation means that the relative importance of the solid and solution phases in terms of radionuclide behaviour also varies widely in space and time. At one extreme, with high particulate concentrations, a large proportion of the radionuclides will be particle associated, even for low K_d elements. At the other extreme the converse will be true.


Element	Freshwater	Seawater

I	3 x 10²	10¹
Na	10²	10¹
Ru	10²	5 x 10³
Sr	10³	10²
Cs	10⁴	2 x 10³
Pu	10⁵	5 x 10⁴
Lanthanides	5 x 10⁶	5 x 10⁵

^aValues are the best estimates from Coughtrey et al. (1985).

The chemical characteristics of the water are important in determining the radionuclide behaviour and K_d. Radionuclide species in solution may depend on the ionic composition and ionic strength of the water, the presence of organic ligands, the redox state (Eh) and acidity (pH); an important contrast in nuclide behaviour is thus observed between aerobic and anaerobic waters. The major differences in ionic composition and ionic strength existing between saline water and freshwater, and within the latter between alkaline and acid waters, play an important role in determining radionuclide speciation and behaviour. Also, individual radioelements vary widely in the complexity of their solution chemistry, e.g. Pu may exist as three or four oxidation states and many ionic species, whereas Cs is a single monovalent ion.

The sediment grains (and bedrock grain surfaces) differ in their potential for adsorbing radionuclides, which results in a K_d variation with lithology, i.e. grain size and composition. In addition, the nature of the grain surface can be important, i.e. presence of organic and Fe/Mn oxide coatings. Minerogenic sediment grains, mainly silicate minerals or rock fragments, are derived from the crustal rocks and their derivative soils and sediments. These are eroded by rivers, in shallow seas and estuaries and are transferred from one aquatic environment to another. A proportion of this minerogenic matter is provided via atmospheric deposition. Another component is of biogenic origin, produced partly in the aquatic environment and partly introduced by terrestrial erosion. This includes skeletal minerals (mainly carbonates) and organic matter. Chemical precipitates form a hydrogenous component and can be locally important. Direct anthropogenic inputs of sedimentary material also occur, e.g. sewage solids. In addition, the solid phase may be primary, e.g. fuel debris. The size and density of these sediment grains must be taken into account for determining radionuclide behaviour. Typically, there is a wide range of grain sizes present in sediments. These include mud (silt + clay), which extends in size from an arbitrary boundary with the `solution' phase (usually 0.2 or 0.4 µm) up to 0.06 mm, sand (0.06

2 mm), gravel, etc. (> 2 mm). In addition, the presence of colloids of sub-micrometre size can be important for radionuclide pathways. Because of the increase in surface area per unit mass as grain size decreases, the fine-grained sediments have a higher adsorptive capacity for radionuclides than coarse, i.e. muds have higher activity concentrations than sands. Grain size also determines the potential mobility of the radionuclide particulate fraction. In terms of their dynamic behaviour, sediments are grouped into granular sediments (sand and gravel) and cohesive sediments (mud). The grain size is broadly related to the energy of the environment. In many situations, the fine-grained sediment is present not as individual grains but as biological aggregates (faecal pellets) and chemical flocs, which substantially change its dynamic behaviour. Occasionally, to facilitate comparison of activity concentrations of sediments with varying grain size, the concentrations are normalized to the Al, Sc or ⁴⁰Kcontent. This assumes that these elements are mainly contained in clay minerals and that their average concentration in these minerals remains roughly the same. These assumptions may not be valid in shallow marine and terrestrial aquatic environments, especially in glaciated terrains where Al- and K-bearing minerals may be common in coarse sediments (e.g. feldspars).

The characteristic features of radionuclide behaviour in the aquatic environment are the redistribution by transport of the solution and solid phases, the chemical interactions between the phases, and their biological cycling. Transport, in addition to leading to radionuclide redistribution, results in their dilution, fractionation and mixing, as well as vitally affecting the residence time in the aquatic environment. In addition, in the surface environments, accumulation of the particulate phase occurs by sediment deposition (sedimentation). This induces post-depositional chemical and physical changes, called diagenesis, which affect the radionuclide distribution.

The transport of the water results from a number of driving mechanisms, all ultimately a response to gravitational forces, modified by Coriolis and friction forces. Water velocities vary widely in space and time in aquatic environments, from as low as µm s^-1 in groundwaters to m s^-1 in rivers, tidal seas and estuaries and, also, in deep ocean turbidity currents. A basic contrast exists in surface waters between current regimes that, at least occasionally, are strong enough to erode sediment from the bed and those which are predominantly weak. The former include river flow, tidal currents and wave base oscillatory currents and the latter, ocean circulation, coastal and estuarine saline density currents and water surface elevation compensating currents, set up by wind shear and wave drift in coastal waters and lakes. Weak currents, however, can be important for transport of sediment introduced to the water column by another process. Transport of radionuclides occurs in these circulation/current/flow systems as a result of advection and dispersion. Advection is produced by the time averaged flow of water. Dispersion is due to a number of processes: molecular diffusion, turbulent eddy diffusion and dispersion due to velocity shear, i.e. the spreading that occurs in the direction of flow as a result of the vertical (and lateral) velocity gradients. The combined effects of the dispersion processes are described by dispersion coefficients applicable to the perpendicular axial directions: D_x, D_y, D_z. The magnitude of the dispersion coefficients varies with velocity, turbulence intensity and secondary characteristics of the different aquatic environments. Transport (flux) of a radionuclide by dispersion processes is related to the magnitude of the concentration gradient and the dispersion coefficient and takes place in the direction of the gradient.

Sediment and associated particulate phase radionuclides respond to the same advective and diffusive circulation processes as the solution phase, resulting in their transport and dilution. However, the sediment and particulate phase response is fundamentally different because a velocity-related threshold (bed shear stress) has to be exceeded before transport occurs, whereas this is not the case for the solution phase. Also, above this threshold, the concentration of mobile particles and their vertical distributions in the flow are dependent partly on velocity-related flow characteristics such as bed shear stress and turbulence intensity. These complex relationships to the flow depend on particle size and density. The state of aggregation of fine particles has a very important influence on this behaviour. As the capacity of the flow to transport sediment decreases, in space or time, above or below the erosion threshold, the excess sediment is deposited on the bed. However, because it takes a finite time for grains to settle through the water column there is a lag between the change in flow and deposition, which can have considerable importance for sediment behaviour. Settling rates vary from µm s^-1 to cm s^-1. Sediment can be transported either in suspension (suspended load), at velocities comparable to that of the water, or in contact with the bed (bedload) at a fraction of water velocity, as mobile bed forms such as ripples, dunes and bars. The sediment in suspension may be actively suspended by turbulence or may settle passively through the water after being introduced by another mechanism. The range of velocity regimes that can be found in aquatic environments means that either one mode of sediment behaviour may predominate, e.g. grain settling in oceans and lakes, or conditions may vary in space or time between various transport modes, e.g. in rivers. An important modifying factor is the availability of sediment: supply limitations may over-ride flow-imposed limits on sediment transport. Net deposition of sediment and associated radionuclides occurs where rates of sediment supply exceed those of transport. Such sediment deposits are extremely important as sinks or reservoirs for radionuclides. The radionuclide inventories preserved in aquatic sediment deposits vary widely, depending on radionuclide sources and on the deposit characteristics, such as lithology, sedimentation rates and residence times. The grain size dependence of sediment transport processes leads to the fractionation of sediment inputs and the production of deposits of different grain size and, hence, of different radionuclide concentrations. Certain environments are characterized by the lithological uniformity of their deposits, e.g. the mainly fine muds of oceans and lakes, whereas others, typically, have a more diverse range of sediment types depending on the local current regime, e.g. rivers and estuaries. Sediment accumulation (sedimentation) rates vary widely, from < mm y^-1 to > m y^-1 and usually correlate with grain size, with low rates in oceans and lakes and variable rates in other environments.

The sedimentation rates can be determined from radionuclide profiles. This can involve conventional radiometric methods based on the decay of an unsupported natural or artificial radionuclide such as ¹⁴C or an unsupported fraction, e.g. excess ²¹⁰Pb. Alternatively, artificial radionuclide profiles can be matched with the record of release of the nuclide, e.g. ¹³⁷Cs from weapons test fallout and, increasingly of use in the future, from the Chernobyl accident. The application of these methods can be affected by processes such as diagenetic, chemical or physical redistribution of the radionuclide and, for artificial radionuclides, the effects of processes which modify the relationship between the initial release to the environment and the signal received in the sediment, such as mixing of sediments labelled at different times. The residence times for sediment deposits also vary widely, from >100 to < 1 y. In short-lived deposits, sediment remobilization leads to reintroduction of particulate radionuclides to the water column. If subaqueous, the deposits are saturated and the pore waters will contain solution phase radionuclides.

In contrast to the high energy environment of coastal zones, transport of radionuclides in the open ocean is generally mediated by biological cycling. The various materials that have adsorbed or incorporated these nuclides, such as biogenic debris and clay, are packaged by zooplankton and can be transferred to the sediment in a few months. Other biological processes, such as aggregation of phytoplankton material, also lead to a rapid transfer of radionuclides to great depths. In the water column, and sediment and bedrock porewaters, the biogeochemical interactions listed earlier can lead to repartitioning of radionuclides between the solid and solution phases, in response to changes in the chemical environment and/or to biological mediation. Particularly important in surface waters are sorption

desorption reactions involving the sediment-bound radionuclides, caused by changes in salinity or redox state, or to the introduction of unlabelled or partially labelled sediment. In addition, physical change in the particulate phase can lead to the same result, e.g. by colloid aggregation/disaggregation or by organic particle degradation. Gain by the solution phase will increase the mobility of a radionuclide.

Transport by diffusion can occur also across sediment/water column interfaces, with loss or gain of activity by the deposits. Other post-depositional (diagenetic) processes in sediments can lead to the physical disturbance or modification of the original sediment and include the effects of biological disturbance by organisms, called bioturbation, and physical processes such as slumping. The effects of bioturbation depend on the number, size and habits of the organisms; they include random mixing of sediment or grain size sorting. These processes also affect the chemical behaviour of radionuclides, mainly via oxygenation of interstitial waters in the burrows and changes in alkalinity due, for instance, to carbon consumption.

Advective movement of porewaters and solution phase radionuclides can occur in sediments (in addition to bioturbation) as a result of consolidation due to the lithostatic stress generated by the overlying sediment. This results in an increase in sediment bulk density, decrease in pore volume and upward migration of porewater. In sediments, and importantly in groundwaters, porewater flow is also a response to regional hydrostatic gradients.

Radionuclide uptake by biota occurs by a number of mechanisms from both solution and particulate phases. Uptake by the primary producers, e.g. phytoplankton, occurs from solution by surface adsorption and metabolic processes. Surface contamination by particulates can also occur with macro-algae. The primary uptake mechanism for invertebrate and vertebrate organisms is ingestion of food. However, for the many invertebrates which are detrital filter- and sediment-feeders, this directly involves particulate radionuclides in general. Respiration also involves intake of solution phase (and particulate) radionuclides. Radionuclide contamination of terrestrial organisms can also occur by feeding on aquatic organisms. Uptake will depend not only on the organism concerned but also on the element involved and its activity concentration. The ratio between the activity concentration in the organism and aquatic environment is defined as the concentration factor.

These different processes affecting the behaviour of radionuclides in aquatic media are illustrated in the following sections.

5.2 FRESHWATERS

Solute transport in rivers is generally described in terms of a one-dimensional, partial differential equation, incorporating advection and eddy diffusion terms, although alternative statistical or systems approaches are receiving increasing attention (Young, 1990). Because of the complex nature of river channel geometry, transport models require detailed calibration. Artificial radionuclides can be used as tracers for studying river flows and for model calibration. Tritium is ideal for this purpose but is rarely used in practice. The main reason for this is the generally dispersed nature of the source term (bomb fallout), with a well-defined point source being the exception. An example of the latter is where tritiated water from the Grafenrheinfeld nuclear power plant has been used to define the dispersion characteristics of a 320 km section of the Main River, Germany (Krause and Mundschenk, 1989). Flow times, flow velocities and longitudinal dispersion (eddy diffusion) coefficients (20

200 m² s^-1) were determined as a function of river discharge from 0.4 to 4 times the mean runoff.

While there is an extensive literature on the relations between the hydraulic characteristics of flow and sediment particle behaviour, there is no coherent, mechanistically based approach to describing sediment transport and dispersion in river systems. A major contributing reason is that sediment transport by rivers is subject to non-hydraulic as well as hydraulic controls. Important non-hydraulic factors include the geology and soils present in the catchment, catchment topography, hydrology, land use and vegetation cover. A further complication is that many of the hydrological factors are stochastic in nature. These include storm duration and spatial effects, rainfall intensity, and antecedent discharge, all of which influence storm-period particle transport. There are also significant stochastic processes occurring within the river channel, e.g. an important source of sediment supply is through bank collapse.

Studies of artificial radionuclides in rivers have frequently highlighted the complex nature of sediment transport and the individuality of rivers. The Great Miami River, Ohio, has been studied over a 25-fold range of river flows and a 6-fold range of particle concentrations (Sprugel and Bartell, 1978). The activity concentrations of ^239,240Puin solution (Bq m^-3 ) and in suspended particles (Bq kg^-1) did not correlate with either river flow or suspended particle concentration. The total Pu inventory was largely a function of suspended particle concentration. The mean concentration of ^239,240Pu in riverborne suspended sediment was 2 to 3 times that in the source material (arable soils containing fallout Pu). The enhanced activities were attributed to size fractionation in the catchment-river system, resulting in a greater fine-grained component in the river. Studies of periodic, pulsed inputs of plutonium to the Great Miami River from the Mound Laboratory in Miamisburg, Ohio, have also given important insights to the dispersion behaviour of Pu (Muller et al., 1977). Mass balance calculations showed that under typical summer flow conditions, ~60 per cent of the effluent is lost through sedimentation within 10 km of the discharge point, and that resuspension of this material between pulses maintains a high `background' ²³⁸Pu flux in the river. The remaining ~40 per cent is transported downstream with each pulse.

Further understanding of the temporary nature of channel sediments comes from a study of the river Danube at Bezdan, near the Yugoslav

Hungarian border (Conkic et al., 1990). The concentrations of ¹³⁷Cs,¹³⁴Cs and ¹⁰⁶Ru in bottom sediments were studied following the Chernobyl accident. Decay-corrected activities showed exponential decreases with time that were the same for all three radionuclides, despite expected differences in chemical behaviour. This highlights the dynamic nature of the sediment deposits and suggests that the relative importance of physical processes was the same for all three radionuclides. Deployment of a sediment trap, higher up the Danube, at Vienna, provided evidence not only of a gradual reduction in Chernobyl activity with time in newly deposited sediments, related to the gradual downstream migration of the most contaminated sediments, but also of marked seasonal variations in activity (Maringer et al., 1989). As expected, the total activity content was influenced by the grain size composition of the sediments, which in turn was related to river discharge. The activity concentration of the < 20 µm fraction was between two and four times that in the bulk sediment. High discharges, related to the passage of spring meltwaters, resulted in a lowering of the ¹³⁷Cs activity in the newly deposited sediment, due in part to an associated coarsening in grain size but also due to a decrease with time in the radionuclide content of the fine-grained fraction (Figure 5.1). The subsequent fall in river flow produced a fining of the sediment in transit and a consequent increase in its radionuclide concentration.

Figure 5.1 Relation between river discharge and the seasonal trend in radiocaesium activities in total sediment and the <20 µm fraction. From Maringer et al. (1989).

Studies, using Chernobyl-derived radiocaesium as a tracer, have also begun to tackle some of the more intractable aspects of the dispersion of riverborne sediments, such as those related to bedload transport and the formation of floodplain deposits (e.g. Walling and Bradley, 1988).

Mass transport of a solute through and within a lake may also be described in terms of advection and eddy diffusion, but the complexities of water motions are far greater than those in rivers. Dynamic processes can be classified according to origin: wind-induced currents including seiches, inflow-induced currents, and convective currents. For large lakes there are, in addition, significant currents produced by Coriolis force, gravitational forces, frictional forces, and the effects of meteorological variations. The need to take account of the various types of current likely to be discernible and the distinction between the advection and eddy diffusion terms are very much dependent on the time and length scales of interest. A particularly important feature of lakes is the degree and temporal dependency of vertical density stratification, which generally is related to temperature but can be related to differences in dissolved salt content (i.e. in meromictic lakes). Transport by molecular diffusion is rarely important in the water column, except possibly associated with the boundary layer (~1 mm) overlaying the bottom sediments and vertically in the thermocline region.

Artificial radionuclides have rarely been used in studies of water movements in lakes. A contributing factor is that the complex temporal variation of the input function of fallout ³H, the radionuclide most frequently used in such studies, occurs on similar timescales to typical mixing times in lakes, thus greatly complicating interpretation. In a study of Lake Tahoe, no significant variation with depth in fallout concentrations was observed, despite it being a deep lake (max. 500 m) with a long hydraulic residence time (Imboden et al., 1977). Far greater success has been achieved through employing experimental injections of tritiated water (Quay et al., 1980).

A more widely applicable approach is to measure the parent

stable daughter tracer pair, ³H

³He (Torgersen et al., 1977). The method is based on the fact that during periods of isolation, water bodies build up an excess of 'He, relative to its equilibrium concentration with respect to the atmosphere, due to the in situ decay of ³H. Consequently, an effective water mass age can be determined and this has been used to calculate (a) gas exchange rates between surface waters and the atmosphere or, more specifically, the piston velocity and hence the surface boundary layer thickness; (b) the extent of degassing or gas renewal at turnover; (c) vertical diffusivities in the well-mixed surface waters; (d) transport across the thermocline during stratification; and (e) renewal times for water masses at different depths. Vertical eddy diffusivities of 0.6

3.6 cm² s^-1 have been estimated for the well-mixed surface waters of Lakes Erie, Huron and Ontario, and an apparent vertical diffusivity of 1 cm² s^-1 for the thermocline region in Lake Constance. The data indicate also that, whereas degassing of the deep waters is virtually complete during turnover in the three Great Lakes, a degassing rate of only ~50 per cent per year occurs in the deep waters of Lake Constance.

Accidental releases of elevated ³H from nuclear power plants have been used as tracers. For example, releases from Douglas Point and Bruce A power stations in Ontario have been used to study the counter-clockwise circulation of Lake Huron (Veska and Tracy, 1986).

In contrast to rivers, lakes are generally regarded as being efficient and permanent sediment traps, related principally to their greater depths, smaller currents and longer hydraulic residence times. Thus lakes are short-lived on the geological timescale, disappearing from the landscape principally through the action of sedimentation. In all but very shallow lakes, bottom sediments away from the high-energy shoreline region are normally subjected to low current velocities and negligible wave action. The transport and fate of sediments within lakes, however, is far from simple because of the many competing processes regulating erosion, transport and deposition. In areas dominated by river action, grain size and the rate of sedimentation generally decrease logarithmically with distance from the river mouth. The distance over which river flow is important varies greatly, depending on factors such as basin topography, stratification, and river discharge. At a distance from the river mouth, wind and wave action become the dominating influence. The associated rate of net sedimentation increases with water depth from zero at some intermediate depth, due to sediment focusing, and involving the resuspension of shallow-water sediments and their transport to deeper waters.

The transport of sediment particles in lakes is studied predominantly through an examination of bottom sediments, rather than direct observation in the water column. Artificial radionuclides with high particle affinities have proved to be useful as tracers in such studies, especially as dating tools providing a measure of sedimentation rates. The first appearance of fallout ¹³⁷Cs, and to a lesser extent ^239,240Pu, in the early 1950s and peak fallout in 1963 have become particularly important stratigraphic markers in lake sediments. In Europe, ¹³⁷Cs resulting from the Chernobyl accident has become a useful additional time marker (1986). Edgington and Robbins (1975) employed the bomb fallout markers, together with ²¹⁰Pb dating, to determine sedimentation rates in Lake Michigan and found good agreement between the three methods. The sedimentation rate in the southern part of the lake varies by more than an order of magnitude from virtually zero in large areas of the western side to >50 mg cm² y^-1 in areas close to the eastern shore, where several large rivers discharge (Figure 5.2a). Radionuclide fluxes at the sediment surface were normalized to fallout fluxes at the air

water interface, which represents the predominant source term. A map of the distribution of this function provides a measure of the relative sediment loss or gain at any point (Figure 5.2b). Particularly high sedimentation occurs offshore to the north and west of the major rivers, whereas bottom topography slopes continuously down to the central axis of the lake. No sedimentation occurs at water depths <50 m, even near the river mouths. This pattern of deposition, which is very different from the classic picture of focusing and delta formation described above, was attributed to the hydrodynamic characteristics of the lake and in particular the pattern of lake-wide currents.

More than 95 per cent of the ¹³⁷Cs and ^239,240Pu that has entered southern lake Michigan as fallout now resides in the sediments, indicating that the efficiency of sediment trapping is very high. This is not always the case, however, and studies with artificial radionuclides have provided data to assess the efficiency of trapping and its main controlling factors. Fallout^239,240Pu has been studied in the sediments of seven lakes in eastern Ontario, Canada, with a large range of mean depths (2 to 19 m) and hydraulic residence times (0.03 to 9 y) (Cornett and Chant, 1988). From 28 to 100 per cent of the ^239,240Pu input to the lakes was retained in the sediments, with the percentage retained being correlated with the hydraulic residence time. In other words, in the lakes with short hydraulic residence times, a significant fraction of the Pu and hence sediment inputs must have remained in the water column long enough or have been resuspended frequently enough to be lost via the outflow. The data also indicated the absence of any sediment focusing, which the authors linked to the shallow nature of the lakes.

Figure 5.2 Sediment characteristics in southern Lake Michigan: (a) mass sedimentation rate (mg cm^-2 y^-1); (b) flux normalization factor for ¹³⁷Cs (sediment flux/flux at lake surface). From Edgington and Robbins (1975); reproduced by permission of the IAEA.

Similar situations exist in Lake St Clair, located between Lakes Huron and Erie, and in a number of man-made reservoirs in the Susquehanna river valley. The latter study (Donoghue et al., 1989) was based on a range of particle-associated tracers (⁶⁰Co, ¹³⁴Cs, ¹³⁷Cs and ⁶⁵Zn), derived from the nuclear power plants at Three Mile Island and Peach Bottom. It is apparent, therefore, that shallow lakes with short hydraulic residence times may be regarded as being intermediate in character between river channels and more typical lake basins in terms of their sediment dynamics.

Artificial radionuclides have also been used as sediment tracers to study other, diverse sedimentological problems. One example involves the sedimentological regime that exists at the eastern end of Lake Erie, which is a high-energy region with a substrate comprising bedrock and coarse-grained sediment. Fine-grained sediment is thought to be transported northwards to the Niagara river, and subsequently Lake Ontario, by the dominant advective flow in this part of the lake. An important radionuclide source, the Western New York Nuclear Service Center (WNYNSC), is located about 50 km south of Buffalo, NY. Water from the site eventually drains into Cattaraugus Creek and thence Lake Erie. Radionuclide inventories and ratios, involving ²³⁸Pu, ^239,240Pu, ²⁴¹Am, ⁵⁴Mn,⁶⁰Co, ⁶⁵Zn,¹⁰⁶Ru,¹³⁴Cs and ¹³⁷Cs have been used to demonstrate the presence of WNYNSC radionuclides in Lake Ontario sediments off the mouth of the Niagara River (e.g. Joshi, 1988). In this region, about 36 per cent and 80 per cent of the 1982 sediment inventories of ^239,240Pu and ²⁴¹Am respectively, are derived from WNYNSC, the remainder coming from atmospheric fallout. The studies also show that nearly all the Pu, ²⁴¹Am and ¹³⁷Cs activity, derived from WNYNSC, ends up in Lake Ontario sediments around the mouth of the Niagara River. These findings are an important illustration of the highly specific nature of artificial radionuclides as tracers. More particularly, they demonstrate that transport processes in surface waters can result in locally concentrated radionuclide activities at sites remote from source, even when separated by a lake basin, which would normally be associated with efficient removal of particle-associated radionuclides.

Another example involves the use of ¹³⁷C s and ²¹⁰Pb profiles in sediments to understand the processes operating in a series of small arctic lakes with depths < 10 m (Hermanson, 1990). Interpretation is based mainly on the inventory in each core relative to decay-corrected fallout. Sediment redistribution processes are shown to vary widely over short distances and include ice rafting effects.

The occurrence of groundwater may be divided into zones of aeration and saturation. The former zone consists of pore spaces occupied partially by water and partially by air. In the latter zone all pore spaces are filled with water under hydrostatic pressure. Over most of the land masses of the earth a single zone of aeration overlies a single zone of saturation and extends upward to the ground surface. In the absence of overlying impermeable strata, the upper surface of the zone of saturation is the water table. This is defined as the surface of atmospheric pressure and would be revealed by the level at which water stands in an open well. Water within the ground moves downwards through the unsaturated zone under gravity, whilst in the saturated zone it moves in a direction determined by the local hydraulic situation. Most natural groundwater discharge occurs as flow into surface water bodies but also occurs via the land surface, including transpiration from vegetation.

Due to advection, non-reactive solutes are carried at an average rate equal to the average linear velocity of the groundwater. However, there is a tendency of the solute to spread out from the path expected due to the advection hydraulics of the flow system. This is called hydrodynamic dispersion and causes dilution of the solute. It occurs because of mechanical mixing during fluid advection and because of molecular diffusion. The latter is significant only at low velocities. Dispersion caused entirely by the motion of the fluid is known as hydraulic dispersion. Dispersion is a mixing process and, qualitatively, has a similar effect to turbulence in surface-water regimes. For porous media, the concepts of average linear velocity and longitudinal dispersion are closely related. Longitudinal dispersion is the process whereby some of the water and solute molecules travel more rapidly than the average linear velocity and some travel more slowly. Thus, the solute spreads out in the direction of flow and declines in concentration.

Generally, tritium can be used to indicate the presence of young groundwaters (less than 30 years). At high continental latitudes, waters having more than about 4 TR fall in this category, while in low and middle marine latitudes the limit is nearer l TR. As a general guideline, it can be said that waters containing over 10 TR contain a thermonuclear test contribution, while 20 TR or more would suggest a component of groundwater recharged since 1961 (IAEA, 1983). On a local scale, ¹⁴C can be used as a tracer of transfer processes from surface systems to either plants or groundwater. A large number of ¹⁴C groundwater studies exist in the scientific literature (e.g. Fontes and Garnier, 1979).³⁶Cl has virtually ideal properties as a tracer for solutes in groundwater and soil water. ³⁶Cl from the Gnome event near Carlsbad, New Mexico (the first nuclear detonation of the Plowshare series, detonated 10 December 1961) illustrates how ³⁶Cl can be used to study the redistribution of radionuclides in the soil profile (Phillips et al., 1990). Atmospherically derived ⁸⁵Kr activity is used in hydrology to identify admixtures of young groundwaters recharged during the last 35 years. The use of ⁸⁵Kr is particularly attractive in combination with ³H, which has a similar half-life but a different history of release.

There are a number of examples where tritium contamination from a localized source has been used to trace groundwater movement and to study surface water

groundwater interactions. Liquid-waste effluents are routinely discharged to the ground at the Hanford Site in Washington State (see site description in Chapter 2). Adsorption, chemical precipitation, and ion exchange attenuate or delay the movement of some radionuclides, such as ⁹⁰Sr, ¹³⁷Cs, and ^239,240Pu. Other radionuclides, such as tritium, ⁹⁹Tc, and ¹²⁹I, are not as readily retained by the soil. These radionuclides move through the soil column at varying rates and eventually enter groundwater. Radionuclide concentrations are reduced by dilution when they reach groundwater. The more mobile constituents move downgradient in the same direction and at a rate nearly equal to groundwater flow.

The maximum extent of radionuclide contamination in the groundwater beneath the Hanford Site can be defined using tritium. Figure 5.3 shows the distribution of tritium in the unconfined aquifer during 1989 (Evans et al., 1990). The highest groundwater tritium concentration measured in 1989 was over 5 x l 0⁶pCi l^-1 in a well in the 200-East Area. The large tritium plume from the 200-East Area moves generally east to the Columbia River, where it discharges. Separate tritium pulses, associated with two major episodes of the Plutonium

Uranium Extraction Plant (PUREX) operations, can be distinguished as lobes with concentrations exceeding 2 x 10⁵ pCi l^-1. The lobe near the Columbia River is a result of discharges to ground during the operation of PUREX from 1956 to 1972. The narrow lobe extending several kilometres south-east of the 200-East Area represents discharges from PUREX between 1983 and 1989.

The travel time of contaminated groundwater from source areas to the Columbia River and the mass of radioactive contamination that is reaching the river have not been determined with a high degree of accuracy. A recent estimate by the US Geological Survey (1987) stated the average travel time is in the range of 10 to 20 years. Contaminants also enter the river along the Hanford Reach as direct effluent discharges. In 1989, the average tritium concentration upstream of the Hanford Site was 63 pCi l^-1, while the average downstream concentration was 129 pCi l^-1, thus substantial dilution occurs when groundwater, containing tritium at concentrations greater than 2 x 10⁵ pCi 1^-1 discharges into the river.

Seasonal rises in the Columbia River stage result in bank storage, which affects groundwater levels and contaminant concentrations in wells near the river. In the eastern part of the site, where tritium exceeds 2 x 10⁵ pCi l^-1, the concentration of tritium in groundwater from wells near the river has been documented to fluctuate as much as a factor of four over the course of a year. The highest concentrations correspond to periods of low river stage. The tritium concentrations are lower when the river stage is higher, during which times bank storage effects from the river dilute the tritium concentrations in the groundwater system.

Figure 5.3. Distribution of tritium on the Hanford Site. From Evans et al.

Another radionuclide migration project was started in 1974 at the Nevada Test Site to determine the potential for movement of radioactivity away from underground nuclear explosions (Ogard et al., 1988). The first field experiment in this project was a long-term single-well pumping test in which the activity of the nuclear explosion was treated as a slug-injection point and the explosion products as tracers. The radionuclides detected in the pumped water from the saturated zone of an alluvium aquifer were ³H and ³⁶Cl. The elution curve of ³⁶Cl preceded that of ³H, and the observed phenomenon was ascribed to an anion exclusion process (Thomas and Swoboda, 1970). Anions, being of the same charge as clays and zeolites in the soil, are repelled by these surfaces and are effectively prevented from entering into the intragranular porosity of the matrix.

A pulse of tritiated water, which was discharged accidentally from an isotope processing plant in the Glatt River Valley, northern Switzerland, was traced through a sewage treatment plant and various rivers and groundwater wells (Santschi et al., 1987). Tritium concentrations were used to test predictions for the transport of conservative anthropogenic trace contaminants accidentally discharged into the sewage system. Mass balance calculations indicated that about 2

10 per cent of the tritium pulse infiltrated the groundwater and about 0.5 per cent of the total pulse reached eight major drinking-water wells in the area. In spite of the complex hydrology of the lower Glatt River valley, tritium breakthrough curves could be simulated effectively by a model developed from an experimental well field.

The accident at the nuclear power plant at Chernobyl also provided an opportunity to investigate the infiltration and migration behaviour of radionuclides in the Glatt River valley. The radionuclides ^99mTc,¹⁰³Ru, ¹³¹I, ¹³²Te, ¹³⁴Cs and ¹³⁷CSwere measured several times during May 1986 in the river and an adjacent shallow aquifer (Waber et al., 1987). The main radioactivity (> 75 per cent) of the river water was found to pass through a 0.05 µm filter. Iodine, Ru and Te were found in the groundwater as a result of river water infiltration, being subject to only slight sorption on particulates and other solid surfaces. This was explained to be the consequence of having formed anionic or neutral species. In contrast, Cs was retained completely by river sediments. Particulate (> 0.05 µm fraction) infiltration from the river into the groundwater system was found to be a negligible process.

Determination of groundwater residence time from radionuclide data is possible only if the initial concentration is known and the radionuclide contents do not change as a result of mixing with isotopically different waters. Increases in content and changes resulting from solution/deposition processes normally preclude estimation of residence times. The estimation of the initial content in the groundwater is difficult although, in principle, a plot of radionuclide content as a function of distance along flow lines in a piston-flow system may permit deduction of the initial concentration and thus, interpretation of changes in terms of decay of the radionuclide through time. The closed system requirement in a groundwater system, however, is the most difficult to satisfy.

The measurement of the parent

stable daughter pair, ³H/³He allows the calculation of groundwater age. Tritiogenic ³He is added to the natural ³He content of the groundwater, and if it is assumed that no tritiogenic ³He is lost by diffusion across the groundwater table, the ³H/³He groundwater age is given by:

= t_1/2/ln 2 ln(2 + [³He]/[³H])

(5.4)

where is the age in years and [³He]/[³H] the concentration ratio.This determination is independent of the initial tritium concentration. This is an apparent age of a water parcel if ³He sources other than ³H decay can be excluded or corrected for and mixing of isotopically different waters is negligible. An example of the application of the method is given by Schlosser et al. (1988). For the bomb ³H peak, the deviation of the ³H/³He age from the age determined by identifying the groundwater layer recharged between 1962 and 1965 was about 3 years (15 per cent). The deviation was explained by diffusive ³He loss across the water table and by flow dispersion.

In practice ¹⁴C dating can be applied in similar fashion to that of ³H dating but in a longer time range because of its longer half-life. Applications have included identifying ancient recharge, calculating velocity of groundwater movement by dating the water at different horizons in the aquifer, and calculating contributions of different components to groundwater blends. The age obtained is again an apparent age only and is subject to substantial corrections due to exchange of carbon in solution with carbon of rock minerals and due to solution and precipitation of carbonate compounds during the transit of water through the aquifer. Nevertheless, a groundwater containing no ³H and 30 PMC (per cent modern C) (apparent age 1 x 10⁴ y) can be designated as ancient recharge with some confidence. Furthermore, differences in apparent age at different points in an aquifer system can be used to calculate minimum groundwater flow rate even though the absolute ages may not be known precisely.

If there are no internal sinks or sources for dissolved chloride and ³⁶Cl in an aquifer system, apart from loss of ³⁶Cl to decay, the groundwater may be dated using the equation:

t = -1/

ln((RR_se)/(R_o

R_se))

(5.5)

where R is the measured ³⁶Cl/Cl ratio, R_othe initial ³⁶Cl/Cl ratio, and R_sesecular equilibrium ³⁶Cl/Cl ratio due to hypogene production. The method was comprehensively tested on old groundwaters from the Great Artesian Basin, Australia, against a hydraulic model (Bentley et al., 1986). The ³⁶Cl groundwater ages ranging from less than 1 x 10⁵ years to over 1 million years were obtained in excellent agreement with a hydrodynamic model, but only in areas of the aquifer in which no groundwater mixing occurred.

One of the principal reasons for the common usage of the distribution coefficient (K_d) is the need, in radiological and transport/dispersion models, for a simple term describing the biogeochemical behaviour of radionuclides. The K_dprovides a reasonable first approximation of this behaviour, but its associated limitations (see Section 5.1) must be recognized in any subsequent modelling exercise. More sophisticated solid-solution models (e.g. surface complexation modelling) do exist and have recently been applied to the behaviour of artificial radionuclides under environmental conditions (Turner et al., 1991).

The importance of suspended particle concentration, when assessing solid-solution partitioning, should not be overlooked. The partitioning of three representative radionuclides is used to illustrate this point, based on typical particle concentrations (Table 5.2). Suspended particle concentrations, in lakes are generally lower than those in rivers, so that a concentration of 1 mg l^-1 is commonplace. At this concentration, even a K_das high as 10⁵ would result in less than 10 per cent of a radionuclide being associated with the solid phase.


Environment		Particle concentration	Particulate fraction (%)

		(mg l^-1)	3 x 10²	10⁴	10⁵

Fine-grained bottom sediment		2.5 x 10⁵	99^a	>99.9^a	>99.9^a
Rivers
	extreme range	<l>10⁵	<197	<1>99.9	<9>99.9
	common range	10¹10³	0.323	1090	5099

^aBased on a sediment porosity of 90 per cent and a sediment dry density of 2.5 g cm^-3.

Thus, what seem to be contradictory statements in the literature can be explained in terms of sediment concentration. A high particle affinity is widely attributed to ¹³⁷Cs and, by way of illustration, its extensive use as a sediment tracer has already been discussed in Section 5.2.2.2. In a study of ¹³⁷Cs released to the Susquehanna River during operation of the nuclear power plants at Three Mile Island and Peach Bottom, the nuclide budget was converted to a sediment budget by assuming all the ¹³⁷Cs was in a sorbed state on sediment particles, which was justified by the high affinity of caesium for particles in freshwater (Donoghue et al., 1989). In contrast, following the Chernobyl accident, more than 90 per cent of the atmospherically derived ¹³⁴Cs in various European lake waters was found to be in solution (Santschi et al., 1987), with the relatively low sediment concentrations in these lakes undoubtedly being an important factor.

The solid-solution partitioning of radionuclides is also a function of the environmental history of particles, since this is likely to influence the form of the radionuclide in the solid phase and, consequently, the reversibility of the solid-solution reaction. Again this is well illustrated by reference to Cs. Sorption by micaceous clays (illite), when present in significant amounts, dominates the environmental behaviour of Cs. This association is attributed to the presence of frayed edge sites, corresponding to partially weathered 1 nm interlayers, which are highly selective for Cs⁺ because of its size and charge. With time, a significant fraction of the Cs becomes `irreversibly' bound by the illite, probably due to fixation at interlayer sites through collapse of the frayed edges and migration along the interlayers. For instance, less than 25 per cent of the ¹³⁷Cs is leached by 1 M HNO₃ from stream channel sediments, contaminated by effluents from the Savannah River Plant (Brisbin et al., 1974).

A number of approaches have been adopted to provide insights into the related issues of the nature of solid-solution reactions, the chemical forms of sediment bound radionuclides, and their remobilization characteristics under a range of environmental conditions. Some of the unique advantages of artificial radionuclides, as chemical tracers in the environment, are well demonstrated by the following examples. One approach has been the application of sequential extraction procedures. Förstner and Schoer (1984) employed the technique in a study of various river sediments, subjected to discharges from nuclear installations in Europe and the USA. They compared the leachability of the associated radionuclides with their naturally occurring stable counterparts and with artificial radionuclides added to the sediments in the laboratory. The results indicated that the extent and ease of leachability for each element decreased in the order: laboratory spiked radionuclides > environmental radionuclides > stable isotopes. This suggests an ageing process, perhaps involving incorporation into lattice positions and recrystallization, and emphasizes the need to consider the environmental history of sediments when predicting solid-solution behaviour. A sequential extraction procedure has also been applied to bottom sediments and sediment trap material from Lake Michigan (Alberts et al., 1989). ²³⁸Pu, ^239,240Pu and ²⁴¹Am were associated mainly with a citrate-dithionate extract, corresponding to the nominal hydrous oxide fraction and probably occurring as mineral coatings. In contrast, ¹³⁷Cs was almost totally associated with the most inert fraction of the sediment, attributed to fixation by clay minerals.

The specific role of Fe

Mn coatings has been examined by a different approach, involving a series of novel field experiments (Cerling and Turner, 1982). The study involved placing contaminated gravels in uncontaminated streams, and vice-versa, within the White Oak Creek watershed, Tennessee. Fe

Mn coatings were known to form on stream gravels throughout much of the watershed, part of which is contaminated by several nuclear plants and a low-level radioactive waste disposal site. ⁶⁰CO was rapidly (e.g. days) sorbed from solution, principally by the Fe

Mn coatings.

Under oxidizing conditions, the sorption process was not reversible over a period of months, although some ⁶⁰Co was lost by abrasion of the Fe

Mn coatings. Under mildly reducing conditions, ⁶⁰Co dissolution accompanied reductive dissolution of the Mn oxide fraction of the coatings, while the Fe oxide component continued to form. Thus the behaviour of ⁶⁰Co was dominated by its well-known geochemical association with Mn oxide, probably through co-precipitation. ⁹⁰Sr was held primarily as an exchangeable cation, although some was associated in a non-exchangeable form with newly precipitated Mn oxide. Adsorption, together with desorption of the exchangeable fraction, was rapid (days), whereas only about half the remaining fraction had been lost after 8 months. ¹³⁷Cs was adsorbed irreversibly by illite, the dominant mineral, as discussed previously.

The examples described are only illustrative, so that the patterns of behaviour that emerge for ¹³⁷Cs and ⁹⁰Sr should not be regarded as being universal. Under the variety of conditions that exist in freshwater systems, different solid-solution behaviours must be expected. There is evidence, for instance, that the solid-solution behaviour and sedimentation of ¹³⁷Cs in marl lakes is dominated by scavenging, associated with calcite precipitation (Lindner et al., 1989). Any process leading to a change in the solid phase composition is likely to affect K_d, and surface water systems are particularly prone to such changes.

The integration of ideas on solid-solution partitioning and other important processes into whole-lake studies is rare. This deficiency has most successfully been addressed by Santschi and co-workers through the use of experiments in which limnocorals and whole lakes have been spiked with artificial radionuclides (e.g. Santschi et al., 1986). Such studies have highlighted that radionuclides are removed from lakes not only by hydraulic flushing and scavenging by sedimenting particles but also through direct adsorption to surface sediments. The latter pathway is dominant for less particle-reactive elements, and it is interesting to note that Cs can fall into this category at the low suspended particle concentrations normally present in lakes (see also above discussion). Whole-lake studies, employing Chernobyl ¹³⁷Cs as a tracer, have confirmed the potential importance of direct adsorption as a removal pathway (Santschi et al., 1990).

The chemical speciation of an element in aqueous solution is important in determining its solid-solution partitioning and general biogeochemical behaviour. As in the case of Cs, which occurs predominantly as the hydrated Cs⁺ ion, speciation may vary only slightly in natural waters and consequently not require special consideration. On the other hand, knowledge of speciation, and more importantly how it varies, might be critical to understanding and predicting the behaviour of an element. Pu is a good example and is therefore discussed at some length.

Pu can exist in aqueous solution in any of four oxidation states

III, IV, V, and VI. A number of novel methods have been developed specifically to determine the oxidation state, the most widely used being that based on the separation of `reduced' Pu(III+IV) from `oxidized' Pu(V+Vl) by co-precipitating the former with neodymium fluoride. Dissolved Pu in freshwaters is invariably a mixture of the two forms. Consideration of the available thermodynamic data led to the suggestion that the oxidized form occurred mainly as Pu(VI). More recently, there has been a growing consensus, based in large part on studies involving the selective adsorption of Pu(VI) by silica gel and on the selective co-precipitation of Pu(V) with calcium carbonate (Orlandini et al., 1986), that Pu(V) is the stable `oxidized' form in natural waters, principally as PuO₂⁺. A comparable situation exists in the case of reduced Pu. Pu(IV) is generally regarded as being of greater environmental significance, although thermodynamic data suggest that Pu(III) may be stable under anoxic conditions, particularly at low pH.

We saw in Section 5.2.4.1 that Pu has a high sediment affinity. There are a number of features of the sorption process that are linked to Pu speciation. Firstly, adsorption to natural particles is almost exclusively via the reduced state, and secondly, natural particles appear to take an active role in redox reactions. A series of experiments with Lake Michigan waters (Nelson et al., 1987) showed that when natural particles were present redox equilibrium was attained within two days and that the same oxidation state distribution was achieved whether Pu was added initially in a reduced or oxidized state. In the absence of the natural particles, equilibrium was not achieved after four weeks. With particles present, not only did the oxidation state distributions converge, irrespective of initial oxidation state, but the final equilibrium distribution was the same as that in situ in the lake, associated with fallout Pu. The role of natural particles in the reduction of Pu(V) to Pu(III,IV) has been studied in more detail by Penrose et al. (1987). Rapid reduction occurred only in the presence of natural sediment (from Lake Michigan), with the rate appearing to be nearly first order with respect to Pu(V) and being proportional to sediment concentration. A small number of sites with homogeneous properties seemed to be responsible for the reaction. Direct microbiological mediation is unlikely, since reduction was unaffected by preheating the sediments to 105°C. Ashing the sediments at 500°C, however, decreased the rate of reduction by 98 per cent, suggesting that organic matter may be involved in the reduction process.

The other dominant factor in terms of Pu speciation is dissolved, or more correctly colloidal, organic matter (COC), comprising mainly humic substances derived from decaying organic matter of terrestrial and aquatic origin. Field measurements and laboratory experiments indicate clearly that Pu(III,IV) is complexed by COC. Figure 5.4 shows the K_d of reduced Pu as a function of COC concentration, determined in laboratory experiments with spiked Pu, a natural sediment and a variety of freshwaters. The critical COC concentration in each curve, above which K_d starts to decrease significantly, ranges from about 0.1 mg l^-1 to about 3 mg l^-1 The other, less well-defined effect of COC is that it appears to influence the redox behaviour of Pu, probably by acting directly as a reducing agent. Field and laboratory studies indicate that the concentration of oxidized Pu is high only when colloidal organic matter is low. The effects of Pu(III,IV)

COC complexation, Pu(V,VI) reduction and the contrasting adsorption affinities of the two oxidation forms work in combination to produce a fairly coherent pattern of behaviour (Figure 5.5). Up to about 10 mg l^-1 COC, dissolved concentrations and hence the K_d of total Pu stay more or less constant, since the effects of Pu(III,IV) complexation and Pu(V,VI) reduction seem to cancel each other out. Above 10 mg l^-1 COC, Pu(V, VI) has all but disappeared from solution so that Pu(III,IV) complexation dominates and the total Pu K_d begins to decrease markedly (by up to two orders of magnitude, as has been observed in freshwater systems).

Figure 5.4 Variation of the K_d of Pu(III/IV) as a function of colloidal organic carbon concentration in a series of laboratory experiments, using natural freshwaters, as indicated, and sediment. The natural COC concentrations are indicated by arrows. From Nelson et al., 1987.

Less attention has been paid to other artificial radionuclides with regard to the direct measurement of speciation. Nelson and Orlandini (1986) have shown experimentally that ²⁴¹Am, which invariably occurs as Am(III) in natural waters, is complexed by colloidal organic matter in the same manner as Pu(III, IV). Orlandini et al. (1990) studied Lake Trawsfynydd in North Wales, which receives waste discharges from a Magnox reactor plant, and found that all the dissolved actinides measured (^239,240Pu(IV), ²⁴¹Am, ²⁴⁴Cm and ²³²Th) were associated mainly with the colloidal fraction. Pu(V) made up less than 25 per cent of the dissolved Pu and, in contrast, appeared to be in true solution. A common approach to the question of speciation in environmental settings is to rely on thermodynamic predictions, e.g. with ¹³¹I and ^103,106Ru.Iodine may occur in the

I, 0 and V oxidation states and Ru as Ru(III) or Ru(VI). Both elements can occur as oxyanions (IO₃^- and RuO₄^2-) and I as iodide. Their frequently observed mobility in surface and groundwater systems is generally attributed to the low adsorbance of anionic species, since natural particles themselves generally have a negative surface charge, especially in the presence of organic coatings.

Figure 5.5 Concentrations of dissolved forms of plutonium as a function of colloidal organic carbon concentration in a laboratory experiment, using data from ANL Pond experiment shown in Figure 5.4. Pu concentrations are normalized to a constant plutonium concentration in the solid phase of 1 pCi g^-1. From Nelson et al., 1987.

5.2.4.3 Redox boundaries as critical features in the biogeochemical behaviour of artificial radionuclides

As we have seen, redox reactions in part determine the speciation and hence the mobility and fate of many artificial radionuclides in freshwater systems. The most important, but by no means only, driving force for such reactions is the microbiologically mediated, oxidative breakdown of organic matter by various oxidizing agents. The radionuclides may themselves undergo redox transformations or may be affected indirectly (e.g. by being associated with redox sensitive phases, such as Fe and Mn oxyhydroxides). The redox conditions in freshwater systems depend largely on the rates of reaction and the mixing processes affecting the availability of reactants. Reducing conditions are more likely to occur in soil waters and sediment porewaters, than in surface waters, due to high net respiration rates and restricted oxygen supplies. In river waters, reducing conditions are generally restricted to badly polluted stretches of river, in which the net rate of respiration exceeds reaeration. In lakes, reducing conditions exist mainly in bottom waters isolated by stratification, and in groundwaters they may be linked to high rates of respiration or the long hydraulic residence times. It is not uncommon for the redox potential in groundwaters to decrease with distance along the flow path, in other words with the length of time since the water was last in contact with the atmosphere.

The best-studied situation is the redox boundary that generally occurs in the vicinity of the sediment

water interface in lakes and reservoirs, principally because it is relatively stable and easy to observe with minimal disturbance to the ambient redox conditions. The results of such studies, involving artificial radionuclides, not only can be applied to other environmental settings but can be used as models for understanding better the behaviour of other elements. One of the best examples is ¹³⁷Cs. Although Cs itself does not undergo any redox transformations, elevated concentrations of¹³⁷Cs have been observed in the isolated bottom waters of Par Pond, a seasonally anoxic reservoir on the site of the Savannah River Plant in South Carolina. Initially, it was tempting to link the mobilization behaviour of ¹³⁷Cs to the well-known reductive dissolution of Fe oxyhydroxides. However, more detailed studies indicated that mobilization was due to ion-exchange displacement of Cs⁺ from sediments (principally clay minerals) by a range of cations, produced as a result of anaerobic diagenesis (Evans et al., 1983). Combined field and laboratory experiments identified three types of ¹³⁷Cs binding sites: (1) surface sites on clays and other minerals, where binding is relatively non-selective being related to the charge density of the ions; (2) sites at the edges of clay interlayers of 1 nm spacing (frayed edges of partially weathered micas), where Cs⁺ is displaced only by cations of similar charge and size; (3) true interlayer sites from which Cs⁺ is not readily exchanged. NH₄+ appears to be particularly efficient at displacing Cs⁺ from the frayed edge sites, and this ties in with the fact that its concentration varies seasonally and with depth in the sediment pore waters and bottom waters of productive lakes. The fraction of ¹³⁷Cs present in each of the three forms identified by Evans and co-workers will vary, depending on the mineral composition of the sediments and to a lesser extent the time since release of¹³⁷Cs, due to the slow fixation in interlayer sites. The former explanation, or more specifically the higher ratio of kaolinite to weathered micaceous clays, has been used to account for the greater leachability of ¹³⁷Cs from Par Pond sediments than from lake and river sediments from Oak Ridge, Tennessee. That is not to say, however, that ¹³⁷Cs significant remobilization, associated with seasonal anoxia, occurs only in situations where micaceous clays are unimportant. One year after the Chernobyl accident, the same process represented the major source of ¹³⁷Cs in the bottom waters of Esthwaite Water, a lake in Cumbria (UK), the sediments of which have a clay mineral composition dominated by illite and chlorite (Davison et al., 1992).

The effect of the same redox boundary on the biogeochemical behaviour of Pu is more controversial, there being disagreement as to whether significant release from bottom sediments actually occurs under reducing conditions (e.g. Sholkovitz et al., 1982; Alberts et al., 1986). This situation highlights the fact that determining the remobilization behaviour of elements under such conditions is far from being straightforward, and that the problem is best tackled by several complementary approaches. More specifically, it is inadequate to examine water column profiles alone, since this approach is open to misinterpretation.

The study of lake sediments as a means of providing information on the sedimentary processes operating throughout a lake basin is considered in Section 5.2.2.2. Of more general interest is the way in which fine-grained sediments in the deeper waters of a lake may represent a more or less continuous historical record of the processes operating in a lake basin, its catchment and beyond. Thus, lake sediments have become the principal natural medium for providing historical records of environmental change. The ideal requirements are that: (a) the environmental change is reflected in some way in the nature or composition of the sedimentary particles; and (b) the record is not altered by the various diagenetic processes typically operating in a sediment, such as remobilization or degradation due to biogeochemical transformations, compaction, physical disturbance due to resuspension and reworking by bottom currents or benthic organisms. Within limits, the historical records can be corrected for the effects of diagenesis by mathematically modelling the constituent processes. Such diagenetic models have become an end in themselves and are capable of providing quantitative information on the various processes operating in a sediment (Berner, 1980).

Artificial radionuclides may act either as the `pollutant', for which the discharge or emission record needs to be reconstructed or, more commonly, as the `tracer' by which the diagenetic model is calibrated. The latter approach applies particularly to studies of the vertical distribution of ¹³⁷Cs in sediments, with the source term, due to fallout, being well defined. One of the earliest and most frequently quoted studies of this type is that of Robbins and Edgington (1975), who looked at sedimentation rates in Lake Michigan by means of ²¹⁰Pband ¹³⁷Cs measurements. Figures 5.6a-c show the data and modelling results for the core with greatest surficial mixing. Firstly, a best fit to the ²¹⁰Pb profile was obtained for a model incorporating sedimentation and compaction only (Figure 5.6a). This provided a timescale for converting the known historical record of ¹³⁷Cs atmospheric fallout in Lake Michigan to a depth scale with respect to the sediment surface (Figure 5.6b). The ²¹⁰Pb results for all eight sites examined indicate that the sedimentation rates have remained unchanged at each of the sites for at least 100 years and perhaps for as long as 7 x 10³ years. Assuming this to be true, there should be a straightforward correspondence between the profile in Figure 5.6b and that of ¹³⁷Cs activity concentration in the sediment (Figure 5.6c), which is clearly not the case. The first appearance and the 1963

64 maximum of ¹³⁷Cs fallout are deeper and the activity concentrations in recent years are greater than predicted. This is accounted for by additionally incorporating a sediment mixing term into the model. Best-fit values for the sedimentation rate at the sediment surface (R_o in cm y^-1)and the mixing depth (S in cm), together with the calculated profile, are shown in Figure 5.6c. The model, with the mixing term, was then applied to the ²¹⁰Pb profile, resulting in best-fit values of R_o = 0.28 cm y^-1 and S = 4.0 cm. Thus fitting the model to the two independent profiles yielded the same values for sedimentation rate and mixing depth, providing supporting evidence for the validity of the model. Edgington and Robbins (1975) have also successfully applied this approach to ^239,240Pu profiles in Lake Michigan sediments.

Figure 5.6 Vertical profiles in a sediment core from southern Lake Michigan. (a) Measured values and calculated profiles of ²¹⁰Pb. (b) The annual flux of ¹³⁷Cs to the lake surface from bomb fallout, plotted using a timescale established from the ²¹⁰Pb measurements. (c) Measured ¹³⁷Cs activities and the calculated profile, assuming rapid surface sediment mixing. (Reprinted with permission from Robbins and Edgington 1975; copyright (1975) Pergamon Press.)

The above approach yields the mixing depth and indicates implicitly that mixing is rapid in comparison with the sedimentation rate. When mixing is insufficient to produce a homogeneous mixed layer, the distribution of activity becomes sensitive to the details of the mixing process. Thus concepts, such as `conveyor belt' mixing or `biotransport', have been developed in addition to the more common approach of random mixing (see Berner, 1980).

Biogeochemical reactions may also be an important influence on radionuclide profiles, especially where concentration gradients are created in the sediment porewaters. Under such conditions, diffusional transport occurs even in the absence of mixing, i.e. by molecular diffusion. Lerman and Lietzke (1975) were amongst the first to consider the role of porewater diffusion on sediment profiles of artificial radionuclides. They developed a model to describe the vertical profiles of ¹³⁷Cs and ⁹⁰Sr in the sediments of Lakes Erie and Ontario. The cores were collected in 1969

70, so that the 1963

64 maxima of ¹³⁷Cs and ⁹⁰Sr were preserved at or close to the sediment surface. The model, which incorporated adsorption, diffusion in the porewaters and sedimentation, produced a good fit to the data, suggesting indirectly that sediment mixing was not important at the coring sites. The need for a porewater diffusion term was manifest most clearly in the much greater penetration of ⁹⁰Sr in the sediment cores, compared to ¹³⁷Cs, attributable principally to the two order of magnitude lower K_d for ⁹⁰Sr.

Later studies have extended the above concepts and have included the use of other radionuclides e.g. ¹⁰⁶Ru (Sickle et al., 1983). In addition, an increasing number of studies have focused attention on major inconsistencies between sediment profiles of ¹³⁷Cs and local fallout histories for a large number of disparate and geographically dispersed lakes. This has raised questions as to the validity of ¹³⁷Cs dating, at least under certain conditions, and emphasizes the need to adopt more than one dating technique when a reliable chronology is required. Such behaviour has been observed in lakes in North America, Europe and Australia. Explanations for the inconsistencies include mixing effects, changes in sedimentation rate, delayed but significant inputs from the catchment, recycling in the water column, sediment focusing, and redistribution due to porewater diffusion. The latter appears to be important at low pH conditions and where there is a lack of micaceous clay minerals. Under such conditions, ¹³⁷Cs may be associated mainly with organic matter and consequently be more susceptible to biogeochemical recycling.

The case study focuses on the behaviour of ¹³⁷Cs and ⁹⁰Sr and a feature of the two radionuclides is the greater geochemical mobility of ⁹⁰Sr, as described in Sections 5.2.4 and 5.2.5. Pre-Chernobyl measurements in the large rivers of Finland show an average removal by runoff of 41 per cent for ⁹⁰Sr and 7 per cent for ¹³⁷Cs of the total amount deposited in the catchment areas (Salo, 1983). As the water surface is about 9 per cent of the Finnish area, these results indicate that ⁹⁰Sr is significantly removed from land areas by runoff, while ¹³⁷Cs is not. Measurements of meltwater runoff during spring 1989 in a Norwegian mountain area (Haugen et al., 1991), combined with estimates of the total amount of snow, indicate that runoff of mainly Chernobyl-derived radiocaesium from the drainage area was 0.01

0.1 per cent.

Large variations have been observed in the relative activity concentrations of ¹³⁷Cs in solution and associated with suspended sediment (Carlsson, 1976; Haugen et al., 1991). These probably reflect real differences in conditions through the seasons. In particular, sediment bound activity is relatively greater when suspended sediment concentrations are high, due to spring meltwaters (see Section 5.2.4.1). Finnish measurements showed that lake sedimentation resulted in a sharp decline in the surface waters concentrations of Chernobyl-derived caesium by the autumn of 1986. There was a further decline in 1987, and by the end of the year concentrations in the two drainage areas, where Chernobyl fallout was highest, were only 3

15 per cent of the maximum values in May 1986. A slightly more modest decrease (down to 10

25 per cent) was measured in the larger rivers. In Sweden, Hammar et al. (1989) observed a similar rapid decrease in concentrations in surface waters, together with a more than two-fold increase in bottom sediment concentrations during the period October 1986 to September 1988. The amount retained in lake sediments varied, e.g. with clay mineral content, but was invariably greater for ¹³⁷Cs than ⁹⁰Sr. In addition to the effects associated with spring meltwaters, there appear to be many other complicating factors, related to winter conditions at high latitudes, that are important in terms of the atmospheric deposition

runoff characteristics of radioactivity. Although poorly understood, some insight into these factors was gained from a study using weapons fallout as a tracer (Lund et al., 1962).

From a population exposure point of view, radioactivity ⁱⁿsurface waters and bottom sediments is unlikely to be important. Direct exposure of bottom sediment may occur, especially during dry spells. The critical areas will be the shores of lakes and rivers. Use of these areas can lead to radiation exposure directly from the radioactive materials on the ground and via inhalation of resuspended sediment. The activity content of the sediment may vary strongly, even quite locally. Variation is particularly marked in rivers, where high concentrations are often found on the inner side of river bends. Such locations may be used for swimming, sunbathing and camping. However, both the size of the potentially exposed population and the exposure time over a year will be quite limited, at least in a North European climate. Boating and fishing activities are also unlikely to pose any significant risk. Pre- and post-Chernobyl studies of ¹³⁷Cs have shown that irrigation is an insignificant exposure pathway unless it takes place in an area otherwise unaffected by contamination.

Post-Chernobyl measurements in drinking water were performed in Finland at a large number of waterworks from 2 May 1986. The dominating radionuclide was ¹³¹I until about the beginning of June. Later on the main contaminants were ¹³⁴Cs and ¹³⁷Cs, but smaller amounts of several other radionuclides were also found. The concentrations of ¹³¹I were up to around 30 Bq kg^-1,but rapidly decreased to about 1 Bq kg^-1 at the end of May. ¹³⁴Cs and ¹³⁷Cs were rarely above 1 Bq kg^-1 and decreased toward autumn. The 1987 values were lower than the 1986 values, and the decrease continued through the summer. The effect of water purification plants upon strontium and caesium concentrations in both pre- and post-Chernobyl studies was found to be of moderate importance. Iodine and strontium concentrations are hardly affected, and the reduction of caesium concentration is at most 50 per cent.

It is easy to show by simplified, pessimistic calculations that the concentrations in water resulting from deposition upon a water surface will not even in the worst circumstances lead to doses via drinking water which could rival those via other exposure pathways, unless the drinking water is used in areas otherwise not affected by contamination.

The total amount of weapons fallout ¹³⁷Cs contained in fish in a typical lake is reported to be as low as 1 per cent of the total amount present in the lake water, although the concentration in fish is much higher (Carlsson, 1976). Typical concentration factors for ¹³⁷Cs, in fish relative to water are 10³

10⁴. The usefulness of concentration factors of this type, however, has been questioned after Chernobyl, since uptake does not take place directly from water but indirectly via numerous nutrition pathways. A direct correlation between concentration in fish and the average fallout in the drainage area (in Bq m^-2) seems to be just as relevant and useful. For trout in the mountain lakes of Norway this 'transfer factor' [(Bq kg^-1 fish)/(Bq m^-2 deposited ¹³⁷Cs)] )] ranges from 0.01 to 0.2 (Tveten, 1991).

Post-Chernobyl measurements both in Norway (Ugedal and Blakar, 1991) and Sweden (Andersson et al., 1990) indicate that the radiocaesium concentration in fish is closely linked to the content in sediment of the same lake. Predictive models perform quite well, in which concentration in fish is expressed as a function of the amount of radiocaesium in sediment (Bq m^-2 ) and various sensitivity parameters are used to characterize the lake. The sensitivity parameters may be height above sea level, area of the lake, residence time of water in the lake, and the potassium concentration in lake water. A certain division into type classes of lakes is necessary before it is possible to determine valid predictive models, however. In particular, mountain lakes seem to be in a class by themselves.

Both from earlier measurements on weapons fallout and from post-Chernobyl studies, it was found that strontium in an exposure context is less important than caesium, as strontium concentrates in bone. Most of the freshwater fish consumed are of species where most of the bones are easily removed, and it can be assumed that less than 10 per cent of the fish bone will be consumed, making the contribution from strontium negligible in most cases. The concentration of ¹³⁷Cs in fish depends, among other factors, upon the feeding habits of the fish, which are different for different species, but which also change with age and size of fish of the same species. Saxen and Aaltonen (1987) found post-Chernobyl that when fish were divided into three classes (benthic, predators and intermediate), the activity rose first in benthic and intermediate classes. In July to September 1986, activity in fish rose sharply, but was in autumn still lowest in predators. By 1987 the activity was highest in predators and was still rising, while for non-predatory fish it had started to decrease. Similar results were found in Sweden.

It was found in Finland in 1986 that for the same species (perch in particular), the smaller fish had the higher concentrations (Saxen and Rantavaara, 1987), and this was explained by differences in diet for different size and age of fish. The youngest perch eat mainly plankton. Later on they turn to insects, worms etc., and even later they start eating small fish. This inverse correlation between concentration and size did, however, not continue to be valid in 1987, when it was found both for perch and pike that the concentration was higher in larger fish. The difference between the years is explained by the fact that in 1986 the caesium had not had time to propagate through the food chain to the larger fish. Swedish investigations (Andersson et al., 1990) in 1987 also found that the concentration was larger in larger fish.

Another correlation, observed in Finland (Saxen, 1990) and in Norway (Ugedal and Blakar, 1991) is that in the same drainage area the concentration in fish is higher in smaller lakes. The concentration is also higher in lakes where the residence time of the water is longer, when the lakes are otherwise similar (Andersson et al., 1990). These authors found this factor to be the most important next to fallout levels. They also found that concentrations in fish are lower in hard water or water with high concentrations of phosphorus or potassium. Experiments on dumping large quantities of potassium in lakes to decrease caesium uptake to fish have been performed in Norway (a slight decrease could perhaps be seen) and Sweden (no decrease found).

Experiments to determine the `biological caesium half-life' have been performed in Norway. It was found that excretion of caesium is very dependent upon water temperature, and one basin-type experiment (Tveten, 1990b; Christensen, 1989), performed over a whole winter season, showed no decrease at all. Long-term measurement of caesium half-life has been performed in one mountain lake in Norway continuously since Chernobyl (Brittain, 1991), showing that the half-life decreased from about 8 to about 2 years from 1986/87 to 1988/89. For the season 1989/90, however, it had increased once more to more than 9 years. The half-life measured in this experiment is a `half-life' of the whole environment, as reflected in fish concentrations, since new activity from the drainage area will enter the lake, particularly during the spring melt. In the basin-type experiment, the fish had been transferred from its original contaminated environment to an environment free of radioactive contamination.

Typical freshwater fish consumption in Finland is reported to be about 4 kg y^-1. It is concluded (Saxen and Rantavaara, 1987) that the average intake of radiocaesium in 1986 via fish was of the same order of magnitude as via beef. In 1987, the relative contribution from fish was larger (about 65 per cent of the total average intake via foodstuffs) than in 1986; mainly because the activity levels in agricultural products decreased more rapidly than in freshwater fish. The radiocaesium concentration in cultivated rainbow trout in Finland was quite low (average of about 40 Bq kg^-1 in the most affected areas). Norwegian cultivated trout and salmon had no increase in radiocaesium content after Chernobyl, as they are cultivated in ocean waters.

The Scandinavian case study provides a representative picture of the state of knowledge concerning radionuclide behaviour in the freshwater environment. The main radionuclide and exposure pathways are reasonably well defined, especially for common nuclides such as ¹³⁷Cs and ⁹⁰Sr, but a detailed knowledge is frequently lacking. Therefore the effects of short- and long-term environmental perturbations on the fate of radionuclides in the future are often difficult to predict.

5.3 ESTUARIES AND INTERTIDAL ENVIRONMENTS

Radionuclides enter estuaries as both solution and particulate solid phases, together with the water and sediment, from any or all of four source environments: sea, river, atmosphere and the estuary itself. The magnitude and radionuclide composition of these inputs varies widely between estuaries (Table 5.3), as well as varying with time in any one. For many estuaries the dominant sources of artificial radionuclides is fallout, introduced by rivers from the land and by currents from the sea, as well as by direct deposition on the estuary. Where nuclear industry discharges occur, they can be the dominant source, as river, estuarine or marine inputs, depending on the location of the discharge point.

Within an estuary, the physical and chemical pathways followed by the radionuclides in solution and particulate phases are determined by three main groups of processes. Firstly, transport in the water column of both phases by the estuarine circulation results in dispersion, dilution, mixing and, for the particulate phase, fractionation of the radionuclides, as well as enabling chemical interactions between the two phases. Secondly, deposition of the particulate phase in the estuarine sediment deposits results in the long-term accumulation of radionuclides in the estuary. In these deposits further chemical interactions between phases are possible. Thirdly, uptake by biota of radionuclides can occur. The last category is not considered in this section.

In essence, an estuary is a narrow coastal inlet in which the sea is in contact with a river, resulting in a variation in salinity between seawater and freshwater. The seaward boundary is nominally set by the opening out of the coastline, although the geochemical effects associated with the plume of estuarine water extend the influence of the estuary into the shelf seas. The landward boundary can be put at either the limit of penetration of salt or at the limit of tidal changes in water level. The distinction is important, since there can be a considerable difference in the relative position of the two limits, e.g. in the Amazon, tidal rise and fall of water levels can be detected 850 km upstream whereas salt penetration is limited to the area of the plume outside the mouth. An estuary, therefore, can include three zones which have been given a variety of names: freshwater, also limnetic or riverine; brackish or mixohaline; and seawater or euhaline zones.

In addition to morphology, the other major determinant of the estuarine environment is the water circulation system. This has two basic components, a tidal and a non-tidal or residual circulation. The tidal circulation consists of reversing, landward and seaward tidal currents generated by the tidal change in sea level beyond the estuary mouth. These reversing currents are accompanied by changes in water level in the estuary itself (high and low tide). This results in a cyclical change in the volume of water stored in the estuary, the tidal prism, and a cyclical inundation of the marginal intertidal parts of the estuary.

The non-tidal circulation, in turn, includes two components: an outflowing current due to the river discharge and a saline density current generated by the density contrast between freshwater and saltwater. Together, these contribute to a circulation in a vertical plane, with a seaward current at the surface and a headward current at the bed. A horizontal non-tidal circulation can exist also, especially in wide estuaries, generated by the Coriolis force due to the Earth's rotation, i.e. anticlockwise in the northern hemisphere. These non-tidal circulations can be affected and even reversed by wind stress. For further discussion of estuarine circulation and hydrodynamics see, for example, Bowden (1980).

The relative importance of the three circulation components varies between estuaries, principally due to the variation of tidal amplitude around the globe, the size of the river discharge and the estuary morphology. Several different types of estuarine circulation can occur, defined by the longitudinal distribution of salinity in the estuary averaged over a tidal cycle, as shown by the isohalines for the three types in Figure 5.7.

Figure 5.7 Estuarine circulation types based on salinity distribution (contours in ppt) and showing mean flow vectors over a tidal cycle (arrows) and suspended sediment distribution (dots). From Postma, 1980; reproduced by permission of John Wiley.

It is important to remember that the division of estuaries between these circulation types is not rigid, and a single estuary may exhibit variations in its circulation regime, both spatially and temporally, e.g. a well-mixed estuary may be stratified in its upper reaches and this zone may become more extensive on the early ebb tide.

The tidal and non-tidal components of the water circulation system transport the radionuclides within the estuary by advection and diffusion. This results in changes in the distribution in space and time of radionuclide activity concentrations in the water column of both the solution and sediment phases.

The dynamic behaviour can be considered over various timescales, i.e. on timescales appropriate to either the tidal cycle or to the non-tidal circulation (tidally averaged), or longer.

Advective transport is produced by the depth mean flow of water and the fluxes (or discharges) of radionuclide solution phases are proportional to the water discharge (mean velocity times cross-sectional area). In partially and well-mixed estuaries, advection by the tidal circulation produces a landward solution phase transport on the flood tide and seaward on the ebb and, consequently, it is an important mechanism for the input of radionuclides which have a marine source. Velocities and depth and, hence, water discharges and radionuclide fluxes show cyclical variations in magnitude, as well as a reversal in direction, falling to zero around high and low water. Peak velocities are related to the tidal amplitude and estuary morphology and, in macro-tidal areas, are often up to 3 m s^-1and occasionally reach 5 m s^-1. Figure 5.8a shows the variations in ^239,240Pu solution phase flux and associated variables over a tidal cycle at a mid-channel station in the shallow, macro-tidal Esk Estuary, UK (also called Ravenglass Estuary in the scientific literature). In this case, although the estuary is small, the fluxes are high because of the strong tidal currents and its proximity to the Sellafield reprocessing plant which discharges directly to the sea.

Averaging the velocities with time over a tidal cycle removes the tidal component and reveals the non-tidal advective element of the circulation. This is responsible for the non-tidal discharge of water to the sea which is the combined result of the river flow and the compensation for the density-driven non-tidal discharge from the sea. Partially mixed estuaries have the most vigorous two-layer circulations, with a landward flow towards the bed and seawards towards the surface with velocities of 0.01

0.1 m s^-1. Beyond the landward limit of this circulation, at the head of the estuary, i.e. the null point, the non-tidal flow is seawards at all depths. Salt penetrates beyond this point, however, by the tidal dispersion processes described below. In contrast, in salt wedge estuaries, the two-layer flow consists of a relatively fast seaward surface flow and a virtually stationary lower saline layer. In well-mixed estuaries, whilst the non-tidal flow is theoretically seawards with low velocities at all depths, there is usually a weak landward flow at the bed.

The non-tidal circulation, in general, is important in influencing the net output of radionuclides to the sea from the estuary, as well as in providing another transport mechanism for the input of marine radionuclides.

Dispersion results in both transport and dilution of the solution and particulate phases in an estuary. In addition to dispersion due to turbulent eddy diffusion and to velocity shear, dispersion in estuaries is produced by the lateral trapping effect, by which there is a delayed release of saline water draining from marginal intertidal areas back to the main channel during the ebb tide. The magnitude of the dispersion coefficients varies between estuaries and, with time and position, within individual estuaries. It normally has to be determined empirically, from the salinity distribution or from tracer studies. Typically, the longitudinal dispersion coefficient D_x varies from 10 to 100 m² s^-1, increasing seawards and with river discharge.

The dispersion processes mix the water masses and their radionuclide solution phases present in the estuary, producing spatial gradients in salinity and radionuclide concentration, modified by biogeochemical processes for non-conservative radionuclides. This results in the dilution of the radionuclide inputs and their diffusive transport in the direction of the gradient. For marine or river inputs, a longitudinal gradient in radionuclide concentration in the brackish zone is produced, with activity concentrations decreasing away from the end of the estuary which is the dominant source (Figure 5.8c). The geometry of this distribution depends on the intensity of the dispersion. For radioactive discharges direct to an estuary, instantaneous and tidally averaged concentration gradients are set up in the three axial directions away from the discharge point.

Over a tidal cycle, the combined effect of advection and dispersion in a tidal estuary results in a salinity and solution phase distribution which migrates up and down the estuary during the cycle. In addition, non-tidal excursions can occur over longer timescales, because of river discharge variations and the cycles in tidal amplitude. The latter includes the semi-monthly spring

neap cycle, which gives bigger tidal ranges and more intense currents on spring than neap tides, and the semi-annual spring tide cycle. These effects can also be defined by considering the salinity and radionuclide distributions at the same point in time of a tidal cycle, e.g. high tide. Changes in river discharge can produce major changes in distribution which are often seasonally related. In climates with marked seasonal changes in rainfall large non-tidal excursions nearly equal to the length of the estuary can occur regularly, as in monsoon climates.

A net flux of solution phase radionuclides to the sea may occur over tidally averaged periods (or longer), depending on the dispersive and non-tidal advective fluxes. Where the sea is the dominant radionuclide source, i.e. in the simplest case of a one-dimensional steady-state distribution in a well-mixed estuary with concentrations decreasing towards the head, the seaward advective flux is balanced by the headward diffusive flux and the net flux is zero. For dominant fluvial or estuarine radionuclide sources, i.e. with seaward decreasing concentrations, a net seaward radionuclide flux occurs which is the sum of the advective and diffusive fluxes. For example, the relatively large Severn Estuary, UK, receives inputs of ¹³⁷Cs predominantly from reactor discharges to the estuary, with subsidiary inputs from reprocessing waste discharges to the sea and from fallout on the river catchment. This results in a stable averaged longitudinal concentration gradient which decreases seawards. The large tidal amplitude results in a significant diffusive transport of ¹³⁷Cs out of the estuary, which exceeds the advective non-tidal transport by factors of 3

6 (Uncles, 1979).

The flushing time of an estuary is an important property affecting the fate of radionuclide inputs which vary with time. Several definitions exist for the flushing time, e.g. the ratio of the volume of the freshwater in the estuary (calculated from the salinity distribution) to the freshwater discharge into the estuary. This is the same as the residence time for conservative radionuclides in solution. Flushing times for partially and well-mixed estuaries vary with their topography, river and tidal water inputs and are generally in the range of days to a year. For highly stratified salt wedge estuaries and fjords, separate flushing times need to be given for the upper and lower layers, with the latter likely to have extremely long times.

In conclusion, the circulation of partially and well-mixed estuaries causes inputs of radionuclides in solution from fluvial, marine or estuarine sources to be dispersed throughout the estuary, up to the limit of the penetration of salt (for the last two sources). For estuarine or fluvial inputs, there will be a net flux to the sea over periods longer than the tidal cycle. For a dominant marine input, there is no net flux when the input is constant, a net flux to the estuary with increasing input and a net flux to the sea with decreasing inputs. The radionuclide inputs will be diluted in the estuary by the mixing processes associated with dispersion.

Figure 5.8 Radionuclide distributions in the water column, Esk Estuary, UK. (a) and (b) Variation with time over a tidal cycle of ^239,240Pu fluxes and associated variables (Pu fluxes are per metre width in mid-channel). Based on Kelly et al., 1991. (c) Longitudinal variation at high tide of salinity and ¹³⁷Cs and estuary zonation. From Assinder et al., 1985; reproduced by permission of Elsevier Science Publishers Ltd.

The absolute and relative magnitudes of the different sediment inputs vary widely between estuaries and, with time, in any one estuary. At one extreme are delta channel estuaries in which the fluvial source is overwhelmingly dominant and large, e.g. Mississippi, 3.3 Gt y^-1.At the other extreme, with a dominant marine source, will be estuaries with high tidal ranges and low river discharges. Partially mixed estuaries typically have important inputs from both fluvial and marine sources, as well as estuarine inputs, e.g. Chesapeake Bay (Schubel and Carter, 1977): river 1.1 Mt y^-1, estuarine 0.6 Mt y^-1, marine: 0.2 Mt y^-1; and the Seine Estuary, France (Avoine, 1986): river 1.5 Mm³ y^-1 (mud), marine 6 Mm y^-1 (half mud and half sand). Fluvial inputs to a particular estuary will vary with river discharge, which may have a pronounced seasonal control. River sediment discharge may be dominated by flood events with a low frequency of occurrence, e.g. Chesapeake Bay rivers received 50 years of average sediment discharge in one storm, leading to extensive sediment deposition (Schubel, 1974). However, whilst such discharges might be mainly contained in a large estuary they would be flushed through a small one. Marine inputs will vary with the tidal regime. Again, low-frequency storm events may increase marine inputs significantly, both through increased wave activity and sediment suspension and from increases in tidal amplitude due to wind stress and low pressures.

The advective transport of the sediment and radionuclide particulate phase is complicated because, as in other aquatic environments, the mobility of the sediment is affected by a number of factors, both flow and non-flow-related (see Section 5.1). Since the concentration of the mobile sediment, mainly suspended load, and its grain size distribution are both a function of flow and because a flow related threshold for transport exists (erosion threshold), the particulate radionuclide fluxes and their characteristics vary much more widely with flow than they do for the solution phase. In tidal estuaries, the cyclical change in depth mean velocity over a single tidal cycle results in the advective transport of sediment, landward on the flood and seaward on the ebb, accompanied by deposition during decreasing velocity periods and erosion with increasing velocity. However, because of lags between the sediment response and the velocity changes (settling and scour (erosion) lag), concentration does not fall to zero in the low-velocity periods (Figure 5.8b). The sediment behaviour will define the behaviour of the particulate phase radionuclides. In general, outside of the turbidity maximum (see below), depth mean suspended sediment concentrations in estuaries range from <1 to 10³ g m³, depending on the velocity regime and the availability of sediment. The low sediment concentrations in the Esk (Figure 5.8b) are due to supply limitations, as the measuring station was in a gravel bed reach. A detailed discussion of estuarine sediment transport is given by Dyer (1986).

Just as for radionuclides in solution, an important aspect of the behaviour of the particulate phase radionuclides in estuaries is the mixing by dispersion processes of the different inputs, leading to their dilution and diffusive transport. However, for the particulate phase, significant mixing is not limited to brackish water but can occur throughout the estuary as a result of dispersion due to the effects of settling and scour lag, i.e. sediment deposited from the flood tide will tend to be eroded by a different body of water on the ebb, joining another suspended sediment population. Also, mixing will occur with older contaminated or uncontaminated sediment which is being eroded at the estuary margin or bed. Older contaminated sediments can be an important secondary source of radionuclides in shallow macrotidal estuaries.

Estuarine waste discharge or atmospheric particulate inputs will also be mixed with the other inputs by the dispersion processes. Depending on their relative magnitudes, inputs from any of these estuarine sources will either enhance or dilute the activity of the suspended sediment population.

For tidally averaged conditions, sediment will be transported by the non-tidal advection and dispersion transport systems. Net dispersion transport will be determined by the gradients in sediment and/or radionuclide particulate phase concentrations. This is likely to be seawards in the case of rivers with high suspended sediment discharges and landwards with small rivers and/or shelf systems with strong tidal and wave currents and relatively high suspended sediment concentrations. There will also be an advective sediment flux to the sea, due to the net outflow of water with the non-tidal circulation. However, this may be more than offset by net inputs of sediment by the density component of the non-tidal circulation, which occurs because of the higher suspended sediment concentration towards the bed, and/or by tidal asymmetry in shallow estuaries. In the latter case, a shortening of the period of the flood tide relative to the ebb can result in higher velocities on the flood and higher sediment discharges. In the Esk Estuary, UK, this resulted in a net sediment input of 18 t/tide (Kelly et al., 1991). In any one estuary, the relationship between sediment inputs and outputs will change with time depending on the variations in sediment discharge of the marine currents and river.

In partially and well-mixed estuaries, the two mechanisms that can cause a net landward transport of sediment/particulates, i.e. the non-tidal circulation and the tidal asymmetry, lead also to the development of a zone of enhanced suspended sediment concentration in the upper part of the estuary, i.e. the turbidity maximum. There, suspended sediment concentrations can be as high as 10³10⁴ g m^-3 in macrotidal estuaries, but are generally lower in mesotidal estuaries (10² g m^-3). This zone extends beyond the landward limit (null point) of the non-tidal circulation, because of dispersive mixing. The sediments of this zone are muds, due to the decrease in velocities towards the head of the estuary. Consequently, this should be a zone of high radionuclide activity concentrations 'for the suspended particulate fraction. Such turbidity maxima have been described, from a sedimentological point of view, from many estuaries in Europe and North America, e.g. Gironde, France (Elbaz-Poulichet et al., 1982). The position of the turbidity maximum migrates with the tidal and non-tidal excursions. The passage of a turbid zone downstream on the ebb tide, unrelated to high velocities, can be seen in Figure 5.8b.

Sediments in estuaries are essentially non-conservative in behaviour and a fraction of the sediment and particulate radionuclide inputs will be deposited in the estuarine sediment. Net sediment deposition (over tidally averaged periods) is due to the existence of longitudinal and lateral gradients in mean velocity in the estuary, decreasing towards its head and sides.

Combined with the settling lag effect, the gradients lead to sediment of a particular grain size being transported landwards beyond the mean velocity required to actively resuspend it. In addition, the gradients in carrying capacity, reinforced in tidal estuaries by lower velocities on the ebb than flood, mean that less sediment of a particular size can be transported.

The net effects of advective and dispersive processes on recent particulate phase radionuclide inputs are best seen in the spatial distribution of activity concentration in the surface layer of the bottom sediments. For example, the longitudinal distribution of ^239,240Pu in surface sediments of the Western Scheldt Estuary, Netherlands (Figure 5.9a) (Duursma et al., 1985) is the result of the effects of the circulation system on three different inputs: a dominant marine input of Pu from reprocessing discharges in France and UK, a local estuarine source of Pu from nuclear wastes and a fluvial input of fallout Pu. The activity concentrations have been normalized in order to minimize the effects of grain size variation, by standardizing them against the Al content. In another example, from the Ulhas River Estuary (Trombay), India (Figure 5.9b), Patel et al. (1975) show the net effects of the circulation system on the distribution of ¹³⁷Cs discharged from a single source in the estuary. The shape is distorted as a result of a horizontal tidal circulation system.

The radionuclide flux to the sediment surface and, hence, the radionuclide inventory in the deposits depends on both the activity concentration, which is a function of grain size, and the mean sedimentation rate. In addition, the frequency of deposition events will also affect inventories for short-lived radionuclides.

The size dependence of the estuarine sediment transport processes leads to the size fractionation of the original sediment inputs and the concentration of these fractions in deposits which are separated in either space or time within the estuary. The fractionation process is not precise and particle populations, both in transport and deposited, have a wide range of grain sizes. The width of this distribution, or sorting, varies with the nature of the transport process, being narrower in granular than cohesive sediments. Because grain density will also affect dynamic behaviour, relatively dense grains will be concentrated together with larger normal (light) density grains, e.g. as seen with natural radionuclide-rich `heavy minerals', such as thorium-bearing monazite. This should also be the case with the uranium-based fuel debris `hot particles' in the Esk Estuary, UK, derived from nuclear fuel reprocessing wastes (Hamilton, 1985).

The location of the sediment deposits of different grain size in an estuary will be determined by the current regime or, in general terms, by the energy of the sedimentary environment, i.e. sands are found in higher energy environments and muds in lower. Thus, muds accumulate in relatively deep channels and basins, and sands in shallower channels. Muds are also deposited in the shallow marginal intertidal areas which are covered only during the low-velocity period at the end of the rising tide. The upper parts of these intertidal banks are covered infrequently enough for salt-tolerant vegetation to grow, forming salt marshes. This vegetation has an additional role in reducing the velocity of the water over the marsh, encouraging sedimentation. The distribution of sediment grain sizes over an estuary will be reflected by the distribution of radionuclide activity concentration, with lower activity sands and higher activity muds.

Figure 5.9 (a) Longitudinal distribution of ^238,240Pu(normalized to Al) in surface sediments in the Western Scheldt Estuary, Netherlands (1979

84; pCi/kg = 37 mBq/kg). Duursma et al., 1985; reproduced by permission of the Commission of the European Communities. (b) Distribution of ¹³⁷Cs in surface sediments, Ulhas River Estuary, India (1970

71). (pCi/g = 37 Bq/kg). From Patel et al., 1975; reproduced by permission of Academic Press Inc.

Relatively shallow estuaries should be in a state of near equilibrium, with sedimentation only occurring in response to the migration of the channel systems after occasional erosional episodes, due to river floods and storm tides, or because of changes in the regional environment, e.g. land subsidence, sea-level rise and coastal evolution. Frequently today, sedimentation is the result of human disturbance of this equilibrium, e.g. sedimentation in marginal areas following construction of training walls and groynes or dredging. In general, sedimentation rates may be locally high on short timescales but are mainly low and interrupted by periods of reworking of deposits. For example, in the Hudson estuary, US, long-term average sedimentation rates of 1

3 mm y^-1, determined from radionuclide profiles, appear to be keeping pace with regional subsidence rates (Olsen et al., 1984

85). However, in areas which are recovering from dredging, rates are as high as 700 mm y^-1.

Direct measurements of sedimentation on the intertidal banks in the Esk estuary, UK, showed how radionuclide deposition rates can vary widely over short (monthly) timescales at a single site, e.g. 50

750 Bq m^-2 d^-1 of ²⁴¹Am, due to the effects of changes in tidal amplitude and storm occurrence. These factors affect grain size, and therefore activity concentration, as well as the sedimentation rate (Kelly and Emptage, 1991). There was also a variation between sites in the same intertidal bank environment. Over a two-year period, the sedimentation rates ranged from 4 to 100 mm y^-1 (or 5

82 kg m^-2 ) and ²⁴¹Am deposition rates (at a smaller number of sites) from 21 to 91 kBq m^-2 y^-1. In deep estuaries and fjords, sedimentation will be more continuous, but often sediment supply rates are lower and the overall sedimentation rates are also low.

A variety of radiometric methods have been applied to determining estuarine sedimentation rates from profiles of artificial and natural radionuclides (Figure 5.10), including decay of unsupported natural radionuclides, such as excess ²¹⁰Pb (e.g. Sharma et al., 1987: Clifton and Hamilton, 1982) and the matching of the observed profile with the history of release to the environment. The latter method may be a simple matching of peaks such as with weapons test fallout (Delaune et al., 1978), or may involve matching against the complex history of a nuclear industry operational discharge by statistical fitting techniques (Hamilton and Clarke, 1984). Analytical models have also been used to generate a signal for comparison with the profile from the source term and incorporating factors to allow for the effects of processes such as bioturbation and transport lag terms (Stanners and Aston, 1981). Such radiometric methods of dating sediments and determining sedimentation rates have their limitations, due to uncertainties in the source terms and the possibility of pre- and post-depositional mixing of sediments labelled at different times. In the case of estuaries, a particular problem is likely to be the mixing of freshly contaminated sediment with eroded older contaminated sediment in the marine, terrestrial or estuarine environments.

The residence time for the radionuclides in these sediment stores is obviously an important factor in terms of their radiological consequences. Residence time will be determined largely by the radioactive half-lives of the radionuclides and the physical stability of the deposits.

Figure 5.10 Radiometric dating of radionuclide profiles in intertidal sediments. (a) Excess ²¹⁰Pb S. Carolina saltmarsh, US (Sharma et al., 1987). (b) Weapons test fallout ¹³⁷Cs, Louisiana salt marsh, US. Delaune et al., 1978; reproduced by permission from Nature, vol. 275, pp.532

533. Copyright © 1978 Macmillan Magazines Ltd. (c) Fuel reprocessing discharge ¹³⁷Cs, sediment profile (blocks) and modelled profiles (lines), Esk Estuary, UK. From Stanners and Aston, 1981; reproduced by permission of Academic Press Inc. (pCi/g = 37 Bq/kg).

In a steady-state estuary, channel and lateral deposits will be reworked by the migration of the channel systems on time scales of 1

10² y and longer for large delta systems. Thus, sediment inputs will be balanced by sediment outputs to the sea and the fluvial sediments are effectively bypassed to the sea. This is the condition for the Mississippi and other delta channel estuaries. However, for most estuaries, the steady-state condition does not exist, due to changes in relative sea level caused by the post-glacial worldwide rise in sea level and local tectonic and isostatic crustal movements, both up and down. Consequently, deep estuaries, e.g. Chesapeake Bay and fjords formed by the deglaciation and Holocene rise of sea level, are still accumulating sediment in their basins. In shallow estuaries and parts of estuaries, continued sedimentation may be keeping pace with subsidence, as has been suggested for eastern US intertidal marshes (Sharma et al., 1987).

In contrast, on short timescales, individual estuaries,can vary from a condition of net gain to one of net loss, as the magnitudes of the sediment inputs and outputs change. In strongly tidal estuaries this can give a spring-neap cycling in the net balance, whereas in estuaries where there is a seasonal imbalance in river water discharges the cycle will be seasonal, e.g. net gain when river discharges are low and net loss when discharges are high. Such a relationship is important: it can result in the under-representation of the river sediment in the estuarine deposits, i.e. the fluvial sediments are bypassed direct to the shelf sea during high river flow periods.

In estuaries, many elements and, consequently, radionuclides, exhibit non-conservative behaviour, i.e. they undergo a change in the distribution between the particulate and solution phases, due to sorption or flocculation/precipitation reactions, or to biological uptake. Non-conservative behaviour needs to be differentiated from the effects of mixing of waters or sediments without particle-solution reactions. Such a mixing process will result in concentrations in each phase lying along a straight line between those of the two end members for that phase. Non-conservatism is most clearly demonstrated by comparing the ratio of solution phase activity concentration to the concentration of a conservative component such as salinity. The resultant distribution (Figure 5.11) shows departures from the mixing or dilution line if non-conservatism occurs, indicating solution loss or gain relative to the particulate phase. It is essential, therefore, that the end members for any mixing processes going on are well defined, in order to rule out mixing as an explanation. Theoretically, it should be possible to demonstrate non-conservatism from particulate phase data (suspended or bottom sediments) but it is often difficult to isolate this from mixing effects. This problem, mixing versus non-conservatism as an explanation of particulate phase pollutant distributions, has been debated for some time with respect to stable element behaviour in estuaries. It is now considered that the best evidence for non-conservatism comes from field data on solution phase distributions or laboratory data on partitioning of elements between particulate and solution phases using environmentally representative polluted sediments and waters (e.g. Duinker, 1980). Evidence from suspended particulate phase and K_d distributions is particularly difficult to interpret because of the existence of grain size gradients in the same direction as the salinity gradient, with finer sizes towards the head of the estuary giving higher particulate activity concentrations.

Figure 5.11 Particle

solution sorption reactions. (a) Desorption from fluvial sediment of ⁵¹Cr and ⁶⁵Zn, Columbia River Estuary, US. From Evans and Cutshall, 1973; reproduced by permission of the IAEA. (b) Desorption from marine sediment of ^239,240Pu Esk Estuary, UK. From Assinder et al., 1984; reproduced by permission of Selper, Ltd. (c) Absorption scavenging by (fluvial?) sediment of ^239,240Pu, Chesapeake Bay, US. From Sholkovitz and Mann, 1987; reproduced by permission of Academic Press, Inc. (pCi/l= 37 Bq/m³; dpm/100 kg = 0.17 Bq/m³).

Four processes can be recognized in estuaries which lead to non-conservatism, i.e. mixing of fresh and saline water, scavenging by sediment, degradation of organic matter, and biological uptake. Sorption reactions generally characterize the first two situations. In water mass mixing, these reactions are a response to the change in the geochemical environment which is associated with a corresponding change in K_d The reactions are able to take place because particles equilibrated with a solution of one salinity come into contact with a solution of another, either by tidal mixing of suspended sediments or as a result of the tidal and non-tidal excursions of the boundary between saltwater and freshwater in the estuary. A particular radionuclide may undergo either desorption or adsorption, depending on the relative levels of particulate and solution phase radionuclides in the inputs of different salinities. Sediment adsorption of radionuclides (scavenging) can occur in response to a decrease in the K_d or with constant K_d when uncontaminated or under-contaminated sediment is added to the suspended sediment population. Organic degradation can result in a change of radionuclide species via redox reactions or organic complexation which affects their partitioning between the solution and particulate phases.

Non-conservative behaviour can vary with radionuclide and between different estuaries, as well as with space and time in the same estuary. Important secondary factors will be those which determine the availability of the sediment bound radionuclide on the appropriate timescale, e.g. the suspended sediment concentration and the fraction of the sediment bound activity which has reaction kinetics of the appropriate rate, i.e. rapid on the timescale of the residence time of sediment in the water column (hours, days) and the physical and biological processes which lead to sediment resuspension. The relative importance of these has been studied with radiotracers in artificial enclosures in a shallow marine environment (e.g. Li et al., 1989).

Desorption of radionuclides from a fluvial sediment input has been demonstrated for the Columbia River Estuary, US, which received large amounts of particulate and solution phase radionuclides from the Hanford plutonium production plant between 1944 and 1972 (Evans and Cutshall, 1973). Field data showed that, for the longer-lived neutron activation products, the behaviour of ¹⁴Mn and ⁶⁵Zn was non-conservative whilst that of ⁵¹Cr and ¹²⁴Sb was conservative (Figure 5.11 a). Laboratory experiments with environmental materials showed that about half of the ⁵⁴Mn and one-third of the ⁶⁵Zn could be desorbed by a slow reaction (1 week) from suspended river sediments. This rate is slow compared with typical desorption reactions and it suggests that the release mechanism may be complex. In general, remobilization from fluvial sediments can occur for radionuclides of heavy metal elements for which there is good evidence of non-conservative behaviour in estuaries, especially those with high organic loads. For example, in the Scheldt Estuary, Netherlands and Belgium, Mn, Cd, Zn have been shown to be released from contaminated fluvial sediments (Duinker et al., 1982). For the first two metals, redox reactions form species of different solubility in the almost anaerobic zone generated in the upper estuary by the oxygen demand of the organic matter (sewage), whereas Zn may be being released directly by degradation of Zn bearing organic particles.

However, in the Elbe Estuary, Germany, Cd, Pb, Cu, Cr, Zn show conservative behaviour, in keeping with a sediment and water mixing model (Förstner and Wittmann, 1979).

Desorption of Cs from fluvial sediments has been proposed for many estuaries, especially in Europe and America, using a variety of evidence, but very few of these examples provide incontrovertible evidence for caesium desorption. Since the experimentally determined K_d for Cs is around 10³ l kg^-1 in seawater and 10⁵ l kg^-1 in freshwater, it is to be expected that desorption should occur as a result of exchange with seawater ions, where fluvial sources of Cs dominate. However, Cs is known to occupy adsorption sites on sheet silicate minerals (clay minerals/micas) with different degrees of exchangeability, at least in freshwater sediments (Evans et al., 1983). Thus, it is important to consider the availability of labile Cs with the appropriate reaction kinetics when assessing the likelihood of desorption. For example, Zucker et al. (1984) considered that labile, reactor-derived, ¹³⁴Cs was being desorbed from sediments in the James River Estuary, on the evidence of the ¹³⁴Cs : ⁶⁰Co ratio in sediment deposits, whereas fallout ¹³⁷Cs, fixed in interlattice sites, was not desorbed. Desorption of ¹³⁷Cs (which may have had a reactor source) also was inferred by Linsalata et al. (1985), from solution activities in the Hudson River Estuary, US, which increased non-linearly with salinity, although the seawater end-member concentration was not determined.

Evidence for desorption based on particulate phase activities alone has to take into account the possibility that the observed trends could be produced as a result of mixing of sediment from different sources. Much of the evidence from US estuaries comes from variations in ratios of Cs : Pu radionuclides in bottom sediments, e.g. Columbia River (Beasley and Jennings, 1984), Hudson River (Olsen et al., 1981) Savannah River (Hayes and Sackett, 1987). At least in the Savannah River, Olsen et al. (1989) consider that the bottom sediment radionuclide ratio is evidence for dilution of the fluvial sediment by Pu-bearing marine sediment rather than for Cs desorption. This is supported by their experimental determination that the fallout Cs in the fluvial sediments is non-exchangeable, and from a consideration of the Pu isotope ratios in the inputs. Such an explanation is compatible with the landward transport of marine sediment by the non-tidal circulation of partially mixed estuaries. Desorption of ¹³⁷Cs from bottom sediments, together with io6Ru and ¹⁴⁴Ce, has been described also from the Ulhas River Estuary (Trombay), India (Patel et al., 1978). There, levels in a surface sediment layer were found to decrease with a half removal time, ignoring radioactive decay, of 2 y following decreases in discharges from the nuclear installation. Since the environmental half-lives for ¹⁰⁶Ru and ¹⁴⁴Ce were different, it was argued that dilution by new sediment could not explain all of this decrease. However, a fuller study, based on core total inventories, would be needed to provide absolute evidence for desorption. The evidence from the Loire, France (Jeandel et al., 1981) also is ambiguous in that the suspended particulate phase seems to show non-conservative behaviour whilst the solution phase remains conservative. This does not support the case for desorption and the data might be explained by dilution with uncontaminated sediment from an estuarine source.

Desorption of plutonium from marine sediments has been extensively investigated in the Esk Estuary, UK, where the discharges to the sea from the Sellafield reprocessing plant provide marine sediment with higher activity concentrations than on the fallout-contaminated fluvial sediment input (e.g. Assinder et al., 1984; Eakins et al., 1985). In particular, plutonium in solution shows enhanced levels in the estuary due to desorption from the sediments of marine origin, as compared to the range of levels predicted by dilution curves (Figure 5.11b; note that the seawater activities vary during the tidal input).

Following a recent reduction in the levels of Pu discharged from Sellafield, the levels of Pu present in solution in the estuary, after desorption has occurred, can be higher than in the seawater entering it. The behaviour of Pu contrasts with the more conservative behaviour of ¹³⁷Cs and ¹⁰⁶Ru from the discharges. Desorption occurs because the tidal excursion on a spring tide in this shallow, 13 km long estuary, brings low-salinity water into contact with sediment deposited from full-salinity seawater. It is assumed that desorption follows resuspension of these sediments into the low-salinity water, as well as from the dispersion of suspended marine sediment into low-salinity water during the mixing of salt and freshwaters. Laboratory experimentation with Esk sediment has shown that only a small labile fraction (< 5 per cent) of the plutonium on the marine sediment is involved in a rapid (< 1 h) exchange reaction in the freshwater (Hamilton-Taylor et al., 1987). The experiments have also shown that it is primarily Pu(IV) species which are involved in the sorption reactions. However, the Pu behaviour is further complicated by redox reactions, between Pu(IV) and Pu(V), both on the sediment surface and in solution (Mudge et al., 1988).

Estuarine scavenging of Pu has been proposed for the Connecticut River, Delaware Bay and Chesapeake Bay Estuaries, US (Figure 5.11c) (Sholkovitz and Mann, 1987), from a study of the distribution of solution and particulate phase activity concentrations. The sources of the solution phase Pu for these estuaries are fallout Pu in the marine and fluvial inputs, with the highest activities in the former. The origin of the sediment is not clear but is likely to be older, uncontaminated or lightly contaminated estuarine sediment and, maybe, fluvial sediment. In these examples, the loss of Pu to sediment appears to be inversely related to the flushing time in the estuary, being greatest in Chesapeake Bay with a 6

12 month residence time. A similar non-conservative behaviour of Pu was observed in estuaries in France, e.g. Seine, Loire and Gironde estuaries (Jeandel et al., 1980, 1981). At the time, coagulation was considered as one of the possible causes but, in the absence of a high dissolved organic carbon content in the river, there seems no reason to consider it as anything other than an adsorption reaction. The dominant sources of the Pu in solution in these French estuaries appear to be marine, with a fallout origin alone in the Gironde and fallout plus industrial discharges in the others. In the Gironde, at least, the sediment particles which are scavenging the Pu are considered to be of fluvial origin (Elbaz-Poulichet et al., 1982). In the Seine and Gironde estuaries, the particulate fractions show high activity concentrations persisting down to low salinities and this may be a residual grain size effect (not removed by the normalization procedure) combined with scavenging and mixing effects, although other explanations were also considered (Jeandel et al., 1981).

The difference in behaviour of marine Pu inputs in these estuaries and in the Esk Estuary, where desorption occurs, may be due to differences in the speciation of the Pu from the fallout and discharge sources (Sholkovitz and Mann, 1987) and/or to the different characteristics of the estuarine systems, i.e. no significant uncontaminated sediment inputs are available in the Esk. Interestingly, it has been suggested that both reactions may be occurring in the Savannah River Estuary, US (Olsen et al., 1989), where the Pu levels in the solution phase suggest that desorption is occurring at very low salinities and adsorption at higher brackish salinities. It is considered that the scavenging leads to a threefold increase in particulate ^239,240Pu activity concentrations. Since the suspended sediments are depleted in ⁷Be whilst enriched in Pu, it is deduced that the frequency of the resuspension events and scavenging process is slow relative to the ⁷Be half-life (53 d).

Natural radionuclides, also, can be involved in estuarine sorption reactions. Highly particle-reactive radionuclides should show adsorption. This is clearly the fate of the direct atmospheric input of excess ²¹⁰Pb to estuaries (i.e. unsupported by ²²⁶Ra). Also, scavenging of dissolved excess ²¹⁰Ph from marine inputs has been suggested for the Savannah River Estuary (Olsen et al., 1989).

Thorium is also largely present on particles in aquatic systems. Adsorption of thorium isotopes onto sediments occurs in the Ribble Estuary, UK, from nuclear fuel fabrication plant discharges direct to the estuary (Hunt, 1977

92). Uranium, however, mainly behaves conservatively in estuaries (e.g. Toole et al., 1987). Exceptions to this exist where there is a significant fluvial input of anthropogenic U from phosphate processing, e.g. Charente Estuary (Martin et al., 1978), where loss of solution phase U occurs in the estuary by co-precipitation with phosphate. In contrast, radium (²²⁸Raand/or ²²⁶Ra) desorption from river sediments has been shown from several estuaries including Hudson (Li and Chan, 1979), Scheldt (Kirchmann et al., 1985), Mississippi (Key et al., 1985) and Amazon (Moore and Scott, 1986). Differences in the clay mineralogy were considered to be the reason for the considerably higher amounts desorbed in the Mississippi. A reduction of Ra concentrations at higher salinities is the result of mixing of the enriched estuarine water with seawater which has a low Ra content due to the biological removal of Ra by plankton.

Colloidal Fe and humic acids, present in the operationally defined `solution' phase of freshwaters, are known to flocculate or coagulate in the low-salinity zone of estuaries (e.g. Sholkovitz et al., 1978). Experiments with freshwater with a high dissolved organic carbon content showed that dissolved Pu was lost to the coagulated Fe humic acid colloids when mixed with seawater (Shen et al., 1983). This is likely to be due to the Pu being associated with the original organic colloids in solution. This mechanism is considered to be involved in the loss of Pu from solution in the organic rich river water in the Mullica River Estuary, US (Sholkovitz and Mann, 1987).

Non-conservative behaviour of dissolved constituents in estuaries can occur through biological uptake. This will affect particularly the major nutrients, such as C, N, P, K. In addition, plankton blooms potentially could have an effect on micronutrients such as trace metals. As well as being involved in metabolism, passive adsorption onto cell surfaces could occur (see Wolff, 1980, for a summary). Such studies have not been carried out in estuaries.

Diagenetic, or post-depositional processes modify the distribution of radionuclides and their partition between solution and particulate phases in estuarine deposits.

Mixing of sediment by bioturbation in estuaries is due to a variety of sediment-inhabiting organisms, e.g. annelids, arthropods, molluscs and echinoderms. Although the fauna is characteristically species-poor, because of the variable salinity, it can have high population densities, with several species having densities in the range of 10³

10⁴ m^-2. The depth of this bioturbated zone in estuaries varies between a few centimetres (e.g. Corophium) to 10s of centimetres for larger annelids and molluscs. The intensity of sediment disturbance depends on the relative rates of bioturbation and sedimentation, but it can be sufficient to totally destroy the original sedimentary structure and smooth out variations in radionuclide fluxes to the sediment.

Advective movements of porewater fluids will also take place, due to consolidation, surface infiltration, lateral recharge and groundwater flow. In the Esk Estuary, UK, percolation down open burrows resulted in the rapid labelling of the upper 5 cm of sediment with Chernobyl ¹³⁴Csfrom rainwater and with Sellafield-derived ⁹⁵Zr from seawater (Kelly and Emptage, 1991). Labelling due to percolation to greater depths down desiccation cracks and by lateral recharge have also been used to explain the presence of short-lived radionuclides (⁹⁵Nb) at depth (70 cm) in Esk sediments (Carr and Blackley, 1985). This mechanism can result in the extension of the zone of contaminated sediments.

Changes in the porewater chemistry can result in redistribution of radioactivity between the particulate and solution phases. The advective porewater transport described above could change the salinity in the porewater, leading to sorption reactions. Also, changes in redox state are a feature of sediment deposits, from oxic to anoxic conditions, due to biogeochemical reactions. This is also true of sub-aerially exposed intertidal sediments, especially muds. It is well known that such elements as Fe and Mn are mobilized from sediments by these processes and redistributed by diffusion within the sediment body or lost to overlying waters, e.g. in the Mississippi delta half the Mn has been lost from the sediments (Presley and Trefry, 1980). Similar fluxes of Mn have been described from other estuaries, e.g. Tees and Conway Estuaries, UK, (Elderfield and Hepworth, 1975). In the last two instances it was suggested that Cu, Zn, Co, Ni, Pb and Fe were also released. Conversely, it is considered that anoxic bottom sediments scavenge U from solution in surface waters by precipitation of reduced U phases, e.g. Narbada Estuary, India (Borole et al., 1982) and Ogeechee Estuary, US (Maeda and Windom, 1982).

It is to be expected that radionuclides of the metallic elements listed above, as well as others with more than one redox state, can similarly undergo redox remobilization. Although radioactive tracers have been used in studies of the anaerobic bacteria sulphate reduction process which underlies the redox changes (³⁵S and ⁵⁵Fe) (King, 1983), detailed studies of radionuclide diagenesis have not been carried out in estuarine areas. However, studies in other aquatic environments have shown that Cs remobilization takes place by exchange with anaerobically produced NH₄ (Comans et al., 1989). In contrast, similar studies have been unable to demonstrate the occurrence of remobilization of Pu (from both fallout and nuclear industry discharges) and americium by redox-related reactions (Buesseler and Sholkovitz, 1987; Malcolm et al., 1990).

The interaction between the radionuclide sources and the system of hydrodynamic, sediment dynamic and geochemical processes in an estuary determines the balance between the inputs and outputs. Several estimates have been made of Pu budgets for individual estuaries. These emphasize the important role of estuaries as transient or permanent stores of particle-reactive radionuclides.

A budget for Pu over a single tidal cycle for a part of the Esk Estuary (Kelly et al., 1991) shows how the particulate phase behaviour of a highly particle-reactive radionuclide like Pu can dominate the budget when there is an adequate supply of sediment (Figure 5.12a). The only significant Pu source was the Sellafield reprocessing plant marine discharges, mainly introduced to the estuary via fresh marine sediment, but also partly via eroded older contaminated estuarine sediment. In this well-mixed estuary, tidal advection produced relatively large fluxes of ^239,240Pu (and other radionuclides) in and out of the estuary over the tidal cycle (input 283 MBq per tide). However, overall, there was a net input of sediment (18 t per tide) and of Pu (85 MBq per tide) to the estuarine sediment deposits. The solution phase was a small component of the budget, although it showed a net output to the sea, due to the desorption of Pu from suspended sediment in the estuary. The extent to which these data indicate the state of balance of the Pu budget over a longer period is not known. Not only is it a question of the sediment budget, but also of the activity concentrations of the sediment inputs, since declining discharges could mean that the higher activity concentration of the eroding older contaminated sediment in the estuary could result in net output of activity, as has been claimed (e.g. Burton and Yarnold, 1988). However, changes in the discharges are likely to be buffered by the availability of older contaminated sediment on the floor of the Irish Sea.

From sedimentation rate measurements, the annual net sediment input to the main depositional areas in the whole Esk Estuary, i.e. intertidal banks, is estimated at 15.5 kt y^-1. This gives a net input to these areas of about 40 GBq y^-1 of ^239,240Pu, if the activity concentration of the input is assumed to be the same as measured in the tidal budget study.

A mean annual Pu isotope budget has been estimated for the Western Scheldt (Duursma et al., 1985) (Figure 5.12b). Four different sources supply the Pu: catchment fallout (river input), direct fallout, estuarine and marine nuclear industry discharges. Their balance shows that direct fallout to the estuary is not a negligible term. However, the budget emphasizes the important role of nuclear industry discharges, even when the discharge points are remote, as is this case where the marine sources are the discharges on the west coasts of France and UK. The marine input given is the net value needed to balance the budget and an estimate of the gross input would be higher, including a flux equal to the mean tidally averaged advective and dispersive transport to the sea of the solution and, perhaps, the particulate phase Pu. Overall, the budget again shows the importance of the estuarine sediments as a sink for Pu, with a deposition of 1.2 GBq y^-1of ^239,240Pu.

A similar approach was used for determining the Pu budget for the Hudson River Estuary by Olsen et al. (1984

85) who considered that about 0.8 GBq y^-1 of fallout ^239,240Pu was delivered in the particulate phase by the river and about the same amount from the sea, at a higher activity concentration on a smaller amount of sediment, i.e. giving a total deposition to the sediment deposits of about 1.5 GBq y^-1.Since two-thirds of the total sediment input (1.5 Mt y^-1) is removed by dredging, there is a major loss of Pu from the estuary, although a net gain occurs in non-dredged areas. Any sediment scavenging of Pu from solution, as described from other US east coast estuaries (Sholkovitz and Mann, 1987), is probably only a small fraction of the Pu delivered as sediment labelled with fallout.

The inventories of artificial radionuclides stored in the estuarine sediments are the result of the 50 y of inputs from fallout and nuclear industry liquid discharges (Table 5.4). The inventories are relatively low in estuaries mainly contaminated by nuclear weapon test fallout, ~2 GBq km^-2 for ¹³⁷Cs and 0.1

0.2 GBq km^-2 for ^239,240Pu, but considerably higher inventories are found where discharges are high, with ~800 GBq km^-2 ¹³⁷Cs and ~70 GBq km^-2 for ^239,240Pu in a total of 1.2 Mm³ of contaminated sediment in the Esk Estuary (Burton, 1988; Kelly and Emptage, 1991).

Figure 5.12 Radionuclide budgets for estuaries. (a)^239,240Pu budget for a single spring tide (12.5 h), Esk Estuary, UK. From Kelly et al., 1991. (b) Estimated annual ²³⁸pu and ^239,240Pu budgets, Western Scheldt Estuary, Netherlands. From Duursma et al., 1985; reproduced by permission of the Commission of the European Communities.

Table 5.4 Estimated inventories in estuarine sediments for artificial radionuclides from fallout and nuclear industry liquid discharges

The inventory for ^239,240Pu in the Esk Estuary, together with the estimated annual sediment input given above, would indicate a residence time of 9.5 y for the Pu-contaminated sediments if the estuary was in a steady state. In reality this is an underestimate, and the estuary is still gaining activity, at least for the actinides. The residence time for contaminated sediments in the Hudson Estuary is greatly shortened by dredging, with two-thirds of the inventory in Table 5.4 having been removed for offshore dumping (Olsen et al., 1981).

A wide variety of other coastal environments exist besides estuaries, although none exhibit the full range of physical and geochemical processes seen in estuaries. In particular, the particle

solution reactions in the water column associated with the mixing of fresh and saline water are unimportant. The behaviour of radionuclides in these environments can be considered as variants of their behaviour in estuaries and, consequently, a detailed discussion is not provided. The main environments concerned include beaches, embayments, lagoons and barrier saltmarshes.

Beaches are essentially high-energy environments whose physical processes are dominated by wave activity, with a wave-generated circulation system. The sediments are characteristically coarse-grained sands and gravels, derived from the seabed offshore, from coast erosion and, via longshore transport, from rivers. Beaches are themselves the source of sediments for aeolian dune systems built behind the beach. Beaches are often transient stores for sediments and they typically undergo a cyclical development which is often seasonal, with removal during storms and accumulation during quiet weather periods. Longer-term storage occurs where there is construction of spits and barrier island complexes. The coarse grain size means that radionuclide activity concentrations normally will be relatively low on beaches. However, their total inventories may still be significant, e.g. along 20 km of coast close to the discharge point from Sellafield, UK, the beaches (7 km²) had inventories of 1800 GBq ^239,240Pu, and 3690 GBq ²⁴¹Am (1983/84) (Eakins et al., 1990). Also, where activity is associated with organic particles this may end up on beaches in the flotsam deposited at the tide-line, as was found to have occurred after an accidental release of activity from Sellafield (DOE, 1984), which contaminated organic objects in the flotsam with up to 69 kBq ¹⁰⁶Ru.

Embayments, characterized by relatively large intertidal areas, are typical of shallow coastal seas and/or areas with large tidal ranges. The sediment sources are principally from offshore or longshore transport. The tidal energy and degree of exposure to wave activity determine the distribution of sediment grain size and, thus, the relative activity concentrations in the deposits. Typically, however, there will be a zonation from an outer zone of sand banks, succeeded by mud banks and, finally, salt marshes.

Lagoons are morphologically similar to some estuaries, in being partially enclosed from the sea, but by definition lack the freshwater input. They are developed by the construction of spits/barrier islands, as on the coasts of the eastern US and south-east North Sea. Barrier salt marshes are similar in origin but lack the open water lagoon. Lagoons can also originate from coral reef construction. They are essentially low-energy environments, with weak circulations due to tidal discharges through the restricted inlets and density currents generated by the development of hypersalinity (in warm climates). Consequently, sediments are characteristically muds, both intertidal and subtidal, except for deltas of coarser sands formed at tidal inlets and where storm washover occurs over the barrier with the open sea. In coral reef areas, the sediments will be characteristically composed of carbonate minerals unlike the predominantly siliceous minerals elsewhere. With the fine grain size will be associated relatively high radionuclide activity concentrations. Also, in the lagoon environment the potential for rapid reworking of the sediment is low and sediment residence times will be on a geological timescale, determined by the evolution of the coastline. Although coral reef lagoons are unusual environments they have a particularly significant place in the story of the radioactive contamination of coastal environments, as coral islands in the Pacific Ocean have been a favourite site for nuclear weapons testing since the early days of the nuclear industry.

The future levels of radioactivity in coastal areas will depend not only on the future releases of radioactive materials to the environment but also on the behaviour of the already-contaminated areas, which can act as secondary sources of solution and particulate phase radionuclides for coastal environments. These releases can be further exacerbated by human activity, insofar as changes in land use, increase or decrease in agricultural activity, mining or construction, will affect sediment yield from the catchment and fluvial sediment input to the estuary. In addition, the impact of contaminated estuarine and coastal areas will be influenced by the future management of these areas, with major changes in use potentially affecting radiation exposure pathways. The possible changes include:

5.4 COASTAL, SEMI-ENCLOSED BASINS, SHELF AND CONTINENTAL MARGINS

Coastal zones and shelf seas are considered here to represent the region from the low-water mark along open coasts to the shelf/slope oceanic boundary. This section also considers semi-enclosed basins representing distinct marginal zones having limited exchange with adjacent coastal and shelf regions. General descriptions of the physical, chemical and biological characteristics of such environments can be found in a number of texts such as Dyer (1986); the main processes of interest for the understanding of radionuclides are summarized in Section 5.1.

The sources of artificial radionuclides in the environment are described in the General Introduction. In regions remote from nuclear installations, inputs of artificial radionuclides in the marine environment have been dominated largely by fallout from nuclear weapon tests performed in the atmosphere. The coastal marine environment can also receive direct inputs of artificial radionuclides arising from the discharges of nuclear facilities: nuclear power plants, nuclear reprocessing plants and industrial users of radioactivity. The largest quantities of low-level radioactive waste discharges have resulted from nuclear reprocessing plants, either by direct discharge of the liquid effluents into the marine environment (Sellafield, La Hague, Bombay) or via rivers (Hanford and the Columbia river in the Pacific; Marcoule and the Rhône river in the Mediterranean Sea). Of these sources, the Sellafield reprocessing plant has had the greatest impact and the largest number of studies carried out (see General Introduction and review in IAEA, 1985a). In contrast, routine releases from nuclear power plants and industrial users of radioactivity have had a much lower impact. The Chernobyl accident in 1986 produced a significant input of artificial radionuclides in the marine environment, both by direct fallout, especially on the Baltic, Mediterranean and Black Sea, and by indirect inputs, for example via the Dniepr into the Black Sea and thence into the Mediterranean Sea (Figure 5.13). As regards the open ocean, in addition to nuclear weapons test fallout, one must address the inputs due to the disposal of low-level radioactive wastes in the deep ocean, and be aware of the accidental releases from satellites re-entering the atmosphere and from nuclear-propelled submarine failures (IAEA, 1992a).

Figure 5.13 Distribution of ¹³⁷Cs (mBq l^-1) in the surface waters of the Black Sea and Aegean Sea in 1988. From Polikarpov et al., 1991.

The physical and chemical form of radionuclides in the source term affect their initial behaviour and distribution in the receiving environment. It may be difficult to predict the chemical form of any specific radionuclide in view of the many processes which take place at a major reprocessing site, and the differences in technology which may occur with time and between sites. Of particular importance is the occurrence of particulate forms. In one study over 95 per cent of the Pu and Am in the Sellafield effluent was associated with a particulate fraction, comprising a ferric hydroxide floc, with a significant quantity present as discrete `hot' particles, defined as clusters of alpha-tracks revealed using the CR-39 dielectric detector (Pentreath et al., 1986). Such `hot' particles, unequivocally having an effluent origin, have been observed in environmental samples and may persist for several months following release (Hamilton, 1985). It cannot be assumed that radionuclides from different sources will behave similarly. For example, ⁶⁰Co released from an experimental facility at Winfrith, UK, behaves differently from ⁶⁰Co released from Sellafield. Most of the Pu discharged from Sellafield is in the reduced form (III+IV), but there is a progressive change to the oxidized form on mixing with seawater. The K_d values of the reduced and oxidized forms differ by about two orders of magnitude. Studies carried out in the late 1970s showed that between 70 and 90 per cent of the plutonium present in the filtrate fraction was in an oxidized form, either Pu(V) or Pu(VI), while that adsorbed to the particulate fraction was almost entirely in a reduced form, Pu(III) or Pu(IV) (Nelson and Lovett, 1978). Little variation in oxidation state distribution was evident over a concentration range of 0.05

50 Bq m^-3. In contrast to coastal zones, it has been reported that the two oxidation state groups are present in almost equal concentrations in the open ocean. Little variation with depth is observed except near the bottom of the water column where the distributions change rapidly and become similar to those found in shallow waters. It is now generally accepted that the reduced category is mainly Pu(IV), very likely as Pu(OH)₄ adsorbed on negatively charged suspended sedimentary material, while the oxidized category is predominantly Pu(V) probably as PuO₂⁺ and its complexes.

Of special interest is the comparison of the role of colloidal or pseudo-colloidal solutions in freshwater and seawater systems. The formation of colloidal solutions is always observed in solutions in which insoluble hydrolysis products can be formed. Pseudo-colloidal solutions consist of adsorption formations of various dimensions on foreign colloidal particles present in the solution. Compared to the freshwater environment, relatively few data have been reported on the association of radionuclides with colloids in seawater. One of the earliest studies to link the enhanced mobility of Pu and Am to colloidal aggregates was carried out in the lagoon waters of Bikini Atoll by Nevessi and Schell (1975) who reported that almost all of the Am in the soluble phase was very probably in the colloidal form.

Later work showed that about 12 ± 8 per cent (standard deviation) of the Pu in lagoon samples was present in a colloidal state. The presence of colloidal organic carbon (COC) is known to have a profound effect on the affinity of a number of radionuclides for the particulate phase (see Section 5.2.4.2). For example, the K_d coefficients of both Pu and Am have been shown to decrease significantly at COC concentrations in excess of about 1-10 mg 1^-1. In open waters, however, COC concentrations are unlikely to reach these levels and the fraction of Pu and Am associated with colloids or pseudo-colloids is considerably smaller. Moreover, recent studies based on ultrafiltration techniques suggest that in saline environments colloidal Pu and Am appear in the < 10 000 Dalton size fraction, in sharp contrast to the freshwater environment. There is also evidence that the colloidal fraction of both elements is in a chemically reduced form, i.e. Pu(IV) and Am(III). Clearly, differences in chemical form may have a significant impact on the initial and subsequent behaviour of radionuclides in coastal waters, which is likely to be reflected in resulting pathways.

The behaviour of most radionuclides is linked to particle dynamics because of their affinity with particle surfaces. Thus, for radionuclides with a high K_d, such as Pu and Am, their fate will largely be controlled by sedimentation processes (sediment accumulation, resuspension and transport). In the Irish Sea, for instance, the distribution of plutonium broadly reflects the distribution of settled sediments, although the influence of local water movements is also important (Figure 5.14). Radionuclides with a low K_d (Tc, ³H) will be controlled mainly by water mass circulation. Nuclides which have an intermediate K_d such as Cs will be predominantly influenced by the water circulation but also affected by sedimentation processes.

Surveys of radiocaesium concentrations in filtered waters from throughout the Irish Sea have been conducted on an annual basis for many years (Mitchell, 1967

77; Hunt, 1977

92). The extensive data available show that general levels reached a peak in the years between 1975 and 1978 and have been declining more or less steadily ever since. In Figure 5.15 the distribution of ¹³⁷Cs in the waters of the Irish Sea in the years 1974, 1976, 1981 and 1986 is reproduced and gives some indication of the temporal evolution of these levels as well as the decrease in concentration with increasing distance from Sellafield. Concentrations in the waters of the Irish Sea attained 5

50 Bq kg^-1 and were more than three orders of magnitude higher than representative fallout levels at similar latitudes. Data on the distribution of other radionuclides in the waters of the Irish Sea are not as extensive, though the distributions of ^239,240Pu and ²⁴¹Am in filtered waters have been determined in a number of surveys. For both elements the fall-off in concentration with increasing distance from Sellafield is noticeably more rapid than for ¹³⁷Cs. The geographical distribution of Sellafield-derived radioactivity in the Irish Sea is in close accord with the pattern of water circulation once account is taken of the affinity of certain radionuclides for sedimentary material and the distribution of the fine sediments themselves. For example, the influx of Atlantic seawater from the south is clearly evident from the shape of the isolines of ¹³⁷Cs concentrations in seawater depicted in Figure 5.15. It has been shown that the caesium discharged from Sellafield mixes with the seawater transported through the Irish Sea in a northerly direction, exits through the north channel and follows a well-defined path around the west and north coasts of Scotland where it enters the North Sea.

Figure 5.14 Inventory of ^239,240Pu in the Irish Sea (MBq m^-2) adapted from Woodhead, 1988. (a) Total core; (b) at depth > 5 cm; (c) at depth > 71 cm.

The relative importance of biologically mediated removal, transformation and regeneration mechanisms of radioactive materials is a function of the production of oceanic biomass. At any given location, most of the oceanic biomass resides in the upper few hundred metres. Thus, the zone of most active biogenic particle production, particularly near oceanic margins, is in the surface layers where primary production occurs. Therefore, to understand fluxes and transfer of radionuclides it is important to evaluate the biologically mediated processes which regulate their biochemistry of uptake, removal and regeneration. This mechanism is of particular importance in upper ocean and deep marginal basins such as the Mediterranean Sea. The effect on the behaviour of radionuclides from global fallout has been studied over a period of 20 years but the Chernobyl accident in 1986 provided additional information that helped to quantify the effect of these processes. Three such studies, involving the use of sediment traps, have been reported for the North Sea, the Black Sea and the Mediterranean.

Primary production can occur on the sediment surface in coastal waters, in contrast to the open ocean, since in shallow waters the photic zone reaches the seabed. Films of microalgae, such as diatoms and other photoautotrophic microbes, develop on the surface sediment particles as light intensities increase during the spring. This benthic primary production has a cyclical pattern synchronized to that in the plankton with benthic microbes rapidly oxidizing particulate matter sedimented from the water column and returning nutrients, and in some cases sorbed radionuclides, in soluble form, to the water column. Indeed during the winter this process resupplies the water column with the nutrients which prime the plankton production cycle for the next spring.

The importance of scavenging and removal of surface-introduced radionuclides by incorporation into plankton and their particulate detrital products (e.g. faecal pellets, moults, eggs, carcasses and algal flocs) has been derived from both laboratory and field studies. In the north-western Mediterranean during the mid-1970s, freshly collected euphausiid faecal pellets contained on average 5.3 Bq ^239,240Pu kg^-1. Applying these relatively high concentrations to estimates of zooplankton faecal pellet production rates, plutonium fluxes via sinking faecal pellets were derived. This in turn led to the computation of an upper mixed layer residence time due only to sinking faecal pellets of 3.6 y for ^239,240Pu in this part of the Mediterranean. Only recently have direct measurements of the vertical flux of ^239,240Pu been made by sediment traps moored in the Gulf of Lyons (Fowler et al., 1990a). These plutonium fluxes, which were measured in an area where particulate mass fluxes are relatively high, resulted in a relatively short ^239,240Pu upper water layer residence time of 2.5y. The similarity in the two residence time estimates is noteworthy; however, characterization of the sediment trap material indicated that its composition was only partially biogenic. In this region, lateral transport of lithogenic material in the form of nepheloid layers also contributes to the vertical particulate flux, and that of associated radionuclides.

Figure 5.15 Distribution of ¹³⁷Cs in filtered surface seawater in the Irish Sea (a) July 1974 (pCi l^-1); (b) January 1976 (pCi kg^-1); (c) March 1981 (Bq kg^-1); (d) April 1986 (Bq kg^-1). The broken line represents the approximate position of the 0.5 Bq kg^-1 isopleth in each case. Adapted from Mitchell, 1967

77.

The most convincing data supporting the involvement of biota in the transfer and transport of artificial radionuclides in enclosed seas comes from experiments carried out prior to and following the Chernobyl accident (Fowler et al., 1987, 1990b). In the Mediterranean, high-resolution, time-series, sediment trap measurements demonstrated that fission products arriving at the sea surface as a single pulse were rapidly transported to 200 m in approximately 7 d (Table 5.5). Examination of samples containing the bulk of the radioactivity indicated they were composed mainly of copepod fecal pellets, e.g. 70 per cent of the sample by mass. Fresh pellets collected from copepod grazing above the traps were found to contain similar radionuclide concentrations and ratios as those in the trap material, confirming that such pellets were responsible for removing the radioactivity to depth (Table 5.6). Furthermore, isotopic ratios in pellets, water and copepods indicated that particle-reactive radionuclides like ¹⁴⁴Ce, ¹⁴¹Ce, ¹⁰⁶Ru, ¹⁰³Ru were scavenged to a far greater extent by sinking faecal pellets than the caesium nuclides (Table 5.6). This and other Chernobyl studies in the Black Sea have clearly demonstrated the rapidity (18

200 m d^-1) by which sinking biogenic particles can transfer surface-introduced particle-reactive radionuclides to depth (Buesseler et al., 1990). Whereas these studies underscore the importance of micro-crustacean faecal pellets in radionuclide flux, pellet deposition from salps, a gelatinous macroplankton species common in the Mediterranean, would also be an effective mechanism since Mediterranean salps produce pellets highly enriched in natural radionuclides and at rates greater than those of planktonic crustaceans (Krishnaswami et al., 1985).

Table 5.6 Selected radionuclide activity ratios of Chernobyl-derived radionuclides in different samples from the north-western Mediterranean (from Fowler et al., 1987, 1990b)


Sample	Collection	Activity ratio	¹⁰³Ru/¹⁴¹Ce	¹⁰³Ru/¹³⁷Cs	²⁴¹Am/^239+240Pu
	date (1986)	¹³⁷Cs/¹⁴¹Cs

Air	3 May	58	110	1.9	0.13
Unfiltered	7 May	11	8.0	0.73
seawater
Zooplankton	6 May	1.7	14	8.2	0.22
Fresh faecal	6 May	7.0	18	2.5	0.09
pellets
Trapped	11 May	0.27	1.8	6.8	0.37
particles
at 200 m

The effectiveness of this biological packaging mechanism in removing surface-introduced radionuclides and transporting them to depth should not be underestimated. For example, based on total wet and dry fallout deposition data obtained near the Mediterranean trap site, it was calculated that 50 and 75 per cent of the total Ce and Pu deposited on the sea surface had fluxed through 200 m in the form of large particles approximately 1 month after the accident. In contrast to these particle-reactive radionuclides, only a small fraction (0.2 per cent) of the ¹³⁷Csdeposited in the region was carried downward by the same particles. What is not evident is to what depth the biogenic particles transported these radionuclides. There is some evidence from deep sediment cores that even up to 1 year after the accident, Chernobyl radionuclides had not reached the sediments in certain areas of the north-western Mediterranean (Thommeret and Huynh Ngoc, 1990). This may result from a gradual release of radionuclides from particles through desorption or decomposition processes as the particle sinks into deeper waters, as has been demonstrated for Chernobyl radionucldes in the Black Sea.

Benthic infauna have evolved a number of feeding strategies to exploit the organic content of the suspended particles over the seabed. The major nutrient sources are abundant in the top few centimetres, whereas deeper sediment represents a relatively poorer food source. The benthos can influence the behaviour of radionuclides in a number of ways. Filter-feeding forms increase the fluxes of particulate bound radionuclides across the sediment-water interface. Other surface-feeding forms may deposit a proportion of this material at depth in the form of faecal material in a manner of a conveyer belt. Conversely, deeper burrowing forms evacuate low-activity sediment onto the surface. The presence of burrow systems, opening at the sediment

water interface, results in increased ventilation of the sediment by irrigation. This in turn influences the chemical cycling of redox-sensitive elements such as Fe, Mn and Pu (see Section 5.4.3). Bioturbation has been shown to be the dominant mechanism by which Sellafield-derived radionuclides are transported below the sediment-water interface (Kershaw et al., 1992; Section 5.4.4.1). The representation of bioturbation, a complex interaction of many different processes, by a single coefficient has been criticized by many authors and other models have been suggested (Berner, 1980; Jumars et al., 1981; Aller, 1983; Wheatcroft et al., 1990). However, these models lack general applicability and the commonest approach has remained that of modelling bioturbation as a diffusive process. The diffusive model assumes that processes affecting bioturbation are temporally and spatially continuous. They are clearly not, as inputs to the sediment are often episodic, especially anthropogenic inputs, or vary over a seasonal cycle and often occur in limited areas. Further, the infauna are not evenly distributed either in the horizontal or vertical planes, patchy distributions being common, and abundance and biomass vary with seasonal and other temporal cycles. The diffusive model also assumes that sediment particles are mixed randomly and isotropically over local distances, that is over dimensions similar to that of the grain sizes themselves. However, again, these are not good assumptions as most mixing is over non-local distances and usually most strongly in the vertical direction.

The extent to which sediments retain contaminants depends, in part, on the nature of the contaminant

particle interaction and on the effect of diagenetic processes. Diagenetic reactions encompass all of the chemical and physical reactions which occur in sediments that bring about significant changes in the chemistry of both the solid (sediment) and liquid (interstitial water) phases. Assessment of the relative mobility of different radionuclides has often been made by comparing only the solid phase distribution of the elements in sediment profiles. Few studies have, however, used the most sensitive method for examining radionuclide mobility directly, that is, by the study of the composition of interstitial waters. This is no doubt due to the significant analytical/technical problems that are inherent in such studies. It is necessary to ensure that measurements are representative of chemical conditions within the seabed (temperature, pressure and oxidation changes) and that sufficient sample is collected to determine the concentrations, chemical associations (e.g. colloids) and oxidation states.

The diagenetic changes which occur in shelf sediments are driven by the bacterially mediated oxidation of the organic matter which is deposited in the sediments. The major chemical change which occurs in most sediments is the change from oxic to anoxic conditions. The significant drop in the redox potential of the system may, of course, have a direct effect on radionuclides which can occur in a number of different valence states, for example plutonium.

The major diagenetic processes that occur in sediments are the reduction of oxygen, manganese, nitrate, iron and sulphate and finally fermentation. These processes tend to occur in the order of decreasing free energy yield and this results in zones of activity arranged horizontally with depth in the sediments, with oxygen as the first electron acceptor in the sequence.

As a result of diagenesis, significant changes occur in the chemistry of particles, especially their surfaces. Radionuclides adsorb to particle surfaces. Some, like caesium, may also be absorbed into the mineral lattice of sediment particles, readily exchanging with potassium. The nature of the interaction will of course determine the subsequent behaviour of any particular radionuclide. For instance, a radionuclide adsorbed to the iron and manganese surface coatings of sediment grains will be subject to mobilization, as the nature of the surface changes, on transition from an oxic to an anoxic regime. The metal oxides will be solubilized and the radionuclide released to react with different mineral substrates. It will be the kinetics of the reaction in competition with the rate of transport which will dertermine the extent to which any radionuclide will be relocated or lost from the sediment. Diagenetic processes also bring about significant changes in the composition of the interstitial water, with marked increases in the concentrations of total carbon dioxide, phosphate, ammonium, ferrous ion, manganese and dissolved organic matter. The speciation of dissolved radionuclides may alter in such a way as to increase mobility. For instance, interaction of radionuclides with dissolved organic matter may enhance solubility. It will be in fine-grained sediments that diagenetic processes are best expressed, and these are also the sediments which contain the highest concentrations of radionuclides.

Two shelf areas have received detailed study of both sediments and the contained interstitial water: (a) the north-west Atlantic shelf (Sholkovitz and Mann, 1984); and (b) the north-east Irish Sea (Malcolm et al., 1990). The north-east Irish Sea study site has a distinct analytical advantage in that the concentrations of artificial radionuclides in the sediments are relatively high due to the proximity of a major source: the discharges from the Sellafield reprocessing plant. In such studies, however, relatively large volumes of interstitial water have to be extracted under controlled conditions, which makes the sample collection operation very demanding technically. The higher concentrations present in the north-east Irish Sea enabled the analytical separation of the higher and lower valence states of plutonium adding a useful dimension to the study in terms of particle reactivity

the reduced form of plutonium adsorbs to particles more readily than the oxidized form (Section 5.4.1).

Figure 5.16 Solid-phase profiles of (a) ^239,240Pu and (b) ²⁴¹Am, (kBq kg ^-1),from a sediment core close to the Sellafield discharge pipe. The concentrations profiles show a close correlation with the discharge histories (TBq) and do not appear to be influenced by iron (c) or manganese (d) cycling. From Malcolm et al., 1990.

The data shown in Figure 5.16 derive from a site near to the Sellafield discharge point in the Irish Sea. The two artificial radionuclides ^239,240Pu and ²⁴¹Am show major subsurface peaks of concentration which are not related to the solid phase distributions of iron, manganese or organic carbon but may reflect the input due to the discharge from Sellafield (the discharge history is shown on Figures 5.16a and 5.16b). This would imply that diagenetic changes in these sediments are not able to remobilize significant amounts of plutonium or americium in timescales of several years. Examination of the interstitial water data again shows little evidence of any correlation between the artificial radionuclide concentrations and the major indicators of diagenetic processes in these sediments (Figures 5.16c and 5.16d). Only the interstitial water profiles of iron and manganese are shown but nitrate, phosphate and dissolved organic carbon were also measured. Association of the transuranic elements with dissolved organic matter has been suggested as an important factor in determining the behaviour of these elements in freshwaters. However, there is only limited evidence for this in marine interstitial waters. Unpublished studies in the north-east Irish Sea sediments using ultrafiltration techniques suggest that some plutonium and some americium is associated with material of molecular weight greater than 10 000 Daltons (Malcolm, unpublished data; Section 5.4.1). The rate of diffusion of high molecular weight complexes is likely to be very slow, so redistribution of the plutonium and americium is likely to be of limited importance.

In the surface sediments at this site between 27 and 69 per cent of the plutonium is present in the oxidized form. This contrasts with the underlying sediments where all of the plutonium is in the reduced form. The apparent distribution coefficient is within the range expected for reduced plutonium in non-carbonate marine sediments, and indicates the very strong particle association of this radionuclide. Close inspection of the distribution coefficient profile suggests that mobilization may be taking place but the signal is not sufficient to suggest major relocation of plutonium or americium within the sediments.

The studies from the north-west Atlantic shelf also suggest that diagenesis does not have a significant effect on the mobility of plutonium; interstitial water plutonium was found to be in equilibrium with the solid phase plutonium concentrations. The apparent distribution coefficient varied with the carbonate content of the solid phase.

The behaviour of caesium in marine sediments has not been well studied in interstitial waters but it has generally been assumed that caesium is more mobile than elements like plutonium. There was some evidence for this in studies reporting the solid phase concentrations of the elements, suggesting that the greater penetration of caesium into sediments was due to its greater mobility. The part that diagenesis plays in this process has not been defined. One study in Buzzards Bay (Sholkovitz and Mann, 1984) suggests that caesium in marine sediments behaves in the same way as caesium in freshwater sediments and that ion exchange reactions are important. Notable in this connection is the increasing mobilization of caesium with increasing concentrations of ammonium, a product of diagenetic reactions in anoxic sediments. While the Buzzards Bay data strongly suggest mobilization, it was not possible to quantify the process.

Natural series radionuclides are used extensively to study the rates of particle transport in marine sediments. An assumption underlying many of these studies is that the nuclides being used are strongly particle associated and not subject to diagenetic remobilization. The assumption appears to be justified in the use of thorium isotopes, though there has been little study of thorium mobility in marine sediments. Interstitial water data from Buzzards Bay (Cochran et al., 1986) for thorium shows depth-invariant low concentrations of ²³²Th and ²³⁸Th in some nearshore sediments, whereas ²³⁴Th did show remobilization associated with the redox-driven dissolution of iron and manganese. The implications for the use of ²³⁴Th as a tracer have not yet been fully considered.

In contrast to thorium, the behaviour of uranium in sediments is more widely influenced by diagenetic processes. Uranium, like plutonium, is present in a number of redox states in the environment. Reduction of U(VI) and U(IV) occurs together with the reduction of iron and manganese. This leads to the removal of uranium from interstitial waters at the same depth as iron and manganese are dissolved. At greater depth in the sediment, uranium is released from the sediments while the organic matter with which it is associated is oxidized and the interstitial waters alkalinity increases accordingly.

The use of ²¹⁰Pb for estimating sedimentation rates and sediment mixing also relies on the assumption that it is predominantly particle-associated and non-mobile. This is likely to be true in sediments which are anoxic below a very narrow surface oxidized layer. This includes the sediments in California borderland basins where the use of ²¹⁰Pb was first established. However, in the suboxic layer of sediments, where there is mobilization of Fe and Mn, Pb may be mobilized. The implications of this for the use of Pb in measuring various rates has not been sufficiently studied.

More studies of the behaviour of radionuclides in interstitial waters and sediments are required, as they will provide increasingly important tools for the study of sediment biogeochemical processes.

In general, the effect of discharging effluent from a point source into a high-energy coastal environment will be to dilute and disperse. However, locally enhanced concentrations can occur and the degree to which these are maintained, increase or even decrease will depend on the discharge history and the timescales of the various geophysical, chemical and biological processes responsible for nuclide migration.

A considerable volume of work on the mechanisms controlling the behaviour and transfer of radionuclides in coastal regions of the open sea has been generated by the discharges from the Sellafield nuclear fuel reprocessing plant (Kershaw et al., 1992). Some of these studies have been referred to in an earlier section (5.4.1).

The history of Sellafield releases is provided in the General Introduction, and the resultant distribution of Cs in seawater with time is described in Section 5.4.1. Although plutonium from Sellafield is found throughout the Irish Sea in both reduced and oxidized states, research based on selective precipitation has shown that more than 90 per cent of the ^239,240Pu in filtered surface waters exists in the oxidized forms and suggests that plutonium may be transported throughout the Irish Sea as a soluble species by residual currents. It should not, however, be inferred that this is the dominant mechanism of transport. In inshore waters the situation is greatly modified. Whilst the total concentration is similar to that observed in corresponding offshore samples, the proportion of particle-associated to soluble plutonium is much higher. This is fully consistent with the fact that plutonium becomes rapidly reduced in the presence of scavenging particulates. The reduced state is considered to be Pu(IV) and is mainly, though not exclusively, associated with the particulate phase. In inshore waters, tides and currents cause large quantities of sediment to undergo a continuous cycle of resuspension and deposition, with the result that scavenging is enhanced.

The speciation of plutonium and americium in seawater and the mechanism by which they can, in certain circumstances, form colloids, is fundamental to an understanding of how transuranic nuclides become dispersed in the marine environment and warrants further study.

The bulk of Sellafield-derived transuranic radionuclides now reside, in association with deposited sediments, relatively close to the outfall and, in particular, in the mud patch extending south from St Bees Head as well as in the local estuaries. A smaller fraction, composed of plutonium adsorbed onto fines in suspension and plutonium in a soluble or colloidal form, is dispersed throughout the Irish Sea with an estimated 3

17 per cent of each year's discharge being removed through the North Channel (Pentreath et al., 1986). Radionuclide measurements were made at three selected sites along the eastern shore of the Irish Sea in the period 1968

87. The variation in the concentration of ¹³⁷Cs in Fucus vesiculosus samples closely reflects that of the ¹³⁷Cs discharges from Sellafield, which were at their peak in the mid-1970s. There is also evidence from these and other data which suggests that the decline in the concentrations of ¹³⁷Cs and other fission and activation products in environmental materials is most rapid close to the Sellafield outfall. Examples of the temporal evolution in anthropogenic radionuclide concentrations in algal indicators can be cited to confirm that a general decline in these concentrations is taking place in the biological compartment throughout the Irish Sea.

Substantial reductions in the concentrations of a number of radiologically important radionuclides in fish and shell-fish sampled in the eastern region of the Irish Sea have been observed since the mid-1970s. Concentrations in both fish and shell-fish diminish with distance from Sellafield, the rate of reduction being least for nuclides which are relatively mobile in seawater. A similar pattern is observed for seawater and algae. Variations in concentrations between fish species sampled in a given area are comparatively small and can be explained in terms of residence time in the area, as well as feeding habits. On the other hand, substantial variations are observed between shell-fish species. For example, molluscs tend to concentrate the transuranics to a considerably greater extent than crustaceans, which in turn accumulate them more than fish (IAEA, 1985b).

The principal feature observed in concentrations of radionuclides in surface sediments sampled close to the outfall is a general reflection of the reduction in discharges since 1978. As expected, the rate of reduction is least for nuclides which both exhibit strong sediment-seeking properties and have long half-lives (McCartney et al., 1992). The concentrations of ²³⁸Pu, ^239,240Pu and ²⁴¹Am in Newbiggin surface sediments sampled in the period 1977

87 confirm that these elements are subject to dispersion, albeit slow. A similar, though weaker trend, was evident in surface sediments sampled at Heysham. Further afield, the reduction in transuranic concentration is not always evident; for example, at Garlieston in SW Scotland, the concentrations of Pu and Am in silt, if anything, are still increasing, whereas those of fission products such as ⁹⁵Zr

⁹⁵Nb, ¹⁰⁶Ru, and ¹³⁷Csare clearly falling.

Prior to the Chernobyl accident in 1986, the radionuclides¹³⁴Cs and ¹³⁷Cswere regarded as near ideal tracers for continental shelf water circulation because: (a) the half-life of the latter (30 years) is long compared to typical transport timescales on the shelf; (b) the half-life of the former (2 years) is distinctly shorter than that of ¹³⁷Cs, which means that the ratio of these two nuclides can be used to provide an alternative determination of transport times; (c) being soluble in seawater, they are not easily scavenged from the water column by particulate matter in suspension; (d) both had been discharged from Sellafield in relatively large, known quantities for many years; and (e) neither occurs naturally, though a low background had resulted from earlier nuclear weapons tests in the atmosphere. Therefore, caesium isotopes have been widely used in marine transport studies in the north-east Atlantic, and in particular in the Irish Sea and the North Sea. For example, the ¹³⁴Cs/¹³⁷Cs ratio has been used to estimate the turnover time for water entering the Irish Sea and to trace the path of Sellafield effluent to the coast of Greenland.

The activity ratio ²³⁸Pu/^239,240Pu in releases from nuclear fuel reprocessing plants has varied systematically and differed from that of fresh fallout and weapons-grade plutonium. It has been used to identify source terms and transfer factors. Presently, this ratio lies in the range 0.22

0.25 in marine environmental materials, including surface sediments, sampled close to Sellafield (Hunt, 1977

92) and is only slightly higher than the ratio measured in similar materials from the east coast of Ireland. Analyses of the ²³⁸Pu/^239,240Puratio in sediment cores and, in particular, in the linings of Maxmuelleria lankesteri burrows, serve to emphasize the role of bioturbation in the incorporation of the transuranics within the deeper sediments (Kershaw et al., 1984).

Attempts have been made to estimate transport rates and availability times from isotope ratios in environmental materials (e.g. Hunt, 1985; Schell et al., 1989; Mackenzie and Scott, 1982) with varying degrees of success. It does appear that particle-reactive radionuclides are being transported away from the vicinity of the Sellafield discharge point in both particulate and dissolved phases. Estimates of actual rates of removal are complicated by the time-varying isotope ratios in the discharge, variations in the actual quantities of each radionuclide discharged, differential half-lives, grow-in (e. g. ²⁴¹Pu

²⁴¹Am) and the mixing of materials contaminated by recent discharges with those contaminated previously.

In most cases the present distribution of contaminated sediments and biota must be considered as representing a temporary 'snap-shot', resulting from complex interrelated processes varying both spatially and temporally. The area of muddy sediments lying in deep-water between the Isle of Man and Ireland may be considered as a probable long-term sink, but there is little evidence to suggest that sediment in any other region of the Irish Sea will act in this way (Kershaw et al., 1988).

Studies, conducted on soil samples, airborne deposits, muslin screens, mosses, lake sediments and sheep faeces taken in West Cumbria, UK, since the late 1970s have shown that a small fraction of the plutonium and americium discharged into the Irish Sea from Sellafield has been transferred into the atmosphere and returned to land. The considerable volume of work undertaken has recently been reviewed (McKay and Pattenden, 1990). Bubble scavenging in the water column, together with droplet ejection from bubble bursting at the surface, may be involved in the transfer mechanism (Walker et al., 1986). A similar phenomenon has been reported for heavy metals.

The soil samples were taken along transects running inland from the Cumbrian coast to about 20 km and were analysed for ¹³⁷Cs and Pu. The transect measurements show enhanced values of ^239,240Pu near the coast, between 3 and 11 times the weapon fallout baseline. These relax to the baseline value at about 5 km and further inland. Preliminary estimates have been made which suggest that 40

80 GBq of excess Pu had been deposited in a coastal strip approximately 5 km wide and 40 km long by about the year 1980. In contrast, the caesium values along the transects can be accounted for on the basis of levels in bulk seawater with little or no evidence of marine aerosol enrichment. Sea-to-land transfer has also been reported near the Cap de la Hague facility on the French Channel coast and the Dounreay reprocessing plant on the northern Scottish coast. In the latter case transport inland is rather limited and the postulated mechanism is wind-blown spume (stable foam). Fine-sediment is an important component in the sea-to-land transfer process. For this reason the relative importance of this pathway will increase, as discharges from Sellafield continue to be reduced, as a result of the legacy of sediments contaminated in the 1970s and 1980s.

The degree to which radionuclides released into coastal waters are transported across the continental shelf is dependent upon the chemical properties of the radionuclides and the nature of the physical environment in which dispersion is taking place. A good example is given by the dispersion of radionuclides from the Hanford reactor on the Pacific coast.

The operation of Pu-producing reactors at Hanford, Washington, beginning in 1944, introduced several artificial radionuclides to the Columbia River and subsequently to the adjacent continental shelf and slope (Figure 5.17). An estimated 300 000 Ci y^-1of artificial radioactivity was discharged to the river by the mid-1960s, but decreasing reactor operations and more stringent discharge controls essentially eliminated radioactive discharges to the river by 1981. Radioactive decay of shortand moderate-lived nuclides, sediment export and river flow combined to reduce artificial activities in river sediments to levels which, by 1984, could no longer easily be measured. Studies since the mid-1970s have focused on chemicals which are of concern due to their persistence and relatively toxic nature, such as the isotopes ²⁴¹Am and ¹³⁷Cs and the natural isotopes ²¹⁰Po and ²¹⁰Pb.

Figure 5.17 General study area off the Washington and Oregon coasts. NB = Neah Bay/Cape Flattery, Wa. ONP = Olympic National Park, Wa. JdF = Juan de Fuca submarine canyon. Puget sound is the inland body of water east of the Olympic National Park. The Hanford Nuclear Reservation is located on the Washington side of the Columbia River and just east to this chart.

The artificial radionuclides have served as instructive tracers of Columbia River-derived material in the adjacent ocean. Seasonal changes in position of the Columbia River plume were clearly evident in early studies of ⁶⁵Zn activity. ⁶⁵Zn activity in offshore sediments (Cutshall et al., 1973) showed that the predominant transport of Columbia river-derived solids was along a midshelf silt deposit trending north-north-west from the river mouth towards Quinault submarine canyon (Figure 5.17). Integrated ⁶⁵Zn activities were at least two-thirds of the total ⁶⁵Zn discharged by the river, suggesting that fine-grained riverine suspended sediment was transported offshore and not trapped in the estuary. The presence of the short-lived isotopes ¹⁴¹Ce/¹⁴⁴Ce and ⁹⁵Zr/⁹⁵Nb in sediments from Astoria Canyon was one of the first clear indications that radionuclides entering the ocean at the surface could be transported to depths of over 1 km within a few days (Osterberg et al., 1963).

Pu (and ²⁴¹Am) activities and inventories in Washington shelf and slope sediments are 5

15 times those in sediments of the north-west Atlantic and Gulf of Mexico (Carpenter et al., 1987). The Washington

Oregon coasts are in the latitude band of high atmospheric fallout of artificial nuclides, but even inventories of 2.2 ± 0.5 mCi km^-2 estimated from terrestrial soil cores cannot account for the exceptionally high Pu activities in Washington shelf and slope sediments. Shapes of Pu activity profiles in both shelf and slope sediments show that Pu inputs to the sediments are continuing to the present

the input cannot be modelled as a pulse in the past. The continuing Pu input to the sediment is supported by a time series of Pu activities in mussels.

Inventories of Pu isotopes in bottom sediments from Hanford to the Columbia River estuary were estimated based on activities measured in 50 cores (Beasley and Jennings, 1984). ²⁴⁰Pu/²⁴²Pu and ²⁴⁰Pu/²³⁹Pu isotope data established that only about 3.5 per cent of the Pu in fine-grained sediment in the river estuary came from Hanford reactors. Thus high Pu activities in offshore sediments are not due primarily to additional Pu input from reactors at the Hanford reservation. Monthly samples of water collected near the mouth of the Columbia River between July 1978 and July 1979 were analysed for Pu and Am to estimate the quantities of these nuclides exported to the adjacent coastal zone during this year (Beasley et al., 1981). From the depositional history of Pu activities recorded in a fine-grained, non-bioturbated sediment core from Youngs Bay (within the estuary), 4

8 Ci of ^239,240Pu and 1

2 Ci of ²⁴¹Am were estimated to have been discharged by the river to the adjacent coastal ocean since the late 1950s. Such riverine discharge of Pu and Am is small compared to total Pu sedimentary inventory of 35 Ci estimated for the Washington shelf alone.

The importance of non-riverine Pu source is seen in that, while Al contents remain relatively unchanged in moving from the estuary to and along the shelf, Pu activities of the fine-grained sediments increase from about 20 dpm kg^-1 to 130 dpm kg^-1. Further evidence for major additional non-riverine Pu in shelf sediments is also seen by comparing ²⁴⁰Pu^/239Puisotope ratio signatures of offshore sediments with ratios in sediments accumulating in the estuary and further up rivers (Beasley et al., 1982). No measurements of dissolved Pu concentration profiles have been reported for ocean waters off Oregon or Washington, but it seems reasonable that such profiles will resemble those found around large parts of the North Pacific in having a pronounced subsurface maximum around 400 m deep (Figure 5.18) due to dissolved Pu removal from surface waters by association with largely biogenic particles and Pu release to deeper waters after sinking and decay of most of the organic particles. This Pu remains in solution at mid-depth, due to lack of particles to scavenge it. Even when the equilibrium partition coefficient (K_d) of an element between particulate and dissolved phases is as high as 10⁷, over 50 per cent of the element should still be in dissolved form when particle concentrations are < 0.1 mg 1^-1. Significant scavenging of particle-reactive, dissolved chemicals occurs when open ocean water enters the Washington coastal zone and encounters 10

100 times higher particulate concentrations due to the Columbia River discharge and the biologically productive nearshore environment. Frequent resuspension of bottom sediments by storms, bottom currents and benthic organisms also increases particle concentrations and enhances scavenging in near-bottom waters. Boundary scavenging has only been confirmed for ²¹⁰Pb, ²⁴¹Am and Pu off the Washington coast but should also be an effective mechanism for removal of other particle-reactive elements such as Be.

Figure 5.18 Pu concentration profiles (dpm 100 kg^-1 seawater) in an east

west section the Pacific at 30

35º N in 1973. From Bowen et al., 1980

Waters from 200

300 m depth are now known to be upwelled onto Washington and Oregon coastal zones (Hickey, 1989). Conservative estimates based on water volumes suggest that the advective supply of these isotopes considerably exceeds their observed sedimentary inventories and fluxes. Sedimentary inventories of these chemicals on the Washington coast are therefore limited by boundary-layer scavenging reactions rather than by supply of dissolved chemicals. Particle concentrations in coastal water and published sediment

water Pu distribution coefficients combine to suggest that only 18

20 per cent of Pu flowing past should be removed to sediments, and this amount is indeed adequate to supply the observed sediment inventories (Beasley et al., 1982).

Pu sedimentary phase associations have not been determined for Washington slope sediments, but in adjacent shelf sediments selective chemical leaching tests have shown that both Pu and ²¹⁰Pb are primarily associated with hydrous Fe

Mn oxide phases rather than with clays or organic phases. Calculations of maximum Pu flux due to biological production on the Washington shelf based on upper limits for total organic matter production and Pu concentration factor in the organic matter, and assuming all the biological matter reached the sediment and remained there, show it is unlikely that biological processes have substantially mediated the scavenging of Pu to Washington coast sediments. The key role of Pu scavenging by reductant-soluble hydrous Fe-Mn oxide phases has also been shown for soils and Lake Michigan sediments.

Because of boundary scavenging, sedimentary inventories and fluxes of excess ²¹⁰Pb and Pu (Beasley et al., 1982; Carpenter et al., 1987) off Washington are 5 to 10 times greater than off the US east coast. This enhanced boundary scavenging off the Washington coast is attributed to a combination of several reasons (Carpenter and Peterson, 1989): (a) concentrations of many dissolved chemicals are greater in mid-depth North Pacific than North Atlantic waters; (b) more upwellings, advection of deep water, and production of biogenic particles occur along the Washington coast than off other coasts; (c) the Columbia River discharges relatively large amounts of fine-grained, reactive detrital particles which include high concentrations of expandable smectite clays which have high ion exchange capacity; (d) lower O₂ concentrations and higher organic C fluxes to Washington coasts sediments may lead to greater dissolved Mn fluxes from the sediment in the overlaying water column, enhancing trace element scavenging by hydrous Mn oxides near the sea floor.

Excess ²¹⁰Pb/^239,240Puinventory ratios in eight representative cores from the Washington shelf average 100 ± 19, even though absolute values of both inventories, sediment texture and accumulation rates vary much more. This surprisingly narrow ratio suggests mechanisms supplying and removing the natural isotope ²¹⁰Pb and the artificial isotopes Pu are similar.

Selective chemical leachings of shelf sediments have shown that both Pu and²¹⁰Pb are primarily associated with hydrous Mn and Fe oxides. Pu activities and inventories in slope sediments also correlate strongly with excess ²¹⁰Pb, with inventory ratios increasing from 100 in shelf sediments to 300

380 in deeper slope sediments. These reasonably constant ratios, for given water depths and similar phase association, permit estimation within a factor of two of the total Pu inventories and prediction of sites of accumulation of Pu from data on the more easily measured ²¹⁰Pb. Geographical distributions of ²¹⁰Phinventories thus lead to the prediction that Pu inventories should be 2

4 times greater in sediments from the submarine canyons off Washington than in sediments from nearby open slopes at comparable water depths.

Average ²⁴¹Am/^239,240Pu inventory ratios are twice as large in slope sediments as in shelf deposits (Carpenter et al., 1987). This increase is attributed primarily to a greater particle affinity for dissolved Am than Pu (and hence Am deposition). The conclusion of preferential scavenging and downward transport of Am over Pu by particles is supported by data from the Mediterranean, off central and southern California, the abyssal Atlantic and Pacific, and by measurements of solid

water partition coefficients.

In contrast to Pu and ²⁴¹Am, ¹³⁷Csinventories in both Washington shelf and slope sediments result primarily from residual Cs retention on particles discharged by the Columbia River. Rough estimates of riverine supply of ¹³⁷Cs-labelled particles show the river discharge is more than enough to maintain offshore sedimentary inventories, even allowing for desorption and decay of some of the riverine Cs. ¹³⁷Csactivities in surface slope sediments average slightly less than activities in adjacent Washington shelf sediments. The finer-grained, more clay-rich deposits of the slope do not contain more ¹³⁷Csthan mid-shelf silt deposits. This difference in ¹³⁷Csactivity between shelf and slope sediments is opposite to that of ²¹⁰Pb, Pu isotopes and for ²⁴¹Am, which are all higher in slope than shelf sediments. ^239,240Pu/¹³⁷Cs activity and inventory ratios in Washington slope sediments are 2

3 times ratios in shelf sediments. Higher ratios are due to a combination of preferential scavenging of dissolved Pu (predicted by the approximately 1000-fold greater sediment

water partition coefficient for Pu than for Cs) and to greater ¹³⁷Csloss by decay and desorption. Partial desorption of ¹³⁷Csduring offshore transport of riverine-derived particulates is supported by laboratory tests of the leachability of ¹³⁷Csfrom Columbia river sediments by artificial seawater.

Marginal basins and seas with limited connection to the ocean often exhibit characteristics which are atypical of more open coastal and shelf regions. Two such semi-enclosed seas are considered here: the Mediterranean and Baltic Seas. Work carried out on the Black Sea is referred to elsewhere within Section 5.4. A review of Black Sea oceanography has recently been published (Izdar and Murray, 1991).

The Mediterranean is a small-scale ocean basin with limited exchange with Atlantic water through the Straits of Gibraltar. It received direct and indirect input of Chernobyl radionuclides (Section 5.4.2) and receives input via the Rhône River of artificial radionuclides from the Marcoule reprocessing plant. In the Rhône estuary water column, the sharp decrease of ¹³⁷Cs from the surface to bottom waters shows an inverse correlation with the salinity profiles and illustrates the dilution of the Rhône River freshwater discharges in the marine surface layer. The predominant fraction of the Rhône sediments (about 2 x 10⁶ t y^-1) accumulates up to 60 km westward of the Rhône mouth. Thus, ¹³⁷Csassociated with fine-grained terrigenous material released by the Rhône is deposited permanently on the continental shelf. However, more recent studies exhibit a more complex behaviour of ¹³⁷Cs, specially at the air-sea interface.

In 1987, a comparison of the 50 cm depth seawater concentration with the surface microlayer (top 150 µm), along a cross-section through the Rhône dilution area, revealed a ¹³⁷Cs enrichment factor of 4 for the fraction less than 0.4 µm by comparison with the fraction greater than 0.4 µm.

Prior to the Chernobyl accident, elevated. ¹³⁷Csconcentrations were restricted to the coastal region influenced by the Rhône outflow. The enrichment factor of the surface microlayer is consistent with data proposed by Eakins and Lally (1984) for marine aerosols. Temporal trends were related to variations in river flow, and thus to rainfall, as well as to the dates of discharge of effluents with a low level of radioactivity.

In April and May 1986, the fallout from the Chernobyl accident added significantly to the Mediterranean marine stock of ¹³⁷Cs. Radionuclides such as ¹³⁴Cs, ¹⁰³Ru, ¹⁰⁶Ru, ^110mAg, ¹²⁵Sb, ¹⁴¹Ce and ¹⁴⁴Ce were also detected, but concentrations decreased very rapidly with time, approaching the minimum detectable limit in seawater, sediments and marine organisms. In contrast, Chernobyl-derived ¹³⁷Cswas still detectable, especially in sediments, two years after the date of the accident. These temporary inputs of radionuclides, which may be considered as tracers, allowed the study of various processes in the marine environment: tracing of telluric inflows in both dissolved and particulate form into seawater masses, followed by the penetration and distribution of radionuclides in the sedimentary layer, and the response of marine organisms.

Direct deposition of caesium from the atmosphere onto the seawater surface provided a source for tracing the vertical transfer of radionuclides through the water column (see Section 5.4.2.1). In the water of the Gulf of Lyons this penetration had reached, in December 1986, only the first 200 m. Twenty months after the date of the accident, a significant increase in caesium content seemed to affect the water masses as far as 400 m. The movements of water masses in this area are characterized by the occurrence of upwellings in winter, combined with the occurrence of winds such as the Mistral or Tramontane. These upwellings homogenize the hydrological characteristics of the whole water column, thus causing the vertical variations in caesium contents to disappear. However, the Liguro

Provençal basin drift along the continental shelf in this area seems to play a leading role in supplying the surface waters of the Gulf of Lyons with radionuclides originating from the Ligurian Sea. Indirect telluric inflows from Mediterranean rivers, after the leaching of continental deposits by rainwater, can also be considered as secondary coastal sources which are delayed in time (Fukai et al., 1983). Following the increase in rainfall in December 1986, the ¹³⁷Cs distribution in surface waters of the Gulf of Lyons allowed the dispersion of river waters to be defined along the north-east to south-west line linking the Rhône prodelta to Cape Bear.

The coastal sediments also responded quickly to the injection of new radionuclides. Thus, in September 1986, three months after the Chernobyl accident, the surface layer of sediments sampled at river mouths indicated the presence of ¹³⁷Cs at higher concentrations than before, and the appearance of ¹³⁴Cs, ¹⁰⁶Ru, ¹¹⁴Ce and ^110mAg. Caesium isotopes originating from the Chernobyl fallout were detected up to 10 cm below the sediment surface showing high apparent diffusion coefficients.

This apparent mobility may limit the use of caesium as an accurate temporal marker in sediments. But caesium is a particularly useful tracer of particulate telluric inflows and sediment transport. Consequently, the measurements made on suspended matter introduced by the Rhône and collected in sediment traps show no significant difference from those made on sediments sampled at the same point. This observation reflects the predominance of telluric inflows compared to direct fallout on the surface of the water. However, the map of the Rhône prodelta down to depths exceeding 2000 mdoes not reveal the passage of continental shelf particles labelled with ¹³⁷Csdown to the bathyal region. Sediment transport must certainly have happened, in relation to the presence of a deep nepheloid layer along submarine canyons.

Following the Chernobyl accident, certain marine organisms showed a rapid response to the injection of the new radionuclides. Subsequently, the concentrations of these various radionuclides decreased substantially, reflecting their rapid removal from the water column. Mytilus sp. radionuclide concentrations in the wake of the Chernobyl reactor accident confirm the significance of this genus as a bioindicator for the monitoring of the Mediterranean coastal marine contamination by radionuclides (Figure 5.19, Calmet et al., 1988). Figure 5.19 shows a decreasing east

west gradient emerging in May 1986 of radioactivity concentrations in Mytilus sp. samples. This gradient reflects the heterogeneity of Chernobyl fallout deposition in the north-western basin surface water, primarily affecting the Tyrrhenian Sea and the Gulf of Genoa. The marine radionuclide inventory thus created within the surface water was subsequently dispersed throughout the north-western basin by means of the east

westbound Liguro

Provençal current.

The Baltic Sea is the largest brackish water area in the world. As it is a semi-enclosed sea with only a limited water exchange with the world ocean via the North Sea through the Danish straits, Kattegat and Skagerrak, persistent contaminants will remain in this area for a long time. A characteristic feature of the Baltic Sea is its vertical stratification into two layers, with the surface exhibiting a low-salinity water body. The permanent halocline, at a depth of between 40 and 60 m, prevents the vertical exchange and mixing of water due to the different densities of the water types. Surface water supplied by river runoff moves into the Skagerrak with 8 to 9 PSU (practical salinity unit), while water below the halocline reflects the inflow from the North Sea with 17 to 19 PSU. The mean residence time of water in the Baltic is estimated to be 20 to 30 years.

Artificial radioactivity in the Baltic Sea mainly originates from three sources: (a) fallout from the nuclear weapons tests during the 1960s; (b) inflow of contaminated water from the North Sea; and, (c) fallout from the nuclear accident at Chernobyl. Fallout from weapons testing is a more common source of artificial radioactivity in the Baltic Sea than in other comparable shallow sea areas.

Owing to increasing discharges from reprocessing plants at Sellafield (Windscale) and La Hague during the mid-1970s, the inflow of water from the North Sea into the Baltic Sea resulted in a time-delayed increase of the activity concentration in the Baltic (IAEA, 1986) with a strong correlation between ¹³⁷Csactivity and salinity. A strong correlation between the annual ¹³⁷Cs discharge from Sellafield and the ¹³⁷Csactivity concentration four years later in the Danish straits could also be established (IAEA, 1986). Similar correlations cannot be ascertained for ⁹⁰Sr, proving that the more relevant pathway of ⁹⁰Sr is runoff from river systems. Figure 5.20 shows the temporal evolution of the ¹³⁷Csand ⁹⁰Sr concentrations at a station in the western Baltic from 1970 to 1989. The 1986

90 period is enlarged sixfold in order to elucidate the Chernobyl impact and the seasonal in- and outflow dependent on the freshwater balance. The lower concentration during winter time is the result of inflow of North Sea water; the higher values during spring and summer times is the result of the southward movement of more contaminated water from the northern Baltic. The increase in the activity concentration of ¹³⁷Csfrom 1977 to 1980 reflects the increasing discharges of this nuclide at Sellafield in previous years. The same is true for the overall decreasing values between 1980 and April 1986.

⁹⁰Sr was almost homogeneously distributed in the water column with highest activity concentrations in the western Baltic around 20 Bq m^-3 and 10 Bq m^-3in the Bothnian bay. Plutonium isotopes were determined at very low levels between 2 and 10 mBq m^-3 for ^239,240pu, reflecting global fallout levels.

Figure 5.19 Concentration of ¹⁰³Ru and ¹⁰⁶Ru in Mytilus sp. (Bq kg^-1 dry weight) along the French Mediterranean coast; LD=detection limit. From Calmet et al., 1988.

Figure 5.20 Temporal variations in the concentrations of ¹³⁷Cs and ⁹⁰Sr (mBq l^-1) in water from Schleimündung in the western Baltic, from 1970 to 1989. The horizontal scale has been expanded six-fold over the period 1986 to 1989. (Reproduced by permission of H. Nies.)

Prior to the Chernobyl accident, artificial radionuclides have been detected in sediments only at levels typical for global fallout. The sediments of the Baltic Sea are mostly fine-grain soft mud, especially in basins, where partly anoxic conditions are prevailing in bottom waters. This fine grained material acts as a trap for various substances dissolved in the water column. There exists an inverse correlation between Cs inventories in sediments and the corresponding water depth. This is a consequence of the decreasing accumulation rate of sedimentation with increasing depth.

The Chernobyl fallout has contaminated the sediments of the Baltic Sea to a higher degree in those areas where the contamination in the water has been the highest, namely along the northern coast of the Baltic Sea and the Bothnian Sea and the bay of Bothnia/Baltic Sea. Between 5 and 20 per cent of the Chernobyl fallout deposited at the sea surface in 1986 has been bound to sediments.

Waterborne radioactive contaminants from northern Europe are dispersed by the north-east Atlantic Ocean current system (Figure 5.21). For example, contaminants from the Irish Sea and the Baltic Sea move via the North Sea up along the Norwegian west coast with the Norwegian Atlantic current. At about 70°N, the current branches. An eastern part runs along the Kola peninsula and moves in the direction of the Siberian north coast and Novaya Zemlya; but the major part turns northward and enters the Arctic Ocean through the Fram strait between Spitzbergen and north-east Greenland. The transfer through the Fram strait represents more than 7 million cubic metres of Atlantic water per second. From the Arctic Ocean, the water leaves through the Fram strait as a surface current: the East Greenland current running southwards along the shelf of East Greenland. The current runs clockwise around Greenland, but on its way the water is increasingly mixed into the north-east Atlantic Ocean (Figure 5.22).

Figure 5.21 Transport routes of ¹³⁷Cs deduced from the measurements of the activity concentration distribution in the years 1971 to 1984. Dotted lines indicate different temporal transport routes. From Kautsky, 1988.

Figure 5.22 Main current directions in the northern Atlantic and Arctic Oceans; dotted lines represent relatively warm water and solid lines relatively cold water. From Rey, 1982.

Since the early 1960s seawater concentrations of ⁹⁰Sr and ¹³⁷Cs have been followed at a number of stations on the shelf in the north-east Atlantic, notably at the Faroe Islands and along the East Greenland coast. The main source of radioactive contamination at these stations has been global fallout from nuclear weapons testing in the atmosphere.

While the shelf water at the Faroe Islands is equivalent to the surface water of the north-east Atlantic Ocean, the East Greenland shelf water is almost identical to water in the East Greenland current representing the surface water of the Arctic Ocean. It is possible to estimate the mean residence time of water in the upper mixed layer of the north-east Atlantic Ocean and the Arctic Ocean to 25 years and 22 years respectively. It appears that ⁹⁰Sr concentrations in the Arctic Ocean are 2.1 times those in the north-east Atlantic while the ¹³⁷Cs levels differ by a factor of 1.7 only. Hence ¹³⁷Cs/⁹⁰Sr ratio is lower in the Arctic than in the Atlantic Ocean. The reason for this may be that the runoff from the Siberian rivers is enriched in ⁹⁰Sr and that this runoff plays a relatively more important role for the concentrations in the Arctic Ocean than runoff from rivers do for other parts of the Atlantic Ocean. Another factor could be that ¹³⁷Cs behaves less conservatively than ⁹⁰Sr, i.e. is sedimented to a larger extent than ⁹⁰Sr. Why the concentrations of both radionuclides are higher in the Arctic Ocean than in the north-east Atlantic is not known. The deposition of global fallout is less over the polar region than at the temperate latitudes. Although Pacific water shows two times higher concentrations than Atlantic water, inflow from the Pacific through the Bering Strait (16 per cent of the total inflow to the Arctic Ocean) is not sufficient to explain the higher levels. Two other sources may be considered. One possible explanation is run off from the Siberian rivers, in particular if these rivers have been heavily contaminated by sources other than global fallout. This has been recently suggested by T. B. Cochran following a visit to the Chelyabinsk region in the southern Urals, where it is thought that rivers and lakes may have been used for deposition of high-level radioactive waste in the late 1940s and early 1950s during the early phases of the Soviet nuclear weapons programme. In addition, recent unofficial reports have implied that substantial quantities of radioactive waste may have been dumped by the former Soviet Union in the Kara and Barents Seas. The radionuclide content and packaging of this material is not known in detail. A number of collaborative international projects have been initiated (in 1992) to seek to establish the magnitude of the input and the potential consequences in terms of environmental contamination and dose implications.

⁹⁹Tc is a radionuclide present in the liquid waste from nuclear reprocessing. Its half-life is 2 x 10⁵ y and it is not readily adsorbed by sediment particles. This makes it an excellent tracer of water movement. In the order of 1 PBq of ⁹⁹Tc have been discharged to the sea from nuclear reprocessing in western Europe. Increased discharges began in 1970 when Sellafield commenced to release approximately 40 TBq annually. Observations of ⁹⁹Tc in Fucus collected in the Danish straits (Kattegat) in the period 1967

84 (Figure 5.23) show a significant increase in the concentrations in 1973

74 suggesting a transit time from the Irish Sea of 7 years. Measurements of water and Fucus concentrations have allowed the tracking of Sellafield-derived ⁹⁹Tc through the Norwegian and Greenland Seas and of La Hague-derived ⁹⁹Tc along the Channel, southern North Sea into the Kattegat.

Figure 5.23. Annual mean values of ⁹⁹Tc (Bq kg^-1, dry) in Fucus vesiculosus collected from the Kattegat from 1967 to 1984. (The number of samples and SE are indicated). From Aarkrog et al., 1987.

Extensive use has been made of the more soluble radionuclides ¹³⁷Cs, ¹²⁵Sband ⁹⁹Tc to develop and validate transport models in the Irish Sea, Channel, North Sea and throughout the north-western continental shelf. In some cases, models have been developed to assist in radiological dose assessments, particularly for collective dose assessment (CEC, 1990). In other cases, radionuclide data have been used to validate models developed to examine the movement of water masses, for example the Channel and North Sea (Guary et al., 1988), to understand better the influence of winds and tides on water circulation, and to allow predictions to be made about the fate of other contaminants in these waters (Howorth and Kirby, 1988).

The current state-of-the-art of coastal modelling of relevance to waste inputs by sea dumping or land-based discharges, has been critically reviewed by a recent GESAMP Working Group (IAEA, 1990).

5.5 DISPOSAL IN THE DEEP OCEAN

The major source of most pollutants to the open ocean is atmospheric deposition. This is also true for artificial radionuclides but direct waste dumping and land-based discharges represent additional sources.

The major processes of radionuclide transfer in the open ocean have been mentioned in previous sections. Vertical transfer of radionuclides from the surface water occurs primarily via the mediation of the zooplankton (faecal pellets, moults, carcasses). This type of transfer is described in Sub-section 5.4.2.1. Material of biogenic origin is partly degraded in the water column. Consequently, some elements are redissolved in intermediate waters where they have a longer residence time than in surface waters because scavenging particles are less numerous (see Figure 5.18). For many radionuclides and trace elements enhanced scavenging occurs at the ocean margins (Sub-section 5.4.4.2).

The practice of dumping packaged, low-level, radioactive waste has been carried out at a number of sites in the North Atlantic since 1946 (NEA, 1985).

This has produced significant, potential sources of a wide range of alpha-, beta and gamma-emitting radionuclides to the bottom waters and hence via various pathways, potential sources of radiation exposure to marine fauna and ultimately to man.

Packaged waste has been disposed of in the north-east Atlantic by seven European countries at several sites (Figure 5.24). The emphasis has changed from using relatively shallow continental shelf sites in the early years to disposal on the deep ocean seafloor. The total quantities dumped in the north-east Atlantic in the period 1949 to 1982 amounted to approximately 680 TBq of alpha-emitters and 56 PBq of beta/gamma-emitters, of which 15 PBq was attributed to ³H in the period 1975

82. There have been no further dumping operations since 1982. In the region of 3.7 PBq of `mixed' activity, packaged waste was dumped by the USA in the north-west Atlantic between 1946 and 1967. About 95 per cent of the total activity, including 1.2 PBq of activation products in the `Seawolf' reactor, was disposed of at the 2800 m site (38°30'N, 72°06'W) (NEA, 1985). In comparison, the total alpha discharge from the Sellafield reprocessing plant in the period 1957 to 1981 amounted to 1.3 PBq, with 47 PBq Of ¹³⁴Cs, ¹³⁷Csplus ⁹⁰Sr (NEA, 1985).

Dumping operations in the north-east Atlantic during the 1950s and early 1960s took place without international control. Since 1967, dumping operations have taken place under the auspices of the Nuclear Energy Agency (NEA, formally ENEA: European Nuclear Energy Agency) of the Organization of Economic Cooperation and Development. The OECD Council established a Multilateral Consultation and Surveillance Mechanism for Sea Dumping of Radioactive Waste in 1977 (OECD, 1983) to further the objectives of the London Dumping Convention (LDC). A periodic assessment of the continued suitability of dumping sites is required under the Mechanism, taking account of the provisions in the LDC and the International Atomic Energy Agency (IAEA) Revised Definition and Recommendations (IAEA, 1984).

Two assessments have been published for the current dumpsite at 46°N, 17°W (NEA, 1980, 1985) and a third was carried out in 1990. The later assessments were aided by the creation of the NEA Coordinated Research and Environmental Surveillance Programme (CRESP) in 1981. This was intended to increase knowledge of the processes controlling radionuclide transport to biota and humans. The programme has focused on, but not exclusively, the north-east Atlantic with site-specific studies at the current dumpsite, and the results published as CRESP reports and in a three-volume series on the oceanographic description of the site (Gurbutt and Dickson, 1983; Dickson et al., 1986; Nyffeler and Simmons, 1989). Fourteen countries and two international bodies have been represented on CRESP, with Working Groups covering physical oceanography, geochemistry, biology, modelling and radiological protection.

Figure 5.24 Approximate locations of dumpsites used by European nations in the northeast Atlantic from 1950 to 1982. From NEA, 1985.

To some extent CRESP can be seen as a parallel to the NEA Seabed Working Group (SWG). The SWG was set up to provide a scientific basis for assessing the feasibility of the sub-seabed disposal of high-level (or heat-generating) radioactive waste. The group considered problems associated with site-selection, emplacement, leaching, remobilization, vertical migration to the sediment

water interface, biological interactions and advective and diffusive transport in the water column (NEA, 1988). The last category is of equal relevance to low-level waste disposal. Site-specific studies were carried out at several locations, including Great Meteor East (GME) on the Madeira Abyssal Plain to the south of the current dumpsite (Figure 5.24).

Two other completed programmes are of particular relevance. The Dutch DORA programme formed an integral part of CRESP, providing detailed information on biogeochemical processes and pathways on the dumpsite (Rutgers van der Loeff and Lavaleye, 1986). The German North Atlantic Monitoring Programme (NOAMP) examined a region immediately to the north-west of the dumpsite (Figure 5.25). The principal aims were to study the pathways of suspended matter in the deep ocean and possible vertical transport out of the near-bottom layer (Mittelstaedt, 1986; Dickson et al., 1986; Nyffeler and Simmons, 1989).

There are two major international programmes which have beep initiated recently, and which will contribute to any further assessments of past or future disposals of radioactive waste by improving our knowledge of ocean processes. The World Ocean Circulation Experiment (WOCE) represents an attempt to describe and explain the major features of the ocean circulation. The Joint Global Ocean Flux Study (JGOFS) is designed to further our understanding of vertical fluxes, carbon cycling and benthic exchanges.

The models used in the radiological assessments of the north-east Atlantic dumpsite (46°N, 17°W) predict very low concentrations of radionuclides in seawater and sediments as a result of past dumping practices (NEA, 1980, 1985). These predictions have been supported by measurements of radionuclide concentrations in seawater, sediments and biota from on and around the dumpsite. In general, radionuclide distributions are dominated by global fallout. It follows that the biogeochemical pathways which are described below should, for the most part, be considered as potential pathways with regard to radioactive waste. In practice, it has proven to be very difficult to establish the degree of environmental contamination resulting from the dumping itself.

The current dumpsite is located about half-way between the mid-Atlantic Ridge and the west European continental margin, on the easternmost part of the province at the southern end of the Porcupine Abyssal Plain (Gurbutt and Dickson, 1983). The area is characterized by a flat-bottom valley (> 4750 m) in the western half flanked by a relatively abrupt ridge and trough topography trending approximately north-south, with the crests of hills reaching to water depths of 3750 m. The entire area has been charted with a narrow beam echo sounder, `Sea Beam', providing maps at a scale of 1 : 50 000 and 1 : 100 000 with contour intervals of 20 m (Dickson et al., 1986). `Sea Beam' has also been used in the NOAMP area.

Figure 5.25 Hypothetical diagram of radionuclide transfer into different layers of the water column by the food-chains in which Eurythenes gryllus might be directly involved. From Nyfeller and Simmons, 1989.

Current velocities in the dumpsite and NOAMP regions have been examined using bottom-mounted, long-term current meter moorings. Current velocities and directions in the near-bottom region

extending to about 1000 to 1500 m above the seafloor

are strongly influenced by the bottom topography and exhibit considerable spatial variability. The mean flow in the upper water column, in the upper 1500 m, is influenced by the near-surface flow and is largely independent of the near-bottom flow. The region between depths of 1500 and 3000 m is partially influenced by the near-surface and near-bottom flows and by a thermohaline-driven circulation.

Eddy kinetic energies are greater than the mean flow and the timescales are complex. The sparsity of the current meter records, combined with the observation that the fluctuating component, resulting from eddy activity, is large compared with the mean flow, introduce uncertainties in the reliability of mean flow estimates. However, an analysis of all available records suggests an overall slow northerly mean flow over much of the region (Gurbutt and Dickson, 1983; Nyffeler and Simmons, 1989). The long-term mooring at NEADS 6 (near the foot of the Rockall Trough), whilst showing an overall north-westerly drift, does record significant periods when the residual flow was to the south-west. Although the mean current velocities are relatively low (1 to 3 cm s^-1) the combination of short-term (1 to 3 weeks) high bottom current benthic storms, > 10 cm s^-1, and the semi-diurnal tidal currents can lead to instantaneous velocities of 20 to 30 cm s^-1 close to the bottom (Nyffeler and Simmons, 1989). This causes the resuspension of significant quantities of surficial sediments which may remain in suspension as a bottom nepheloid layer (BNL) for several days. The combination of the considerable topography (1000 m), vertical mixing and lateral advection result in the occurrence of a BNL, up to 1000 m above the seafloor, frequently showing a number of intermediate layers (Dickson et al., 1986). The near-bottom eddy diffusivity on the dumpsite was estimated from measurements of excess ²²²Rnin bottom waters, which diffuses out of the seabed following the decay of naturally occurring ²²⁶Ra. The diffusivity was proportional to the local density gradient and gave a bottom mixed layer of 50 m. The occurrence of concentration maxima at greater distances above the seabed was interpreted as evidence of lateral advection from surrounding hills (Dickson et al., 1986).

The pattern of water transport has also been studied using neutrally buoyant floats. Four floats released into the bottom waters of the western end of the dumpsite in 1976 were tracked for up to 17 days in a southerly to south-westerly direction, in contrast to a bottom float and two mid-depth floats which were tracked in a westerly to north-westerly direction

with mean velocities of 2 to 3 cm s^-1 and an overall cyclonic path apparently influenced by the local topography (Gurbutt and Dickson, 1983). Fourteen SOFAR (sound fixing and ranging) floats launched in 1985 in the NOAMP area were tracked for up to 1 year at a depth of 3300 to 4000 m. These yielded a horizontal diffusivity of 10⁶ cm² s^-1 and a mean velocity of 1 cm s^-1, with a south-westerly tendency. The floats dispersed in an area of about 300 x 300 km over a year with a domain of occupation of 100 km in diameter in 180 days. Similar diffusivities have been obtained for the Iberia Abyssal Plain, although at long time scales dispersion decreases as the floats reach the sides of the basin. In contrast, floats in the Madeira Abyssal Plain have not yet filled the basin and their dispersion increases.

A major difficulty is to estimate vertical velocity of bottom waters. The general average upwelling velocity of ocean waters is about 10 m per year: such estimates are made as a compensation for the downwelling waters (waters that are saltier and cooler will sink to a level appropriate to their density). Large variations can occur in the upwelling velocity of deep waters. It has been shown that these variations can depend on local topography or meso-scale events (Armi and D'Asaro, 1980) which are quite common in the Atlantic. Such events may affect the vertical distribution of radionuclides and particles (Lambert et al., 1983; 1984).

Several attempts have been made to determine waste-derived radionuclides in the bottom waters on and around the dumpsite to estimate the degree of local contamination and the effect of advective and diffusive transport. Tritium has been dumped in large quantities (> 15 PBq), has a high solubility and was mostly disposed of in vented drums. It could be anticipated that detectable levels would exist on the dumpsite. In order to check the assumptions made for the release rate and initial dispersion in the radiological assessment model (NEA, 1985) 176 water samples from the bottom 1000 m at 18 sites in the north-east Atlantic were collected for tritium analysis in 1986. There was considerable horizontal and vertical variability, with many concentrations being not significantly different from zero. Five stations produced inventories in the bottom 1000 m which were statistically significant. There were three sites on the dumpsite and two to the south and south-west (Figure 5.24). Taking a mean value from the five sites, and a 3° x 3° square around them, resulted in an inventory of 1460 TBq, which is approximately 10 per cent of the total decay-corrected inventory in 1986 from all disposals to the deep north-east Atlantic (Nyfeller and Simmons, 1989). The implied movement to the south-west was not predicted from the current meter records, but the limited float data suggest that geographically and temporally limited excursions to the south-west are possible. The inventory estimates are a gross approximation and further measurements would be necessary to attribute greater credence to these preliminary findings.

Water column profiles of ^239,240Pu and¹³⁷Cs in the NOAMP and dumpsite areas showed no enhancement which could be ascribed to dumping operations, i.e. the profiles were consistent with a fallout origin and did not differ between the two regions. The approximately exponential decrease in the ¹³⁷Cs concentration with depth contrasted with the profiles of ^239,240Pu which exhibited marked concentration maxima at about 1000 m, as a result of biological activity and the remineralization of sinking particles. No significant enhancement of ^239,240Pu was observed in near-bottom (+ 5 m) water samples taken from throughout the north-east Atlantic basin. ^239,240Pu/²³⁸Pu ratios were lower than in surface samples taken at the same position.

Most of the seafloor at the dumpsite lies at or a little above the estimated depth of the lysocline (4700 m) and this is reflected in average Holocene sediment accumulation rates of 2 cm ky^-1. Deeper core profiles have yielded slightly higher accumulation rates due to the influx of ice-rafted material in the late Pleistocene. In the hillier areas there is evidence that localized slumping and episodic deposition has taken place.

The extent of bioturbation in the sediments at the dumpsite appears to be of the same order of intensity as other areas of the Atlantic. The irregular profiles of both Pu and ²¹⁰Pb are indicative of episodic mixing events, predominantly by Sipunculida (Rutgers van der Loeft and Lavaleye, 1986). Most of the meio-, macro- and megabenthos are deposit-feeders and occur in the upper 10 cm.

Decomposition of the most reactive organic matter takes place at the sediment surface with dissolution of some CaCO₃ and release of organically bound Pu, Am and Sr. These reactions take place at reduced rates with depth in the sediment. In the 10

70 cm zone there appears to be little biological or chemical change in condition, but below 70 cm Fe and Mn are reduced and associated elements may be released. Significant sulphate reduction is restricted to turbidity deposits (Rutgers van der Loeff and Lavaleye, 1986).

Concentrations of Pu and Am in bottom sediments are consistent with a fallout origin. However, ^239,240Pu/²³⁸Puratios provide equivocal evidence of a possible dumpsite origin for some of the Pu. K_d values of Pu and Am obtained by filtering large-volume bottom-water samples were similar to those from coastal sites.

Many analyses of biota have been completed, including Holothuroidea, Actiniaria, Cephalopoda, Decapoda and Coryphaenoides armatus (rat tail). In general the variety and concentrations of radionuclides present are indicative of fallout origin. Some muscle samples from Coryphaenoides had relatively low ^239,240Pu/²³⁸Pu ratios, suggesting contamination by dumpsite Pu, but the errors on the determinations were considerable. Hypothetical pathways for radionuclide transfer away from the dumpsite have been suggested based on the ubiquitous Amphipod Eurythenes gryllus (Figure 5.25; Nyfeller and Simmons, 1989).

The 91-box SSR model (NEA, 1985) contained representations of the perceived key physical, chemical and biological processes. Generally, pessimistic assumptions were made about the rates of processes in the model (e.g. release rate), particularly when data were lacking. Additional measurements made after completion of the 1985 SSR have allowed a comparison to be made between observed and predicted concentrations on and around the dumpsite. Clearly radionuclide concentrations were overestimated at the dumpsite, suggesting that estimates of dose to local fauna had been overestimated. The effects at greater distances is less clear but it seems likely, on balance, that doses in the far field will also have been overestimated in the 1985 SSR.

5	Radionuclide Aquatic Pathways
	Co-ordinators:	J. Hamilton-Taylor, M. Kelly, P. Kershaw and C. E. Lambert
	Contributors:	A. Aarkrog, D. P. Calmet, S. Charmasson, R. Carpenter, S. Fowler, M. Ivanovich,
		G. Kuznetsov, S. P. Luttrell, S. J. Malcolm, P. I. Mitchell, H. Nies, B. Patel,
		G. G. Polikarpov, I. Rjabov, D. Swift and U. Tveten

5.1 Introduction
5.2 Freshwaters
	5.2.1 Transport and Dynamics in Rivers
		5.2.1.1 Solution phase
		5.2.1.2 Sediments
	5.2.2 Transport and Dynamics in Lakes and Reservoirs
		5.2.2.1 Solution phase
		5.2.2.2 Sediments
	5.2.3 Transport and Dynamics in Groundwater Systems
		5.2.3.1 Introduction
		5.2.3.2 Tracing applications
		5.2.3.3 Dating applications
	5.2.4 Biogeochemical Transformations and Chemical Species in Freshwaters
		5.2.4.1 Solid-solution partitioning
		5.2.4.2 Chemical speciation
		5.2.4.3 Redox boundaries as critical features in the biogeochemical behaviour of artificial radionuclides
	5.2.5 Lake Sediments and Postdepositional Change
	5.2.6 A Case Study Dealing with Exposure Pathways in Scandinavia
		5.2.6.1 Runoff characteristics and surface water and sediment activity
		5.2.6.2 Drinking water
		5.2.6.3 Freshwater fish
5.3 Estuaries and Intertidal Environments
	5.3.1 Introduction
	5.3.2 The Estuarine Environment
	5.3.3 Transport of Radionuclide in Solution
	5.3.4 Transport and Deposition of Radionuclide Particulate Phase
	5.3.5 Particle-Solution Reactions
	5.3.6 Diagenesis
	5.3.7 Radionuclide Budgets and Inventories
	5.3.8 Other Coastal Intertidal Environments
	5.3.9 The Future of Radioactive Contaminated Coastal Environments
5.4 Coastal, Semi-enclosed Basins, Shelf and Continental Margins
	5.4.1 Introduction
	5.4.2 Biomediated Pathways
		5.4.2.1 Biomediation in the water column
		5.4.2.2 Biomediation in the seabed
	5.4.3 Diagenesis
	5.4.4 Transport and Enhanced Boundary Scavenging
		5.4.4.1 Coastal mechanisms
		5.4.4.2 Cross-shelf transport and boundary scavenging
		5.4.4.3 Transport in marginal basins and semi-enclosed inland seas
		5.4.4.4 Long distance transport pathways
		5.4.4.5 Radioactive tracers and transport modelling
5.5 Disposal in the Deep Ocean
	5.5.1 Introduction
	5.5.2 Radioactive Waste Disposal
		5.5.2.1 Control and Assessment
		5.5.2.2 Advection and dispersion
		5.5.2.3 Biogeochemistry
5.6 Recommendations


	Solution phase			Particulate phase
	(Bq/m³)			(Bq/kg)				Dominant
Estuary	¹³⁷Cs	^239,240Pu	Others	¹³⁷Cs	^239,240Pu	Others	Origin	source

Connecticut, 1983		0.002			0.006		Weapons	Fluvial
(n = 710)							test fallout	+ marine
Savannah, 1986)		0.003		41.8	0.439		Weapons	Fluvial
(n = 5)							test fallout
							+ reprocessing
Seine, 1979	12.6	0.003		39.2	2.4	¹⁰⁶Ru134	Weapons	Fluvial/
(n = 1113)							test fallout	marine
							+ reprocessing
Dnieper, 1988	116		⁹⁰Sr 410				Chernobyl	Fluvial
(n = 3)							fallout
Esk, 1981	12660	18.5	²⁴¹Am 53	16260	12790	²⁴¹Am22150	Reprocessing	Marine
(n = 17171)			¹⁰⁶Ru 1050			¹⁰⁶Ru 790909

Refs: Connecticut (Sholkovitz and Mann, 1987); Savannah (Olsen et al., 1989);Seine (Jeandel et al,, 1980); Dneiper (Polikarpov et al,, 1991);
Esk (Assinder et al,, 1985)


Estuary		Area	⁶⁰Co (GBq)	¹³⁷Cs (GBq)	^239,240Pu (GBq)	²⁴¹Am (GBq)
		(km²)	(fallout %)	(fallout %)	(fallout%)	(fallout %)

Hudson, US, 1975		240	8 (0%)	925 (76%)	41 (100%)
Columbia, US,		285	555 (0%)	851 (?)	30 (95%)	9 (0%)
	1977/78
James, US, 1981		380	95 (0%)	766 (83%)
Esk, UK, 1989		5.6	?	4500 (<l%)	380 (<1%)	490 (0%)

References: Hudson (Olsen et al., 1981); Columbia (Beasley and Jennings, 1984); James (Zucker et al., 1984); Esk (Burton, 1988; Kelly and Emptage, 1991).