modelling the distribution of tc-99 along the east coast of ireland · 2013-07-22 · a.i. olbert...

A.I. Olbert and M. Hartnett Modelling the distribution of Tc-99 along the east coast of Ireland

Page 1 of 67

Modelling the distribution of Tc-99 along

the east coast of Ireland

Final report

by

Agnieszka I. Olbert, Ph.D

Michael Hartnett, Ph.D

Department of Civil Engineering

&

Environmental Change Institute

National University of Ireland, Galway

for

Radiological Protection Institute of Ireland

DP/281107


Page 2 of 67

Executive summary

Technetium-99 (Tc-99) is a highly soluble radionuclide with a very long half-life

(t1/2=2.13x105 years). Due to its beta-emitting characteristics it is an important component

of high-level radioactive wastes (Wildung et al., 1979). The main source of this

radionuclide in the Irish Sea waters is the British Nuclear Fuels (BNF) reprocessing plant

at Sellafield (Cumbria, UK). Controlled discharges into the north-eastern Irish Sea began

in 1952 and since then the plant has been the subject of national and international

controversy. Commissioning and operation of the Enhanced Actinade Removal Plant

(EARP) at Sellafield resulted in additional discharges to the Irish Sea and has a well-

recognized impact on the Irish Sea ecosystem (Hunt et al., 1998; Smith et al., 2001).

In this project a validated transport model was used to hindcast the transport of Tc-99 and

ultimately to examine the seasonal and interannual circulation pattern within the Irish

Sea. Findings from the model were used in the analysis of past RPII monitoring and to

assess the adequacy of the historical sampling regime. Recommendation for optimization

of monitoring programme to the RPII is one of the primary goals of this project.

The study has been carried out as a result of collaborative work between the Marine

Modelling Group (MMG) and the Radiological Protection Institute of Ireland (RPII); the

RPII provided an extensive dataset on Tc-99 concentrations within the Irish Sea, the

MMG was responsible for undertaking transport and dispersion modelling of Tc-99.

Simulations of Tc-99 transport covers a period of December 1993 to December 2001.

The hydrodynamic model was validated against various sources hydrographical data. The


Page 3 of 67

performance of the transport model was assessed by comparison against existing data:

distributions of Tc-99 at offshore locations within the Irish Sea compiled from cruise

surveys and time traces obtained from inshore regular sampling at Balbriggan (Ireland).

Although, the Irish Sea is a complex system with large fluctuations in the discharge and

seasonal variations in flow fields, the numerical model demonstrated very good

capabilities to predict migration pattern of Tc-99 on temporal and spatial scales. The

transport model of Tc-99, first of this kind for the Irish Sea, gives detailed knowledge on

spatial and temporal distributions throughout the entire study period.

Additional modelling work involved numerous tests to establish the model’s sensitivity to

input data, in particular, the effect of meteorological forcing on transport patterns, and

travel times under seasonal conditions. Wind action, through modification of advective

transport, dispersion and mixing of material is primarily responsible for transporting Tc-

99 to the east coast of Ireland and is a prime contributor towards high concentrations at

Balbriggan. The model demonstrates strong relationship between concentration peaks on

the east coast of Ireland and strength of the western Irish gyre. Air temperature variation

has greatest effect during summer months through its role in the development of density-

driven baroclinic circulation.

Relationships between discharge time and timing of far field concentrations are highly

variable and depend not only on discharge fluctuations but also, quite significantly, on

sea dynamics that controls the accumulation and removal of Tc-99 mass. Away from the

immediate vicinity of Sellafield, the impact of the fluctuations in the discharge is

progressively less marked because of long residence time and strong mixing in the sea.


Page 4 of 67

Since migration of Tc-99 to the east coast of Ireland depends primarily on variations in

hydrodynamics of the Irish Sea, transit times vary greatly. Time traces show that arrival

time and amount of material transported is highly depended on seasonal release and also

illustrate significant interannual variability.

The above findings were used in an assessment of the current monitoring and also in

optimizing future monitoring programmes. The current monitoring is generally

considered to be quite effective, though the inshore sampling is probably too sparse over

summer months and as a result short-lasting peaks in this period are likely to be missed.

The locations for monitoring stations (Dunmore East, Cahore, Bull Island, Balbriggan,

Carlingford) covering the eastern coastline of Ireland are well selected. Offshore

sampling provides a robust picture of spatial distributions. Future monitoring may be

optimized in a couple of ways: (a) by increasing the frequency of sampling during

summer months at inshore locations, and (b) by changing the timing of offshore sampling

in order to obtain a more comprehensive image of seasonal variations.


Page 5 of 67

1. Introduction

Technetium-99 (Tc-99) is a highly soluble radionuclide with a very long half-life

(t1/2=2.13x105 years). Due to its beta-emitting properties it is an important component of

high-level radioactive wastes (Wildung et al., 1979). Over many decades Tc-99 has been

regularly detected at elevated levels in the Irish Sea. The main sources of this

radionuclide in the Irish Sea waters are discharges from the British Nuclear Fuels (BNF)

reprocessing plant at Sellafield (Cumbria, UK). Radionuclides derived form Sellafield are

now also known to be a major source to the shelf seas of north-western Europe (Orre et

al., 2007) and north east Atlantic (Kershaw, 2004).Controlled discharges into the north-

eastern Irish Sea began in 1952 and since then it has been the subject of national and

international controversy. Following the introduction of stricter discharge authorizations

and new effluent treatment systems in the 1980’s, sharp reductions occurred in releases of

most radionuclides from Sellafield (McCubbin, 2002). Despite initial significant

reductions, the discharge of Tc-99 increased from 1994 and continued to be elevated in

comparison to previous decade. This increase was due to a commissioning and operation

of the Enhanced Actinade Removal Plant (EARP), an additional treatment plant designed

to reduce alpha and beta activity from effluent prior to discharge. Although the activity

has been substantially reduced, the operation of the plant has allowed the continued

treatment of concentrated effluents, which previously had been accumulated on site but

necessitates the discharge of the Tc-99 which is not significantly removed during the

EARP process (Leonard et al, 2004). This step resulted in additional discharges to the

Irish Sea and had a well-recognized impact on the Irish Sea ecosystem.


Page 6 of 67

The presence of Tc-99 in Irish coastal waters and significantly elevated values since the

EARP era is of environmental risks to the marine biota (Hunt et al., 1998) and to the

seafood consumer (Smith et al., 2001). Both British and Irish authorities, concerned about

radiological impact, established their own monitoring systems within the Irish Sea.

Although, in recent years the monitoring of Tc-99 has greatly improved in terms of

spatial and temporal frequency, the transport pathways and residence timescale cannot be

firmly determined from available datasets. Unfortunately, understanding the circulation

pattern and associated transport of Tc-99 material in a dynamically changing

hydrosystem is a complex task and requires a wide range of information. Usually,

historical timeseries and spatial distribution maps provide the best insight into the system.

From our experience, the combination of modelling and data collection provides an

appropriate approach to developing an understanding of hydrodynamics and solute

transport. Using the monitoring data for validation of numerical model and then

improving the monitoring programme based on the numerical predictions is the most

technically and economically effective way of monitoring and managing a system such as

the Irish Sea.

In this project a validated transport model was used to hindcast the transport of Tc-99 and

ultimately to examine the seasonal and inter-annual circulation patterns within the Irish

Sea. The findings from the model study were then used to analysis past monitoring and to

assess the adequacy of the sampling regime. One of the primary goals of this project is to

suggest improvements to the RPII regarding the radionuclide monitoring programme in

the western Irish Sea.


Page 7 of 67

1.1 Project aims and phases

The study has been carried out as a result of collaborative work between the Marine

Modelling Group (MMG) at the National University of Ireland, Galway and the

Radiological Protection Institute of Ireland (RPII) environmental monitoring section.

RPII provided an extensive dataset on Tc-99 concentrations within the Irish Sea and

MMG was responsible for undertaking transport and dispersion modelling of Tc-99

discharges from Sellafield reprocessing plant. Radionuclide Tc-99 originating from the

nuclear fuel reprocessing plant in Sellafield was modelled in this study using advanced

numerical techniques. A hydrodynamic model of the Irish Sea with conservative tracer

transport mode was developed to study Tc-99 migration pattern within the sea, and in

particular to provide detailed information on concentration levels and distributions along

the western Irish Sea. It is expected that model results will assist the RPII’s

environmental monitoring section to:

examine the adequacy/effectiveness of the RPII monitoring methodology

assess the optimum frequencies and densities of data collection.

The key findings from the modelling will be combined with the RPII’s knowledge and

experience to draw inferences about the adequacy of current monitoring scheme and to

introduce improvements to the system, such as:

the basis for RPII marine monitoring future programme

framework for interpretation of monitoring data.


Page 8 of 67

This report focuses mainly on the modelling component of the overall RPII analysis; it

provides details of transport pattern of Tc-99 in the Irish Sea. The hydrodynamic and

transport model developed by the MMG enabled the analysis of Tc-99 spatial

distributions for a specific point in time as well as historic time series of concentrations at

a particular point in space. The outputs from the model were later utilized in

understanding seasonal effects such as the western Irish Sea gyre on the Tc-99

distributions and, in broader timescales, inter-annual variability. This knowledge in turn

was used to estimate transit times and residence times and their variations with respect to

the above external drivers.

The project was carried out in three phases:

Phase 1 – collection of available data

Phase 2 – model validation

Phase 3 –assessment of current monitoring program and interpretation of monitoring data

based on model outcomes.

Details on numerical model setup, validation of hydrodynamics, radionuclide transport

and significant conclusions from the modelling work are presented in the following

sections:

Data collection

Model setup, validation and sensitivity

Tc-99 circulation, transit and residence time

Assessments and recommendations for monitoring programme

Conclusions


Page 9 of 67

2. Phase 1 - Collection of available data

2.1 Data collection

The first phase of the project involved collation of data necessary for hydrodynamic and

transport model. Two types of datasets were needed: data to force the numerical model

and data for model validation. For hydrodynamic simulations, an extensive dataset of

input conditions was required and ultimately acquired by the MMG from various sources.

The dataset includes initial physical conditions (temperature and salinity data), open

boundary conditions (time variable water elevations, temperature, salinity, river

discharges) and meteorological forcing. Validation datasets of water elevations and

temperature timeseries were also obtained. All details of hydrodynamic model setup and

validation results are presented below.

With regard to Tc-99 data, Sellafield discharges and spatial concentrations for model

initial conditions are the primary inputs. The discharges were compiled by the RPII and

incorporated into the model in the form of monthly mass loads from the plant. Initial

distributions of Tc-99 within the Irish Sea were generated from a CIR 11/93 survey data

carried out in December 1993.

Considering availability of discharge data (monthly values 1990-2006), field records for

initial conditions and validation data, it was agreed to perform a simulation for a period

of December 1993 to December 2001. The period covers a spin-up time (approx. 3

months) necessary for the development of steady-state hydrodynamic conditions and an


Page 10 of 67

accurate simulation of transport including the large increase in the rate of discharge of

Tc-99 (March 1994) following the start-up of the Enhanced Actinide Removal Plant

(EARP). The inshore and offshore measurements for model validation were partially

provided by the RPII and partially from field measurements published in literature.

2.2 Chronology of discharges

The existing dataset of Sellafield discharges contains a 55-year long history of annual

loadings. From these records an apparent difference in release rates between pre-EARP

and post-EARP period emerges. The pre-EARP era defines a period starting with the

opening of the Sellafield plant in 1952 up to March of 1994, when the operation of the

new EARP plant was commenced. During this period relatively small discharges of c. 8

TBq per annum in the 1950’s and 1960’s were followed by a significant increase in the

1970’s (40TBq/year). The 1980’s saw a reduction to approximately 5 TBq/year as a

result of stricter discharge authorizations and new effluent treatment systems introduced

in the early 1980’s. In the post-EARP period, the annual discharges increased

dramatically as the new treatment turned out to be ineffective in removing technetium

(Smith et al, 2001). The highest peak discharge was recorded in 1995 (c. 192 TBq) and

the mean annual discharge for years 1994-1997 was approximately 30 times higher than

for the period 1990-1993. Detailed Tc-99 monthly discharges covering the period of

simulation (11/1993-12/2001) are given in Figure 1.


Page 11 of 67

0.0E+00

5.0E+03

1.0E+04

1.5E+04

2.0E+04

2.5E+04

3.0E+04

Nov-9

3

Mar-9

4

Jul-9

4

Nov-9

4

Mar-9

5

Jul-9

5

Nov-9

5

Mar-9

6

Jul-9

6

Nov-9

6

Mar-9

7

Jul-9

7

Nov-9

7

Mar-9

8

Jul-9

8

Nov-9

8

Mar-9

9

Jul-9

9

Nov-9

9

Mar-0

0

Jul-0

0

Nov-0

0

Mar-0

1

Jul-0

1

Nov-0

1

Tc-

99(G

Bq

/mo

nth

)

Figure 1. Monthly discharges of Tc-99 from Sellafield nuclear plant.

2.3 Tc-99 distributions and temporal variations

The spatial distributions of Tc-99 within the Irish Sea were assessed from offshore

samples, while temporal variations were determined from inshore regular sampling. For

the period considered, seven surveys (11/1993-09/1998) were performed and maps of

concentrations at various locations within the Irish Sea were developed (see Figure 2 and

Figure 11 for contour maps). Notably, the amount of collected samples increased with

each survey, from 20 samples taken during a cruise in December 1993 to over 70 samples

uniformly covering the inner Irish Sea from a September 1998 survey. The southern Irish

Sea remains rather under-sampled throughout the survey period; hence greater

uncertainty is associated with Tc-99 values in this region. Although, the spatial


Page 12 of 67

distribution of sampling provides a reasonably robust picture of Tc-99 distributions, the

temporal scale is too coarse to draw inferences about transport pathways.

Samples collected during CIR 11/93 cruise were used to create initial conditions for the

Tc-99 transport model (Figure 2). Inventories from cruises CIR 12/94, CIR 5/95, CIR

10/95, CIR 11/96 and CIR 9b/98 were employed to validate the numerical model and are

shown in Figure 11.

Estimates of dissolved phase Tc-99 inventories range from 6 to 166 TBq and are of

similar magnitudes to the increase in annual Sellafield discharges. This is followed by a

significant difference in concentrations in the vicinity to discharge. The concentration

close to Sellafield in December 1993 was c. 30 mBq/l, and this compares to post EARP

values of c. 200 in December 1994, c. 1800 in December 1995, c. 500 in December 1996

and c. 40 in September 1998. The reduction by an order of magnitude between December

1996 and September 1998 results from greatly reduced discharges for the 6 months

proceeding the September 1998. These data provides some illustration of the response of

the water body in the eastern Irish Sea to fluctuations in discharges.


Page 13 of 67

Figure 2. Contour maps of the Tc-99 (mBq/l) distributions in the Irish Sea during

December 1993, CIR cruise. From McCubbin et al. (2002).

The near-shore measurements were used to create historic time series. The inshore

sampling programme, labeled as the CMP programme, was carried out along five

locations on the east coast of Ireland: Balbriggan, Bull Island, Carlingford, Cahore and

Dunmore East. For the period 1993-2005, in total 139 samples were analyzed, from

which 78 were collected in Balbriggan, 31 in Bull Island, 15 in Carlingford, 11 in Cahore

and 4 in Dunmore East. With regard to sampling frequencies, the Bull Island station is

relatively high monitored with approximately 1 sample per month, however, the sampling

times are outside our analysis period.

The Balbriggan records, with an average sampling frequency of 0.56 sample/month and a

long sampling history, were found to be quantitatively most suitable for an analysis of


Page 14 of 67

high-frequency fluctuations of Tc-99. With regard to the sampling quality, the RPII

claims uncertainties of the measurements as presented in Table 1.

Table 1. Uncertainties of the Tc-99 measurements.

Tc-99 concentration

mBq/l

Measurement error

%


Page 15 of 67

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

09/1

996

12/1

996

04/1

997

07/1

997

09/1

997

12/1

997

04/1

998

07/1

998

09/1

998

12/1

998

04/1

999

07/1

999

09/1

999

12/1

999

03/2

000

06/2

000

09/2

000

12/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

Measurements

Figure 3. Timeseries of Tc-99 concentrations with uncertainty bars at Balbriggan

collected during monitoring programme by the RPII.

2.4 Summary of Phase 1

Phase 1 involved the compilation of necessary datasets. RPII acquired a database of

monthly Tc-99 discharges from Sellafield, and inshore and offshore measurements of tc-

99 within the eastern and western Irish Sea.

MMG purchased high-temporal resolution meteorological data from the ECMWF for the

period of interest. Also, MMG have tested an alternative meteorological dataset compiled

from NCEP reanalysis global atmospheric model. Comparison of model results forced by

both datasets is presented in Section 3.5.


Page 16 of 67

3. Phase 2 - Model validation

In Phase 2 of the project MMG developed a numerical model of the Irish Sea and

implemented it to investigate Tc-99 transport pattern; the simulation of Tc-99 transport

covers a period of December 1993 to December 2001. The hydrodynamic model was

validated against various sources of hydrograph data. The performance of the transport

model is assessed by comparison against a relatively large volume of monitoring data

collected by RPII and others.

3.1 Study area and simulation domain

The Irish Sea is as a semi-enclosed body of water, being a part of the Northwest

European continental shelf extending northwards from St. David’s Head and Carnsore Pt.

on the Welsh and Irish coasts to the North Channel between Larne and the Mull of

Galloway, Gaffney (2001). Its approximate length is 300 km and its width varies from

75-200 km down to about 30 km in the North Channel. The mean water depth averages

around 60 m though several geographical sub-regions, differing in bathymetric

characteristics, can be distinguished. The North Channel, connecting the Irish Sea with

the Atlantic Ocean in the north, is deep and narrow, with maximum depths exceeding 275

m. This channel extends southward, becoming slightly shallower in the St. George’s

Channel, and is 300 km in overall length, 30-50 km in width, with minimum depths of c.

80 m.


Page 17 of 67

3.2 Numerical model description

A hindcast of hydrodynamic conditions in the Irish Sea was successfully performed using

the Princeton Ocean Model (POM) model (Blumberg and Mellor, 1987; Mellor, 2001).

POM is a numerical model simulating the effects of tides, winds and density gradients on

water level and three-dimensional currents. The development of the hydrodynamic

module was followed by the application of a solute transport model, so that predictions of

three-dimensional temperature and salinity distributions could be made. This feature was

necessary for accurate simulation of density driven baroclinic flows. Through the solution

of the advection-diffusion equations the solute transport model calculates the transport of

conservative tracers (e.g. Tc-99). The model has been continuously enhanced and recent

improvements employ generalized open boundary conditions, curvilinear orthogonal

coordinates for horizontal grids and advanced numerical approximation schemes for steep

solute gradients.

Vertical diffusion is determined from transport of the vertical turbulence kinetic energy

and turbulence macroscale defined by Mellor-Yamada (M-Y) equations, while the

horizontal diffusion term is calculated with a variable value of horizontal viscosity and

diffusivity of the Smagorinsky formula. The model also accommodates realistic coastline

geometry and bottom topography. Vertical processes are solved on a sigma coordinate

system capable of resolving coastal and deep ocean regions.


Page 18 of 67

3.3 Model set-up

3.3.1 Hydrodynamic model

Computations were carried out on a 4 km horizontal grid covering a complex area of the

Irish Sea and surrounding waters. The numerical domain consists of 110 and 147 cells in

x- and y-direction, respectively. In the vertical direction, a sigma coordinate system of 11

sigma level was used. Hence the total number of computational grid cells used in the

model is 177,870. For such grid spacings, stability conditions impose the following

external and internal timesteps respectively: Δt = 35 and ΔT = 385 seconds. The

minimum bottom roughness height of 5 mm was assumed. Bathymetric data for the

numerical model were obtained by digitising the Admiralty Chart number 1824A. The

bathymetry map with depths referred to the MWL is presented in Figure 4.

Two open ocean boundaries are specified: (a) along the west and (b) along the south

boundaries as shown in Figure 4. The discharge of freshwater rivers is also included in

the model. A total of 41 data sets representing monthly average discharges were used to

simulate the effect of freshwater input. The model is driven mainly by a variable water

elevation defined along the open boundaries, however, density distributions and

meteorological conditions are also very important forcing factors. The open sea

boundaries used an elevation condition with the tidal spectrum provided by MMG model

from their 5 constituent tidal (K1, O1, M2, N2 and S2) model of the North-East Atlantic.

Input files also stored climatological initial conditions for temperature and salinity as well

as monthly averages of salinity and temperature along open boundaries. These data were

extracted from two sources (1) LEVITUS94 global dataset (Da Silva et al., 1994) and


Page 19 of 67

interpolated from 1 degree resolution to Irish Sea model grid, and from (2) the state-of-

the-art global model, MPIOM, developed by the Max Planck Institute in Germany and

run by MMG.

Figure 4. Bathymetry map of the Irish Sea (depths in meters), positions of numerical

boundaries and locations of monitoring stations

The full set of meteorological conditions was obtained from the European Centre for

Medium-Range Weather Forecasts (ECMWF) from their regional reanalysis ERA-40

atmosphere dynamics model. Data on wind fields, air pressure and temperature,

humidity, cloud cover fraction, evaporation, precipitation, shortwave radiation and

extinction coefficient were applied to the POM model at 3-hour intervals.

Balbriggan

South boundary

West boundary

Cahore

Bull Island

Larne

Carlingford/Green

Sellafield

Dunmore

T1


Page 20 of 67

Wind stresses were parameterized as a quadratic function of the 10 m wind speed.

Changes in the roughness of ocean surface were included through varying drag

coefficient. The bottom stress term is calculated as a quadratic function of the depth mean

flow with frictional parameter of 0.0025.

3.3.2 Transport model

The Tc-99 transport model was initialized from Tc-99 concentrations measured in the

eastern and western Irish Sea over December 1993; this is presented as a contour plot in

Figure 2. The concentration map was interpolated onto the POM model grid (Figure 5).

Tc-99 is input to the model as a single point source loading from the location of Sellafield

in accordance with discharge dataset provided by RPII (Figure 3). The series of Tc-99

concentrations at the Sellafield grid point is shown in Figure 6. At the open boundaries

constant inflows concentrations of Tc-99 are assigned on flood tides representing

conditions outside the model domain. Along the western boundary it is assumed that

flood waters contain 8mBq/l Tc-99 while at the southern boundary it is assumed that

flood waters contain 0.4 mBq/l Tc-99. These values are based on data published in

McCubbin et al (2002).


Page 21 of 67

Figure 5. Initial conditions for transport model. (Tc-99 mBq/l)

0.0E+00

1.0E+03

2.0E+03

3.0E+03

4.0E+03

5.0E+03

6.0E+03

7.0E+03

8.0E+03

9.0E+03

1.0E+04

Nov-9

3

Mar-9

4

Jul-9

4

Nov-9

4

Mar-9

5

Jul-9

5

Nov-9

5

Mar-9

6

Jul-9

6

Nov-9

6

Mar-9

7

Jul-9

7

Nov-9

7

Mar-9

8

Jul-9

8

Nov-9

8

Mar-9

9

Jul-9

9

Nov-9

9

Mar-0

0

Jul-0

0

Nov-0

0

Tc-

99co

ncen

trat

ions

(mB

q/l)

Figure 6. Time series of Tc-99 concentration in the vicinity of Sellafield in the numerical

model.


Page 22 of 67

3.4 Model validation

3.4.1 Hydrodynamic model

The hydrodynamics of the Irish Sea is driven mainly by tides with the M2 and S2

constituents having greatest impact. Tides enter the region through both the St. George’s

and North Channels, with the two paths meeting along a line running westward from the

south of the Isle of Man, McKay & Pattenden (1993). Figure 7 compares modelled and

observed tidal amplitudes while patterns of tidal circulation are compared in Figure 8.

The annual residual circulation of the flow through the Irish Sea is northward, however,

as a result of wind and density gradients action the movement of water is variable with

southward net flow prevailing at times, even on the scale of months. It is known, that

southward flow results from the Western Irish Gyre (WIG) formed over summer months

west of the Isle of Man.

The gyre is an important phenomenon in the western region of the Irish Sea; it is

characterized by summer stratification due to a combination of persistent slack water and

water depths exceeding 100 m. A two-layer system develops typically between April and

October, the surface layer being 20-40 m thick and up to 7C warmer than the bottom

layer, Horsburgh (1999). The difference in temperature is demonstrated in Figure 9

presenting surface and near-bed temperatures in the region. A dome of cold, dense water,

composed of water left over from the previous winter, can be found beneath the

thermocline. Since the dome is static, the sloping density surfaces bounding it can only be

maintained in geostrophic balance by cyclonic surface layer flow, Hill et al. (1997).

Cyclonic residual flows may be of the order of 20 cm/s and are concentrated in jet-like


Page 23 of 67

cores at the base of the pycnocline and immediately above the flanks of a well-defined

dome of dense water. Since the bottom density gradients have been recognised as the

significant factor driving the baroclinic, near-surface flow, Hill et al. (1996), the gyre

becomes progressively faster due to persistent sharpening of the bottom fronts with

proceeding heating season. The dome develops quickly, over a matter of days, Hill et al.

(1997), in late spring, and breaks down equally quickly in early autumn, as a result of

cooling and strong wind events. The existence of the gyre and residual circulation can be

seen in Figure 10.

(a) (b)

Figure 7. Contour of amplitude of M2 tidal constituent in the Irish Sea (a) predicted by

POM, (b) derived from Hartnett (2002).


Page 24 of 67

(a)

(b)

Figure 8. Tidal circulation pattern in the region of the Irish Sea predicted by (a) POM, (b)

extracted from MacDowell (1997). Flows in (a) are presented in m/s, in (b) are in knots.


Page 25 of 67

0.0E+00

2.0E+00

4.0E+00

6.0E+00

8.0E+00

1.0E+01

1.2E+01

1.4E+01

1.6E+01

1.8E+01

2.0E+01

01/1

995

04/1

995

07/1

995

10/1

995

01/1

996

04/1

996

07/1

996

09/1

996

12/1

996

04/1

997

07/1

997

09/1

997

12/1

997

04/1

998

07/1

998

09/1

998

12/1

998

04/1

999

07/1

999

09/1

999

12/1

999

03/2

000

06/2

000

09/2

000

12/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

Surface temperatureBottom temperatureMeasurements

Figure 9. Surface and bottom temperature timetrace at point T1 in the western Irish Sea.

See Figure 4 for point location.

(a) (b)

Figure 10. Residual circulation of the surface layer (a) predicted by the POM model and

(b) similar plot reproduced from Horsburgh (1999).


Page 26 of 67

3.4.2 Transport model

Performance of the transport model was quantified on the basis of how well it reproduces

Tc-99 spatial distributions at an instance of time, and how well temporal variations are

predicted at a point. The validation of Tc-99 was possible due to a relatively large volume

of monitoring data collected by RPII and others. The data, although temporally sparse,

were considered to be generally of good quality (spatially) and the majority of records

were employed in the process of validation. Two comprehensive datasets used here are:

Balbriggan 01/01/1996 – 31/12/2001 - temporal analysis

Research survey vessel and inshore records – spatial analysis.

In the period of November 1993 to November 2005 there were in total 307 water samples

tested for the contamination of Tc-99. From the total amount, over half the samples were

collected during research vessel expeditions labeled CV, CIR and COR, the remaining

samples were taken at 5 coastline locations. The fixed-point sample can be used as an

indicator of extent and magnitude of temporal variations, while less frequent ship records

were used to provide information on spatial distribution.

For spatial validation of the model the following datasets were used: CIR 11/93, CIR

12/94, CIR 5/95, CIR 10/95, CIR 11/96 and COR 9b/98. These data were compiled from

various sources and converted by McCubbin (2002) to contour plots. Field records and

corresponding model outputs are compared in Figure 11.


Page 27 of 67

While the dataset for the eastern Irish Sea has sufficient horizontal resolution to describe

Tc-99 densities and gradients, data for the western part is sparser and hence yields a less

precise picture of Tc-99 distribution fields. The monitoring of other regions of the Irish

Sea is less informative; the North Channel and Firth of Clyde was partially surveyed

during some of cruises, the central Irish Sea has poor records, while St. George’s Channel

was not included in the monitoring programme. For this reason only regional validation

can be performed; however, regions of highest concentrations and gradients, and

therefore of highest interest from a modelling perspective, are sufficiently well

monitored. Figure 11 compares distributions of radionuclide in the Irish Sea at 5 times. In

general, modelled and observed concentrations have similar distributions, and density

gradients match well. Quantitatively, magnitudes are of the same order and prove mass

conservation within the model. The 50, 100, 200 and 500 mBq/l contours are displayed at

the same locations as field records. Even the COR 9b/98 data are quite well reproduced in

spite of the fact that the Tc-99 distributions and magnitudes are markedly dissimilar to

that observed on the preceding surveys. By and large material transport is quite well

predicted by the model, however, the 1998 circulation overpredicts the strength of

southward transport.


Page 28 of 67

(a) (b)

December 1994

May 1995

Figure 11. Contour plots of Tc-99 surface concentrations derived from (a) McCubbin

(2002) and (b) POM numerical model.


Page 29 of 67

(a) (b)

December 1995

December 1996


(2002) and (b) POM numerical model. (Cont’d)


Page 30 of 67

(a) (b)

September 1998


(2002) and (b) POM numerical model. (Cont’d)

The Balbriggan near-shore measurements were used to create historic time series. The

Tc-99 measurements at other stations (Cahore, Dunmore East, Carlingford and Bull

Island) are either too rare or do not coincide with a period of model run, and so were not

used in the validation process. Sampling at Balbriggan started in May 1995 and continued

regularly at approximately two-month intervals until the end of 2006. The records were

found to be generally suitable for the validation, and, therefore, numerical model results

were compared against this dataset.

Although, the Irish Sea is a complex system with seasonal variations in flow fields and

large fluctuations in the discharge, the model reproduces the transport of Tc-99


Page 31 of 67

surprisingly well, even at locations as far away from the source as Balbriggan. The

timeseries hindcast by the model and historic field records are compared in Figure 12.

Over the 5-year time-slice, the timing of peaks and strength of the model results are in

very good agreement with records. For the years 1996 and 1997 the model results are

particularly close to data illustrating substantial variations in concentrations. The

predicted peak by the model for June-July 1998 (43 mBq/l) was not recorded during

monitoring, this could be due to sparseness of monitoring, as the peak seems to result

from a strong southward circulation associated with the development of the WIG, which

transported a substantial amount of material that had accumulated in the Irish Sea since

summer 1997. This discrepancy will be subject to in-depth analysis in the next section.

Generally, low values recorded in 1999 resulting from much reduced loads through 1998

and 1999 are well-reproduced by the model. Despite some minor discrepancies between

model and records, the model is considered as well validated.

In the course of model calibration hundreds of simulations were undertaken, each with

modified settings to improve model predictions. However, there are still some constraints

on further improving model predictions:

The meteorological conditions are of high temporal resolution but crude

horizontal resolution. The model uses the same meteorological conditions at each

grid point throughout the domain. This problem will be analyzed further in next

section.

The model assumes a constant flux of Tc-99 at the open boundaries (8 mBq/l at

west boundary and 0.4 mBq/l at south). In the absence of more data this is the

only solution available for numerical computation.


Page 32 of 67

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)Numerical modelMeasurements

Figure 12. Historic and model timeseries at Balbriggan station.

3.5 Model sensitivity to meteorological conditions.

Meteorological conditions are often the main driver of water circulation in shallow

coastal waters. Brown and Gmitrowicz (1995) highlighted the importance of wind

strength on transport pathways within the Irish Sea. A second meteorological dataset was

acquired during this project to force the hydrodynamic model and assess the effects of

varying meteorological conditions on model output. The new dataset was extracted from

global reanalysis model run by the National Centre for Environmental Protection

(NCEP), US. The frequency of data and data point location were identical to those of the

previously used ECMWF reanalysis model. The performance of the numerical model

driven by the NCEP and ECMWF datasets is demonstrated in Figure 13. The


Page 33 of 67

considerable differences between the two outputs indicate the significant effects of the

meteorological conditions on water movement. The hindcast with NCEP data for years

1995-1997 shows very good agreement with data in respect to Tc-99 concentration, the

disagreement in years 1998-2000 is significant and leads to major overestimations of

southward flow and concentrations.

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

ECMWF Met modelNCEP Met modelMeasurments

Figure 13. Comparison of Tc-99 timeseries modelled using ECMWF and NCEP

meteorological data.

Model sensitivity and accuracy were further analyzed to assess the relative significance

of wind stresses and heat fluxes. Figure 14 (a) compares results of two simulations; (1) a

model forced with a fully developed wind and (2) with zero wind stress. The results show

a dramatic change in circulation and a significant reduction of Tc-99 concentration in


Page 34 of 67

Balbriggan. The temporal distribution of peaks is characteristic; peaks occur in

September and all have a similar profile of rapid increase and gentle decrease. This

repetitiveness is attributed to development of summer stratification and baroclinic

cyclonic circulation that transports material southward along Irish coastline and

ultimately reaches Balbriggan in September.

In the second case study of model sensitivity, the effect of heat fluxes was investigated. A

constant sea surface air temperature of 10 deg C was applied over the entire domain and

throughout the entire simulation period. The response was again quantified by comparing

differences in the concentration of Tc-99 at Balbriggan between the control simulation

and modified air temperature run. Figure 14 (b) clearly demonstrates that the air

temperature has a considerable impact on circulation and material transport. By lowering

summer air temperature all summer peaks occurring regularly between July and

September are greatly reduced; particularly, the peak of exceptionally warm summer of

1997 is underestimated. The low summer temperature leads to faint stratification and a

weak WIS gyre.

Two particular conclusions can be drawn from the tests: 1) wind has a prime effect on

circulation through modification of advective transport, dispersion and mixing of

material; 2) air temperature and solar radiation (not tested here) have a remarkable impact

on thermal stratification and therefore the density driven transport, which in turn is

responsible for higher peaks in Balbriggan.


Page 35 of 67

(a)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntra

tio

n(m

Bq

/l)Control runNo wind stressMeasurements

(b)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntra

tio

n(m

Bq

/l)

Control runConstant heat fluxMeasurements

Figure 14. Comparison of Tc-99 concentrations between control simulation and outputs

obtained from a model driven by meteorological data with (a) wind stresses excluded,

and (b) constant heat flux.


Page 36 of 67

3.6 Summary of Phase 2

Phase 2 of the project examined ability of the numerical model to reproduce measured

water concentrations of the Tc-99. Validation was conducted by comparing numerical

outputs against timeseries data at Balbriggan and distribution maps representing seasonal

fields. Although some discrepancies exist between model output and data, the model

reproduces circulation and transport pattern relatively correctly.

Also, sensitivity tests were carried out to assess model response to meteorological

conditions. Wind stress and temperature gradients are responsible for interannual and

seasonal variations in Tc-99 transport. Air temperature has greater effect during summer

months through its role in a development of density-driven baroclinic circulation.


Page 37 of 67

4. Phase 3 - Assessment of current monitoring program and interpretation of

monitoring data based on model outcomes

Phase 3 uses the model results and data to draw conclusions regarding the

appropriateness of RPII’s monitoring system. In the first two sections below the general

transport of Tc-99 is summarized based on monitoring data, numerical model and peer

reviewed publications. Relationships between discharge rate and peak concentration in

remote locations are investigated in Section 4.3. Travel times and residence times are

estimated in Section 4.4 and 4.5, respectively, while seasonal distribution pattern are

discussed in Section 4.6. Finally, evaluation of the radionuclide monitoring program is

conducted in Section 5. This section involves an assessment of the level of granularity in

space and time with which it is possible to predict concentrations in the western Irish Sea,

and the timescales over which variations in Tc-99 concentrations can be predicted.

4.1 General circulation of Tc-99 – spatial pattern

Temporal and spatial variations in the distributions of Tc-99 throughout the Irish Sea are

highly variable. Variations in concentrations between individual surveys indicate that

they are due to the combined effects of hydrographic variations, discharges and residual

concentrations.

The annual net flow within the Irish Sea is northward and transports Atlantic water

through St. George’s Channel to the west of the Isle of Man. A minor component of the


Page 38 of 67

flow enters the eastern Irish Sea to the north of Anglesey and moves anticlockwise round

the Isle of Man before rejoining the main flow to exit through the North Channel

(Howarth, 1984). However, as a result of variable meteorological conditions and/or

development of seasonal density gradients, the northward annual residual flow can be

disturbed on a scale of months giving a southward net flow. The southward current along

the east coast of Ireland is amplified by the inflow from the western side of the North

Channel (Brown and Gmitrowicz, 1995) and, within the western Irish Sea, by the

cyclonic circulation of the WIG persisting throughout summer months.

The transport pathways of Tc-99 as deduced from monitoring data confirm the general

circulation patterns mentioned above. Comparison of Tc-99 concentrations in the North

Channel and at the southern entrance to the Irish Sea indicates that the Sellafield material

predominantly migrates northward (McCubbin et al., 2002). The material released from

Sellafield travels either south-eastward along the coast of Cumbria during weak wind

days, or northwestward during storm days. Also, taking into account variability resulting

from freshwater discharges, the observed net advection of Tc-99 off Sellafield is in a

north-westerly direction along the southern Scottish coastline and later toward the

entrance of the North Channel.

During winter months the material is further transported northward to the Clyde Sea and

Inner Seas on the Scottish coast, from where it takes approximately 3 years to reach

Pentland Firth on the east coast of Scotland (Kautsky, 1985). During spring and summer

months, however, along with the formation of WIG, Tc-99 is advected southward. The

material is entrained into the gyre north-west of the Isle of Man through mixing processes


Page 39 of 67

with North Channel waters. The pattern of material transport from Sellafield to the east

coast of Ireland predicted by the model is in good agreement with field observations and

particularly with work of Jeffries (1982).

4.2 General circulation of Tc-99 – temporal pattern

The general circulation described in the previous section is well reflected in the historical

data recorded at Balbriggan; in particular, the existence of WIG is evident. Figure 15

presents previously shown field data and model results complemented by local sea

surface temperatures to prove strong relationship between Tc-99 peaks and the strength

of the WIG.

A plume of material originating in the North Channel travels along the Irish coastline for

approximately 4-8 weeks until it reaches Balbriggan typically between June and August.

This is manifested in a sudden increase of concentrations and is usually preceded by a 2-

month period of relatively constant values. This sudden increase at the initial stage may

be substantial; in the course of 2 weeks of August 1997 the concentration raised from

background 10 mBq/l to 65 mBq/l, bringing a mean daily increase rate to 4 mBq/l and

overall increase of 550% in this period. Also, significant increases of 25 mBq/l and 28

mBq/l were observed during August of 1996 and 1998, respectively, giving a daily

increase rate of 0.8 and 0.9 mBq/l. This is not surprising, since the rapid rise coincides

with a peak of sea surface temperature (July-August) when the WIS is strongest and

advection is fastest.


Page 40 of 67

The peak maximum is typically reached around 4-6 weeks after plume arrival and is

usually followed by a rapid decrease. Consequently, lowest levels are observed during

winter months (from December to February concentrations are lowest). The annual cycle

exhibiting seasonal oscillations is pretty well reproduced from one year to another.

The differences in annual peak heights result from variability in atmospheric fields,

fluctuations in discharge and accumulation from previous years. The effect of wind

action on longitudinal dispersion and mixing superimposed upon strong southward

density-driven advection resulting from exceptionally warm summer were the key factors

in the formation of the peak of September 1997. This mechanism was described in

section 3.5. Surprisingly, the effect of variable discharge rates has secondary effects on

the September 1997 peak height as discussed below.

One effect of discharge on the concentration variability at Balbriggan was quantified by

modelling a scenario with constant discharge. Figure 16 illustrates the effect of

fluctuations caused by a constant rate continuous release from Sellafield. The load

represents an average discharge from Sellafield over the period March 1995 to March

1996. As the case study discharge rate is similar to the real discharges the timeseries in

Figure 16 for real discharges and constant discharges are very similar. These outcomes

indicate that away from the immediate vicinity of Sellafield, the impact of the

fluctuations in the discharge is progressively less marked. Two most likely causes are: (1)

the residence time in the eastern Irish Sea is long enough to disperse individual Tc-99

pulses over the region before advecting it afield; (2) complex hydrographic transport


Page 41 of 67

pattern and long travel distances are conducive to material mixing and smooth out

concentration gradients.

Nonetheless, this simple test shows what concentrations would have been like from year

1998 onward if loads had not been reduced. The highest annual discharge of 192

TBq/year was recorded in 1995, subsequently a significant reduction in annual loads was

observed. The releases of 154 TBq/year, 84, 52, 68 and 44 TBq/year were reported

during consecutive years 1996-2000. If loads of 1995-1996 have been maintained over

following years, concentrations greatly exceeding the September 1997 would probably

have been observed.

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

994

04/1

994

07/1

994

10/1

994

01/1

995

04/1

995

07/1

995

10/1

995

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntra

tio

n(m

Bq

/l)

0.00E+00

2.00E+00

4.00E+00

6.00E+00

8.00E+00

1.00E+01

1.20E+01

1.40E+01

1.60E+01

1.80E+01

Control runMeasurementsSST

Figure 15. Historic and model timeseries at Balbriggan station along with corresponding

SST at the same location.


Page 42 of 67

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntra

tio

n(m

Bq

/l)

Control runConst. dischargeMeasurements

Figure 16. Comparison of Tc-99 timeseries modelled using actual discharges from

Sellafield (black) and constant rate discharges (green).

4.3 Comparisons of temporal patterns at four stations

Historical time series from four stations were collated and analyzed; these data were used

to assess spatial variations and travel times of Tc-99 material along east coast of Ireland.

The four stations located in order from north to south are: Larne, Co. Antrim, Greenore,

Co. Louth, Balbriggan, Co. Fingal and Cahore, Co. Wexford (see Figure 4 for exact

locations). Timetraces for Larne, Greenore and Cahore are presented in Figure 17 a-c

while Balbriggan timeseries is shown in Figure 12. There is a very strong relationship

between the Greenore timeseries and Balbriggan. The concentration levels are very

similar, likewise with arrival times; magnitudes are slightly higher for Greenore. Cahore

due to its remote distance from the source has very different distribution than the


Page 43 of 67

Balbriggan timeseries. The plume arrives approximately 3-4 months after Balbriggan

peaks. Peaks are 2-fold smaller for years 1996 and 1998-2000 and an order of magnitude

smaller for year 1997.

As expected, Larne station located on the northern tip of the Irish Sea displays much

different distribution than other stations. The peaks here are associated with northward

flow, and, therefore, highest peaks occur at times of lowest magnitude in the other three

locations dominated by the southward flow. The highest value is noted during winter

months of 1997/1998 and generally, higher magnitudes are observed in off-summer

months.

(a)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

Numerical model

Figure 17. Tc-99 time tracers for (a) Larne, (b) Greenore and (c) Cahore


Page 44 of 67

(b)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

Numerical modelMeasurements

(c)

0.0E+00

5.0E+00

1.0E+01

1.5E+01

2.0E+01

2.5E+01

3.0E+01

3.5E+01

4.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

nce

ntr

atio

n(m

Bq

/l)

Numerical modelMeasurments

Figure 17. Tc-99 time tracers for (a) Larne, (b) Greenore and (c) Cahore


Page 45 of 67

4.3 Relationships between discharge rate and peak concentration at Balbriggan

Prior to the commissioning of EARP, Tc-99 concentrations in a large proportion of the

Irish Sea were reasonably uniform (Leonard et al., 1997). The CIR11/93 survey (Figure

2) separates near-discharge region of concentrations c. 20 mBq/l from the rest of Irish

Sea (1-4 mBq/l). The concentrations detected during the CIR11/93 provide a baseline for

the analysis of Tc-99 concentrations in post-EARP period. By December 1996 surface

concentrations throughout the Irish Sea were elevated by more than an order of

magnitude, compared with pre-EARP levels (Leonard et al., 2004). Leonard et al. (1999)

suggested that concentrations have increased from the cumulative effect of continuous

discharge. Then, as a result of the greatly reduced discharges for the 6 months preceding

the September 1998 (COR9b/98) survey, Tc-99 concentrations in the vicinity to

Sellafield were reduced by more than an order of magnitude compared with those

observed during CIR 11/96 cruise.

In this section the relationships between discharge time and far field concentration are

investigated to examine the impact of transport pathways and material accumulation on

peak height and arrival timing. For this purpose, monthly discharges (from Figure 1)

were grouped into clusters as shown in Figure 18. Each cluster is a block of 3-5 monthly

discharges with details given in Table 2. The numerical model was next re-run where

individual clusters were included or excluded from simulations so their effect on far field

concentrations could be quantified.


Page 46 of 67

Figure 18. Clusters of monthly discharges.

Table 2. Cluster period and total mass of Tc-99 released during a period of one cluster.

Cluster name Period Total mass of Tc-99

TBq

Cluster A 04/1995-06/1995 49.4

Cluster B 07/1995-09/1995 48.8

Cluster C 10/1995-12/1995 48.6

Cluster D 01/1996-03/1996 47.8

Cluster E 04/1996-06/1996 34.6

Cluster F 07/1996-10/1996 68.6

Cluster G 11/1996-01/1997 27.7

Cluster H 02/1997-07/1997 56.3

0.0E+00

5.0E+03

1.0E+04

1.5E+04

2.0E+04

2.5E+04

3.0E+04

Nov-9

3

Mar-9

4

Jul-9

4

Nov-9

4

Mar-9

5

Jul-9

5

Nov-9

5

Mar-9

6

Jul-9

6

Nov-9

6

Mar-9

7

Jul-9

7

Nov-9

7

Mar-9

8

Jul-9

8

Nov-9

8

Mar-9

9

Jul-9

9

Nov-9

9

Mar-0

0

Jul-0

0

Nov-0

0

Mar-0

1

Jul-0

1

Nov-0

1

Tc-

99(G

Bq

/mo

nth

)

A C E F G HB D


Page 47 of 67

Figure 19 shows the results of eight runs, where single clusters of discharges were

excluded (zero discharge during months defined by a cluster). Three clusters in 1995 (A,

B, C) have very little effect on activity in Balbriggan during 1996 and literally no effect

on distributions in subsequent years. The winter/spring release D significantly contributes

to Tc-99 concentration during summer 1996 and influences spring concentrations of

1997. The (E) load generates similar effect to spring release, though overall

concentrations are higher as discharges in this cluster are generally lower. Both

summer/autumn (F) and winter (G) releases markedly affect concentrations during entire

1997 and 1998. The cluster (H) of longest duration but not the heaviest in terms of

discharges is responsible to a high degree for summer 1997 and 1998 peaks; its effect is

evident also in 1999 concentrations.


Page 48 of 67

(a) (b)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)

Numerical modelCluster runMeasurements

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)

Control runCluster runMeasurements

(c) (d)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)


(e) (f)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

ncen

trat

ion

(mB

q/l)


(g) (h)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


Figure 19. Tc-99 timeseries, control run (black) and run with a cluster of dischargesexcluded from simulation (green). These clusters are (a) cluster A, (b) cluster B, (c)cluster C, (d) cluster D, (e) cluster E, (f) cluster F, (g) cluster G, and (h) cluster H. Detailsof each cluster are given in Table 2


Page 49 of 67

A second test was carried out similar to the previous one, but in contrast to the test here,

releases were simulated for one cluster only. Results of this exercise are presented in

Figure 20. Once again 1995 discharges have small impact on 1996 concentrations

(Figures 20 a-c), while the 1996 releases affect concentrations in 1996 and in 1997

(Figure 20 d,e) and also to some extent in1998 (Figure 20 f,g). The highest peak of

September 1997 is due to the discharges of entire year 1996 and first half of 1997, with

the strongest contribution of summer/autumn 1996 discharge (F) and cluster of

discharges directly preceding the peak (H).

This exercise shows that the relationships between discharge time and far field

concentration are highly variable and depend not only on discharge rates and fluctuations

but also on sea dynamics that controls the transport and flushing of Tc-99 material. The

total mass of material of the 1995 discharge (highest of all annual records) that

accumulated over the year was effectively flushed out from the Irish Sea during winter

months, and therefore has insignificant impact on 1996 concentrations. In contrast to

1995, the material accumulated over 1996 within the eastern Irish Sea was transported

southward during autumn towards Liverpool Bay. This short-term variability (over time

scales of weeks) in circulation was driven by the prevailing meteorological conditions

and impeded the exit of material through the North Channel in subsequent months.


Page 50 of 67

(a) (b)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


(c) (d)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

nce

ntr

atio

n(m

Bq

/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-9

9co

nce

ntr

atio

n(m

Bq

/l)


(e) (f)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

09/1

996

12/1

996

04/1

997

07/1

997

09/1

997

12/1

997

04/1

998

07/1

998

09/1

998

12/1

998

04/1

999

07/1

999

09/1

999

12/1

999

03/2

000

06/2

000

09/2

000

12/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)

Numerical modelMeasurementsSeries3

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

09/1

996

12/1

996

04/1

997

07/1

997

09/1

997

12/1

997

04/1

998

07/1

998

09/1

998

12/1

998

04/1

999

07/1

999

09/1

999

12/1

999

03/2

000

06/2

000

09/2

000

12/2

000

Tc-9

9co

nce

ntr

atio

n(m

Bq

/l)

Numerical modelMeasurementsSeries3

(g) (h)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

01/1

996

04/1

996

07/1

996

10/1

996

01/1

997

04/1

997

07/1

997

10/1

997

01/1

998

04/1

998

07/1

998

10/1

998

01/1

999

04/1

999

07/1

999

10/1

999

01/2

000

04/2

000

07/2

000

10/2

000

Tc-

99co

ncen

trat

ion

(mB

q/l)


Figure 20. Tc-99 timeseries, control run (black) and run with a single cluster ofdischarges (green). These clusters are (a) cluster A, (b) cluster B, (c) cluster C, (d) clusterD, (e) cluster E, (f) cluster F, (g) cluster G, and (h) cluster H. Details of each cluster aregiven in Table 2


Page 51 of 67

This study undermines conclusions drawn from a monitoring system solely. For instance,

Smith et al. (2001) maintains that there is a relationship between peak discharges of 1995

and peak values of Tc-99 measured at Balbriggan occurring approximately two years

later. The numerical modelling analysis finds, however, that this is unlikely to be the

situation as demonstrated in Figure 19.

4.4 Travel time estimates

Travel time from Sellafield to the east coast of Ireland is dependent on variations in water

circulation dynamics within the Irish Sea and the discharge time. However, assuming that

the release rate is constant, differences result from variable hydrographics. The effect of

variable hydrodynamics on far field concentrations is investigated with some idealized

experiments. In this study travel time is calculated on the basis of assumed instantaneous

injection of Tc-99 material from the Sellafield plant. The material was released at a

constant rate over a period of 1 week. Since seasonal variability is likely to have an

impact on travel time, the releases were assumed taking place four times a year: 1st

January (winter), 1st April (spring), 1st July (summer) and 1st October (autumn).

Additionally, variability in interannual circulation was considered by introducing these

seasonal releases during the years 1995, 1996, 1997 and 1998. Thus, a total of 16

scenarios were simulated.

Travel time values, presented in Figure 21, exhibit high variability depending on both

season and year of release. Transport times from Sellafield to Balbriggan falls within the


Page 52 of 67

wide range of 30-240 days; with autumn releases resulting in the shortest travel times.

This is not surprising, since southward net flow accelerates in the course of summer with

the amplification of the WIG. The gyre’s retentive character is clearly manifested in

Figure 21 (a) and (b), where material arriving in late spring/early summer has been

entrapped there for 4-6 months.

The ultimate collapse of the gyre reverses drift from southward to northward, which

effectively intensifies over late autumn and winter as a result of prevailing wind

conditions. Northern drift prevents material released in January to be transported to

Balbriggan during winter and early spring. Consequently, the travel time of January

releases are the longest and take between 150-230 days to arrive in Balbriggan.

The travel time value does not reflect the amount of material transported. Spring

circulation patterns are effective in the fast transport of highly concentrated material and

retaining this material over long timescales. Summer currents, although ensuring short

transit times, carry generally much smaller quantities (exception is year 1995) in the first

year after release. In turn, the material returns again in the following year and the peak is

higher compared to the first year.

From the inter-annual analysis, significant differences in travel times between years

1995, 1996, 1997 and 1998 can be seen. The effect of long-term variability in

hydrodynamics on Tc-99 transport is greater than one could anticipate from monitoring

alone. The leading edge of material released in January of 1996 and 1998 will reach

Balbriggan approximately 60 days ahead of material that was released in January of 1995


Page 53 of 67

or 1997; similar patterns can be found for April releases. The timeseries for autumn

releases differ widely for individual years and for this reason are difficult to compare.

There are a limited number of studies published that estimate travel times for the Irish

Sea; moreover, these estimates are divergent.

Smith et al. (2000) suggest a travel time between Sellafield to Balbriggan of

around 3 years, whereas Smith et al. (2001) from a monitoring data deduced two-

year travel time.

Mitchell et al. (1987) found travel times to Dublin of 6-12 months, and proposed

an upper limit of 1.6 years.

Dabrowski and Hartnett (2008) provide for a seasonal variability in their

numerical model and based on a one-year analysis only come up with travel times

to Dublin for 100-200 days.

The simple test conducted here shows why estimates were so different.

In summary, taking into account the seasonal and inter-annual effect, the typical estimate

for the leading edge of plume would be within the range of 60-150, in conservative

approach from 40 to 210 days, and in extreme case from 30-240 days. In respect to peak

concentration, the following travel times are likely to be observed:

winter release 180-260 days

spring release 95-170 days

summer 80-240 days (higher peak is expected in following year)

autumn release 150-360 days.


Page 54 of 67

(a) (b)

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

0 120 240 360 480 600 720 840 960 1080

Time (days)

Tc-

99co

nce

ntr

atio

n(m

Bq/

l)

1995199619971998

0.0E+00

1.0E+01

2.0E+01

3.0E+01

4.0E+01

5.0E+01

6.0E+01

7.0E+01

8.0E+01

0 120 240 360 480 600 720 840 960 1080

Time (days)

Tc-

modelling the distribution of tc-99 along the east coast of ireland · 2013-07-22 · a.i. olbert...

Documents