- Email: [email protected]
Identifying optimal agricultural countermeasure strategies for a hypothetical contamination scenario using the strategy model
Journal of Environmental Radioactivity 83 (2005) 383e397 www.elsevier.com/locate/jenvrad
Identifying optimal agricultural countermeasure strategies for a hypothetical contamination scenario using the strategy model G. Cox a, N.A. Beresford b, B. Alvarez-Farizo c, D. Oughton d, Z. Kis e, K. Eged e, H. Thørring f, J. Hunt g, S. Wright b, C.L. Barnett b, J.M. Gil c, B.J. Howard b, N.M.J. Crout a,* a
Division of Agricultural and Environmental Science, School of Bioscience, University of Nottingham, University Park, Nottingham NG7 2RD, UK b Centre for Ecology and Hydrology e Lancaster, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster LA1 4AP, UK c Unidad de Economia Agaria Servico de Investigacion Agroalimentaria e DGA, APDO 727, Carretera de Montanana KM. 7, 50080 Zaragoza, Spain d Department for Chemistry and Biotechnology, Agricultural University of Norway, P.O. Box 5026, Drobakveien 1432 A˚s, Norway e GSF-Forschungszentrum fu¨r Umwelt und Gesundheit Ingolstaedter landstrasse 1, D-85764 Nueherberg, Germany f Environmental Protection Department, Norwegian Radiation Protection Authority, P.O. Box 55, N-1332 Østera˚s, Norway g Centre for the Study of Environmental Change, Lancaster University, Lancaster LA1 4YT, UK Received 19 November 2003; received in revised form 30 April 2004; accepted 7 May 2004 Available online 23 May 2005
Abstract A spatially implemented model designed to assist the identification of optimal countermeasure strategies for radioactively contaminated regions is described. Collective and individual ingestion doses for people within the affected area are estimated together with collective exported ingestion dose. A range of countermeasures are incorporated within the model, and environmental restrictions have been included as appropriate. The model evaluates * Corresponding author. Tel.: C44 115 951 6253; fax: C44 115 951 3251. E-mail address: [email protected] (N.M.J. Crout). 0265-931X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvrad.2004.05.021
384
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
the effectiveness of a given combination of countermeasures through a cost function which balances the benefit obtained through the reduction in dose with the cost of implementation. The optimal countermeasure strategy is the combination of individual countermeasures (and when and where they are implemented) which gives the lowest value of the cost function. The model outputs should not be considered as definitive solutions, rather as interactive inputs to the decision making process. As a demonstration the model has been applied to a hypothetical scenario in Cumbria (UK). This scenario considered a published nuclear power plant accident scenario with a total deposition of 1.7 ! 1014, 1.2 ! 1013, 2.8 ! 1010 and 5.3 ! 109 Bq for Cs-137, Sr-90, Pu-239/240 and Am-241, respectively. The model predicts that if no remediation measures were implemented the resulting collective dose would be approximately 36 000 person-Sv (predominantly from 137Cs) over a 10-year period post-deposition. The optimal countermeasure strategy is predicted to avert approximately 33 000 person-Sv at a cost of approximately £160 million. The optimal strategy comprises a mixture of ploughing, AFCF (ammonium-ferric hexacyano-ferrate) administration, potassium fertiliser application, clean feeding of livestock and food restrictions. The model recommends specific areas within the contaminated area and time periods where these measures should be implemented. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Model; Restoration; Optimisation; Countermeasures
1. Introduction Following a nuclear accident, rural environments may be contaminated for many years. To protect the population, effective restoration strategies must be developed. To do this, decision makers must consider many factors, such as the radioecological, environmental, economic, and social conditions in the affected area. However, nuclear emergency planning often focuses on short-term (dayseweeks) responses and does not consider longer term (monthseyears) restoration strategies. Ideally, these mid-long term strategies should be considered during the early phase after an incident, as decisions made then may have consequences for the future management of the affected area. There are a number of alternative countermeasures that could be applied following a release and these vary in their effectiveness, mode of operation and cost. Moreover, environmental, agricultural, social and economic conditions can vary across an affected region. Therefore, selecting the optimal combination of countermeasures and determining when and where they are best applied is a complex task. The combination of Geographical Information Systems (GIS), radioecological models, and information about the effects of various radiological countermeasures presents the possibility of computationally assessing the consequences of restoration strategies. The aim of such a model should be to help select actions that reduce doses from radiation to ‘as low as reasonably achievable’ (ICRP, 1973) and produce a net benefit (i.e. the difference between the monetary value of the averted dose and the costs incurred by the action (ICRP, 1988)), while recognising that in reality decision
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
385
makers will have to consider a wider range of social impacts of both contamination and the countermeasure strategy. In this paper we describe a model which is designed to assist decision makers in identifying optimal agricultural countermeasure strategies under user-defined conditions. To illustrate the use of the model we present results of a case study application to a hypothetical scenario in Cumbria (NW England).
2. Model overview The approach described here is fully described by Cox and Crout (2003) (available from www.strategy-ec.org.uk/output/outputs.htm) and is summarised here. The overall spatial representation of radionuclide transfer is an evolution of the model for radiocaesium described by Gillett et al. (2001). The area under study is divided into a two dimensional array of grid squares so that the spatial variation in model inputs and outputs across the region can be represented using a raster-based GIS approach. Each grid square is associated with a record within a database, giving data for radionuclide deposition levels, soil characteristics, topography, agricultural production (animal, arable, and horticultural) and population density. Previously published plant uptake models for Cs, Sr, Pu and Am (Absalom et al., 2001; Mu¨ller and Pro¨hl, 1993) together with spatially attributed agricultural data are used to estimate the activity concentrations of 27 common food products. Soil exchangeable K, pH, percentage clay and percentage organic matter content are used to predict radiocaesium transfer from soils to crops (Absalom et al., 2001); other radionuclides are modelled assuming established transfer factors (Mu¨ller and Pro¨hl, 1993). These models are coupled with information on dietary habits and food sources to calculate the resulting ingestion doses of the population within the region simulated. The collective ingestion dose due to the consumption of food products exported from the region and consumed elsewhere is also calculated. To represent the diverse dietary preferences and sources (e.g. local, regional, imported) of the region’s population, representative individuals are simulated by Monte Carlo sampling attributes from appropriate distributions. For example, this approach can be used to create 1000 simulated individuals in each grid square with differing dietary preferences and sources. The collective dose for the grid square is calculated by scaling the dose for the simulated individuals to the population of the grid square (i.e. by a factor of number of simulated individuals/grid square population). This approach is used in an attempt to more accurately reflect the variations that exist between individuals. It allows the distributions in individual dose to be estimated and integrated to predict collective dose of the population, rather than using mean values (of diet etc.) or defined subsets such as critical groups. Another advantage is that attributes which lead to high dose can be identified under the conditions of a specific scenario. A range of countermeasures are considered within the model, these are: shallow ploughing (20e30 cm) of pastures; deep ploughing (approximately 45 cm) of pastures, edible crops, silage crops; skim and burial ploughing of pastures, edible crops, silage
386
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
crops (Roed et al., 1996); application of potassium fertilisers and/or lime to pastures, edible and silage crops; administration of AFCF to food producing animals (as concentrates to dairy cows, pigs and poultry and as boli to sheep and beef cattle); clean feeding of food producing animals; and dietary advice. Any combination of these countermeasures can be activated within each grid square and their combined effect on ingestion dose is simulated. Environmental constraints (physical and regulatory) have been included within the implementation of countermeasures as appropriate. In addition to the above countermeasures, the model assumes that restrictions on the sale of contaminated food will be triggered at the recommended Council Food Intervention Limits (CFILs, CEC (1989)). The countermeasures selected, and the extent of their implementation, can be user-defined or automatically adjusted to identify an optimal combination of countermeasures for a given scenario. The optimisation of countermeasure strategies is performed by the minimisation of a cost function (C, monetary units) which incorporates the countermeasure implementation costs (I, monetary units) together with the monetary value of the averted dose (A, person-Sv). CZI aA
ð1Þ 1
where a (monetary units person-Sv ) is a coefficient representing the monetary value of averted dose. The outputs from countermeasure optimisation are the spatial and temporal extent of each countermeasure’s implementation that constitutes the optimal balance between expenditure and effective dose reduction. Optimisation is undertaken using Powell’s conjugate vector method (Press et al., 1986). This makes successive line minimisations along the axis of each adjustable parameter (in this case the extent of an individual countermeasure’s implementation). These are used to define the direction for subsequent line minimisations along a composite axis for improved computational efficiency. The method is iterative, and can be numerically intensive, especially if large number of parameters (i.e. countermeasures) is optimised. Given the numerically intensive nature of countermeasure optimization, the recommended approach when using the model is to individually assess each countermeasure to identify those that are likely to be of benefit within the given scenario, prior to optimisation. This allows a reduction in the number of countermeasures included within the optimisation. For each individual countermeasure, the value of the cost function is assessed at regular intervals between the maximum and the minimum extent of countermeasure implementation. If the cost function is negative, or close to zero, at any point, the countermeasure may be of benefit in the scenario, and is included for optimisation.
3. Data sources for Cumbrian scenario A grid square resolution of 25 km2 (5 ! 5 km) was used giving a total of 271 grid squares for the study area (the county of Cumbria).
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
387
A simple deposition pattern was created using a source term derived from Kelly and Clarke (1982) who described potential releases from a pressurised water reactor degraded core accident (Fig. 1). This provides a realistic, large-scale accident. However, it should be noted that this is deliberately hypothetical as this type of reactor site is not present in the study area. The total deposition for the radionuclides considered were 1.7 ! 1014, 1.2 ! 1013, 2.8 ! 1010 and 5.3 ! 109 Bq for Cs-137, Sr90, Pu-239/240 and Am-241, respectively. The spatial distribution of deposition was assumed to be the same for each radionuclide. Deposition was assumed to occur on the 1st May (i.e. in the early stages of the principal growing season). Soil properties (pH, exchangeable K, % clay and % organic matter) were derived from block kriging interpolation of measured soil properties from the Geochemical Atlas of England and Wales (McGrath and Loveland, 1992). The Geochemical Atlas data were sampled at the centre of 5 ! 5 km grid squares for England and Wales (5648 samples); semi-variogram models were fitted to soil properties for England and Wales and used to interpolate at a 5 ! 5 km resolution. The region has extensive upland areas characterised by soils with high organic matter contents and low nutrient status. To estimate agricultural production and practices (e.g. tilled areas), spatially attributed land cover data (Fuller et al., 1994) were combined with county level
Fig. 1. Assumed spatial distribution of radiocaesium deposition (Bq m2) across Cumbria (grid square resolution is 5 ! 5 km).
388
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
production statistics (MAFF, 1996), advised stocking rates and regional crop yield data (Nix, 2002). The predominant agriculture in Cumbria is animal production. Spatial data for population density in Cumbria was taken from the national census undertaken in 1991 (http://www.census.ac.uk/cdu/software/ surpop/). Countermeasure effectiveness together with resources required for implementation, wastes produced and potential environmental restrictions were taken from the countermeasure datasheets described by Howard et al. (2002) and Nisbet et al. (2004). As the practical implementation of most countermeasures considered is similar to routine agricultural practices (i.e. ploughing, fertilisation, animal feeding, spreading of milk to pasture) the cost of a contractor supplying these services (the service cost) was used; these values are collated for the UK annually (Nix, 2002). Current year ‘farm gate’ values of produce and stock were assumed to compensate farmers for food bans; service costs for animal housing and supplementary feeding, and pasture fertilisation and reseeding were used to compensate for loss of grazing as a consequence of ploughing. Where appropriate, the cost of waste disposal (based on the lowest cost option for the local situation) was included within the implementation cost of each countermeasure (e.g. food restrictions all have associated waste costs). Restrictions on the use of farm machinery for countermeasure implementation were imposed where the average slope in a given grid square was greater than 16 (22 grid squares) and deep ploughing was not allowed where the soil depth was less than 50 cm (6 grid square). Cumbria contains a national park, includes a number of environmentally protected (by law) areas and many farms participate within agri-environment schemes. The laws/agreements associated within these areas and schemes dictate that under normal circumstances many of the countermeasures could not be used. In this exercise, we have assumed that the ploughing of, or fertiliser application to, pastures within such areas would not be allowed; this affects countermeasure implementation in 45% of grid squares. The dietary preferences of the general UK population were assumed to be typical of the Cumbrian population, this is characterised by a high intake of dairy and meat products, and a low intake of fruit and vegetables (MAFF, 1990). Food is assumed to be derived from three geographical areas, ‘locally’ (the grid square), the region (production weighted mean activity concentration for the study area), or from outside the study area (uncontaminated). Most of the food products consumed by residents of Cumbria are assumed to be produced outside of the region, as the majority of people in the UK obtain their groceries from supermarkets, who source their products very widely. However, 25% of people were assumed to grow some of their own fruit and vegetables (Stewart et al., 1990), but, on average, this only constituted 10% of their total intake of this food group. For the other food product groups, the majority of the produce consumed was assumed to be derived from outside of the region, with small fractions being sourced from within the region. One percent of the population consumes a large proportion of meat and dairy products (w50 and w80%, respectively) from within the grid square where they live.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
389
The value of a in Eq. (1) was estimated using the approach of Lochard et al. (1996), i.e. g di aZabase di Od0 d0 aZabase
di %d0
where abase (£ person-Sv1) is the value of a when the individuals dose (di) is less than a threshold dose (d0) and g is an aversion coefficient. The value used for abase was 20 000 £ person-Sv1 as recommended for use in the UK (NRPB, 1993). The aversion coefficient was taken from an assessment of the value of the averted dose for public exposure in Hungary (Katona et al., 2003). The threshold dose of 1 mSv yr1 is the dose limit recommended by the ICRP for additional doses resulting from practices (ICRP, 1990). Although radioactively contaminated land does not constitute ‘‘a practice’’, many national authorities use values intended for practices as targets in remediation situations (Linsley, 2002). Above this level of individual dose, the cost of a unit of collective dose increases according to the value of the aversion coefficient. We should emphasise that although we are using standard values of a its estimation is difficult, especially if social and economic factors are to be taken into account. 4. Results The model was used to investigate 3 remediation options: ‘‘Do nothing’’. No countermeasures were implemented. ‘‘Food restrictions’’. Food restrictions were implemented at the recommended CFILs. No other countermeasures were implemented. ‘‘Optimal’’. The implementations of selected countermeasures were optimised as outlined earlier. Food restrictions were also implemented at the recommended CFILs as above. In each case, simulations were performed for a period of 0e10 years after the deposition event. This choice of period was based on the need to simulate the period during which most of the ingestion dose is accumulated balanced against computational requirements. 4.1. ‘‘Do nothing’’ The majority of the dose arising from food products produced within Cumbria is exported outside the region (Table 1). The relatively small local collective dose from ingestion is due to significant sourcing of the local population’s diet from outside of the region, where food products are assumed to be uncontaminated.
390
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Table 1 Collective doses and overall costs for the ‘‘Do nothing’’, and ‘‘food restrictions’’ and ‘‘optimal’’ management options over a period of 10 years post-deposition
Local collective ingestion dose (person-Sv) Exported collective ingestion dose (person-Sv) Averted dose (person-Sv) Implementation cost (million £) Cost function (million £)
Do nothing
Food restrictions
Optimal
245 35 873 e e e
96.2 3156 32 866 2298 C1630
39.5 2887 33 192 159 528
The majority of people receive doses of less than 7.5 mSv over the 10-year study period (Fig. 2). The highest individual doses over this time are approximately 75 mSv. The distinguishing features of people who receive this level of dose are that their rate of milk consumption is above average (213 L yr1 compared with a mean of 177 L yr1), and, more importantly, most (88%) of the milk that they consume is locally produced (i.e. from within their own grid square). The food products responsible for the majority of the local collective ingestion dose are potatoes, lamb, cow’s milk, cereals, fruit, and beef (Fig. 3). The contributions from potatoes and fruit are large, primarily because they are food products that are often home-grown. Cow milk also contributes significantly to the local collective ingestion dose as although the proportion of people sourcing their milk locally, or regionally, is comparatively low, the mean consumption rate is the highest of any food product. The large contribution of lamb and beef is attributable to the high proportion of these products produced in areas with high radiocaesium bioavailability (i.e. upland areas with nutrient poor, highly organic soil).
Fig. 2. Distribution of individual doses received by the population over a period of 10 years postdeposition.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Fig. 3. Contribution of
137
90
Cs and
391
Sr to the local collective ingestion dose by food product.
The exported collective ingestion dose is predominantly due to lamb (Fig. 4). This is due to the high production of lamb in Cumbria and high long-term bioavailability of radiocaesium in upland areas where the grazing of sheep is the dominant agricultural practice.
Fig. 4. Contribution of
137
Cs and
90
Sr to the exported collective ingestion dose by food product.
392
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Radiocaesium is the greatest contributor to dose (approximately 50 times that of radiostrontium, other nuclides are negligible). This is due to a combination of its high relative deposition and long-term bioavailability in Cumbrian agricultural systems, and comparatively low transfer of the other radionuclides to meat (the production of which is a predominant farming practice in the area). The annual collective dose falls with time, approximately 40% of the total collective dose occurs during the first year after deposition; years 2 and 10 contribute approximately 10 and 5%, respectively. 4.2. ‘‘Food restrictions’’ Implementation of CFILs is predicted to cost approximately £2000 million, and avert approximately 33 000 person-Sv over the 10-year study period (Table 1). Most of the averted dose would be due to a reduction in the exported collective ingestion dose. This would be reduced by a factor of approximately 12, compared with ‘‘do nothing’’, whilst the local collective ingestion dose would be reduced by a factor of approximately 2. The maximum individual dose is reduced from approximately 75 mSv to !30 mSv (Fig. 2). The value of the cost function is positive (Table 1) as the resources required outweigh the monetary value of the benefit (i.e. the averted dose). According to ICRP philosophy these measures are therefore unjustified (ICRP, 1988). The majority of the implementation cost is due to the long-term restriction of lamb and beef food products (Table 2), because, as outlined earlier, a significant proportion of lamb and beef in Cumbria is produced on land susceptible to longterm radiocaesium bioavailability. It is predicted that, after 10 years, the radiocaesium activity concentration of lamb would still be above the recommended CFIL in nearly 50% of the grid squares in Cumbria, whilst restrictions on the entry of beef into the food chain would still be required in approximately 40% of the grid squares. Of course it is unlikely that normal agricultural practice would continue in these areas for such an extended period (i.e. market pressures or regulation would probably force changes) so the associated predicted costs are probably an overestimate, although some form of long-term compensation would probably be required. However, predicting such changes is impracticable. Table 2 Cost of food restrictions by food product and waste volumes generated, under the ‘Food Restrictions’ option over a period of 10 years post-deposition Food product
Cost of restriction (million £)
Waste generated
Lamb Beef Cow’s milk Cereals Cow cheese Fruit Potatoes
1500 728 19 8.7 15.2 0.5 0.5
5.2 ! 106 animals 1.7 ! 105 animals 9.4 ! 107 L 2.8 ! 108 kg 7.5 ! 107 L milk 2.7 ! 106 kg 2.4 ! 107 kg
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
393
4.3. ‘‘Optimal’’ The countermeasures suggested for implementation by countermeasure optimisation, their associated costs, and extent are shown in Table 3. The focus of the optimal countermeasure combination is to reduce the activity concentrations in sheep and beef cows, which are major contributors to both the local and total collective ingestion dose. This is achieved by skim and burial ploughing of pasture in those grid squares where it is permitted and practical, and by the administration of clean feed and AFCF in the grid squares where ploughing is restricted (Fig. 5). We should emphasise that the optimal countermeasure strategy requires the various countermeasures to be implemented for different time periods at different locations within the simulated region. The temporal variation is illustrated in Fig. 6 which shows the proportion of the regions grid squares in which AFCF (to diary cows, beef cattle, and sheep) and clean feeding (to beef cattle and sheep) are implemented as a function of time. During the first 4e5 months after deposition the extent of these countermeasures broadly increases. This is because the animal’s activity concentrations fall sufficiently for these measures to be able to reduce activity concentrations below CFILs and consequently the requirement for food bans restrictions reduces. After this period there is a slow but gradual decline in the extent of the countermeasure implementation, although large parts of the region still remain affected 10 years after deposition. The reduction in local collective ingestion dose is 40% greater than by the imposition of CFILs alone, whilst the exported collected dose is reduced by a further 10% (Table 1). Moreover, the cost of implementing the optimal countermeasure strategy is !10% of using only food restrictions. The difference between the costs of ‘‘Food restrictions’’ and ‘‘Optimal’’ is largely due to the reduced amount of lamb and beef products that require restriction and disposal over the 10-year period. The effect of the optimal countermeasure combination on the distribution of individual doses received by the Cumbrian population is to further reduce the number of people receiving the highest doses (Fig. 2). Table 3 Implementation cost and, where applicable, extent of implemented countermeasures under ‘‘optimal’’ management over a period of 10 years post-deposition Countermeasure implemented
Cost (million £)
Logistical extent/waste produced
Skim and burial plough (Pasture) K fertiliser (Pasture) Food restrictions
41.1 34.1 36.0
Clean feed (Sheep) AFCF (Sheep) Deep plough (Silage crops) Clean feed (Beef cows) AFCF (Beef cows) AFCF (Dairy cows) Skim and burial plough (Edible crops)
20.7 11.9 8.0 2.0 2.0 1.4 1.4
116 000 ha 125 000 ha 48 000 t milk 64 000 animals 291 000 ha edible crops 2.5 million sheep 2.9 million sheep 133 000 ha 38 000 beef cows 314 000 cows 111 000 cow months 11 000 ha
394
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Fig. 5. Spatial extent of skim and burial ploughing of pasture (left), clean feeding of sheep 12 months after deposition(centre) and AFCF administration to sheep 12 months after deposition (right). Dark areas indicate where the countermeasure is being applied.
5. Discussion Countermeasure optimisation suggests a set of countermeasures which are an improvement over the sole use of food restrictions; the radiation protection outcome is similar but is achieved at a much lower cost. The countermeasures selected are focussed on reducing the activity concentrations of lamb and beef in particular. These results suggest that ploughing is the agricultural countermeasure of first choice. It is relatively cheap and effective, and in the simulations presented here it is AFCF (Dairy cows) AFCF (Cattle) AFCF (Sheep) Clean feed (Cattle) Clean feed (Sheep)
70
% of Grid Squares
60 50 40 30 20 10
1
2
m 0 on m th on t 3 m hs on t 4 m hs on t 5 m hs o 6 nth m s o 7 nth m s o 8 nth m s o 9 nth m s 10 ont m hs 11 ont m hs 12 ont m hs on t 2 hs ye ar s 3 ye a 4 rs ye a 5 rs ye a 6 rs ye a 7 rs ye a 8 rs ye a 9 rs ye 10 ars ye ar s
0
Time after deposition Fig. 6. Temporal variation in the regional extent of five animal based countermeasures. The vertical axis is the percent of the region’s grid squares where the countermeasure is conducted. Note: The horizontal time axis is not linear, the first year is expanded to illustrate early changes in countermeasure implementation.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
395
selected wherever it is allowed. However, the model takes no account of the logistical requirements of widespread ploughing. Therefore, implementation time may be longer in reality than that assumed in the model. This may influence the effectiveness of the countermeasure. Also the spatial resolution of 5 ! 5 km may have led to an overestimate of the areas which can be ploughed. For example, within the complex topography of Cumbria there will be localised areas of high slope unsuitable for ploughing which are lost when a mean slope is calculated for the grid square as a whole. Where ploughing is not allowed AFCF administration and clean feeding are selected, generally in combination with one another. For example, AFCF can be used to reduce the amount of clean feeding required to lower an animal product to its CFIL; AFCF being administered for a period prior to clean feeding. This does not affect the radiological outcome but does reduce the cost. Much of the benefit, in terms of dose reduction, of the suggested countermeasure is for people living outside of the region, whose individual doses are likely to be negligible. In part this is because any agricultural countermeasure that is implemented will reduce more exported than local dose, simply because more food products are exported from the region than can be consumed locally. It would be possible to extend the analysis to consider external doses and corresponding relevant countermeasures. In this case, any benefits would obviously accrue directly to the population of the contaminated areas. Of course the model presented has many limitations. Of these, perhaps the most noteworthy is the exclusion of some key radionuclides, especially 131I. Although this was outside the scope of our work the model could be extended to consider 131I. Clearly, this isotope has considerable potential for short-term food contamination. In the example we have presented we could anticipate that it would increase the requirement for short-term food restrictions/countermeasures. However, such measures were already recommended for many areas within the contaminated region due to radiocaesium and radiostrontium contamination. Therefore, the inclusion of 131I would probably force adjustments to the optimal strategy rather than its complete revision. If such models were to be considered for operational use then clearly the inclusion of 131I would be required. The model application presented demonstrates the importance of relevant local information in planning effective remediation. Recommendations that a combination of ploughing, AFCF administration, clean feeding and food restrictions are the essential ingredients of a remediation strategy will not be a surprise to experts in the field. However, the model has the potential to identify the optimal combination of these measures, in both space and time, taking into account the implementation costs. This is a high-dimension problem which is very difficult to visualise in a form that makes it manageable without computational tools such as the model presented. Therefore, the model has the potential to form a useful part of the decision making process. It is not intended to provide definitive answers; rather, it allows decision makers to investigate the effects of various options, and to provide information about the likely consequences of countermeasure combinations. The model provides a starting point for the design of a restoration strategy by highlighting the major sources of dose to affected populations, together with recommendations on when,
396
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
where and in what combinations countermeasures should be employed. To our knowledge the approach described is the first to successfully implement a countermeasure strategy optimisation within spatially predictive models. Acknowledgements The STRATEGY project was conducted under a contract (FIKR-CT-2000-00018) within the research and training programme (Euratom) in the field of nuclear energy, and this support is gratefully acknowledged. The paper is the sole responsibility of the authors and does not reflect Community opinion, and the Community is not responsible for any use that might be made of data appearing in this publication. We also wish to acknowledge the contribution of all the STRATEGY project members and the comments of an anonymous reviewer which improved this paper considerably. References Absalom, J.P., Young, S.D., Crout, N.M.J., Sanchez, A., Wright, S.M., Smolders, E., Nisbet, A.F., Gillett, A.G., 2001. Predicting the transfer of radiocaesium from organic soils to plants using soil characteristics. J. Environ. Radioact. 52, 31e43. CEC, 1989. Council Regulation (Euratom) No. 2218/89 amending Regulation (Euratom) No. 3954/87 laying down maximum permitted levels of radioactive contamination of foodstuffs and feedingstuffs following a nuclear accident or any other case of radiological emergency. Official Journal of the European Communities L211/1. Cox, G.M., Crout, N.M.J., 2003. A methodology for the selection and optimisation of countermeasure strategies. Report for the STRATEGY project (CEC Contract N : FIKR-CT 2000e00018). University of Nottingham, UK. Gillett, A.G., Crout, N.M.J., Absalom, J.P., Wright, S.M., Young, S.D., Howard, B.J., Barnett, C.L., McGrath, S.P., Beresford, N.A., Voigt, G., 2001. Temporal and spatial prediction of radiocaesium transfer to food products. Rad. Environ. Biophys. 40, 227e235. Howard, B.J., Andersson, K.G., Beresford, N.A., Crout, N.M.J., Gil, J.M., Hunt, J., Liland, A., Nisbet, A., Oughton, D., Voigt, G., 2002. Sustainable restoration and long-term management of contaminated rural, urban and industrial ecosystems. Radioprotection Colloques 37 (C1), 1067e1072. Fuller, R.M., Groom, G.B., Jones, A.R., 1994. The land cover map of great britain: an automated classification of landsat thematic mapper data. Photogramm. Eng. Remote Sens. 60, 553e562. International Commission on Radiological Protection (ICRP), 1973. Implications of commission recommendations that doses be kept as low as readily achievable. ICRP 22. Pergamon, Oxford. International Commission on Radiological Protection (ICRP), 1988. Optimisation and decision-making in radiological protection. ICRP 55. Pergamon, Oxford. International Commission on Radiological Protection (ICRP), 1990. Recommendations of the ICRP. ICRP 60. Pergamon, Oxford. Katona, T., Kanya´r, B., Eged, K., Kis, Z., Ne´nyei, A´., Bodna´r, R., 2003. The monetary value of the averted dose for public exposure assessed by the willingness to pay. Health Phys. 84, 594e598. Kelly, G.N., Clarke, R.H., 1982. An assessment of the radiological consequences of releases from degraded core accidents for Sizewell PWR. NRPB-R137. National Radiological Protection Board, Didcot. Linsley, G., 2002. Summary of the IAEA Arlington symposium on the restoration of environments with radioactive residues. In: Radiation Legacy of the 20th Century: Environmental Restoration. IAEATECDOC-1280. International Atomic Energy Agency, Vienna, pp. 30e34. Lochard, J., Lefaure, C., Scheiber, C., Schneider, T., 1996. A model for the determination of monetary values of the man-sievert. J. Radiol. Prot. 16, 201e204.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
397
Ministry for Agriculture, Fisheries, and Food (MAFF), 1990. The dietary and nutritional survey of British adults. SS1241, HMSO, London. Ministry of Agriculture, Fisheries and Food (MAFF), 1996. The digest of agricultural census statistics, United Kingdom. The Stationary Office, London. ISBN 0 11 243039 2. McGrath, S.P., Loveland, P.J., 1992. Geochemical Atlas of England and Wales. Blackie Academic & Professional, London. Mu¨ller, H., Pro¨hl, G., 1993. Ecosys-87: dynamic model for assessing radiological consequences of nuclear accidents. Health Phys. 64, 232e252. Nisbet, A.F., Mercer, J.A., Hesketh, N., Liland, A., Thørring, H., Bergan, T., Beresford, N.A., Howard, B.J., Hunt, J., Oughton, D., 2004. Datasheets on countermeasures and waste disposal options for the management of food production systems contaminated following a nuclear accident. NRPBW58. National Radiological Protection Board, Didcot. Nix, J., 2002. Farm Management Pocketbook, thirty-second ed. Imperial College at Wye, Melton Mowbray. ISBN 0-9541201-0-8. National Radiological Protection Board (NRPB), 1993. Occupational, Public, and Medical Exposure. Documents of the NRPB 4 (2). NRPB, Didcot. Press, W.H., Flannery, B.P., Teukolsky, S.A., Vettering, W.T., 1986. Numerical Recipes. Cambridge University Press, Cambridge. Roed, J., Andersson, K.G., Prip, H., 1996. The skim and burial plough: a new implement for reclamation of radioactively contaminated land. J. Environ. Radioact. 33 (2), 117e128. Stewart, T.H., Fulker, M.J., Jones, S.R., 1990. A survey of habits of people living close to the Sellafield nuclear reprocessing plant. J. Radiol. Prot. 10, 115e122.
Identifying optimal agricultural countermeasure strategies for a hypothetical contamination scenario using the strategy model G. Cox a, N.A. Beresford b, B. Alvarez-Farizo c, D. Oughton d, Z. Kis e, K. Eged e, H. Thørring f, J. Hunt g, S. Wright b, C.L. Barnett b, J.M. Gil c, B.J. Howard b, N.M.J. Crout a,* a
Division of Agricultural and Environmental Science, School of Bioscience, University of Nottingham, University Park, Nottingham NG7 2RD, UK b Centre for Ecology and Hydrology e Lancaster, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster LA1 4AP, UK c Unidad de Economia Agaria Servico de Investigacion Agroalimentaria e DGA, APDO 727, Carretera de Montanana KM. 7, 50080 Zaragoza, Spain d Department for Chemistry and Biotechnology, Agricultural University of Norway, P.O. Box 5026, Drobakveien 1432 A˚s, Norway e GSF-Forschungszentrum fu¨r Umwelt und Gesundheit Ingolstaedter landstrasse 1, D-85764 Nueherberg, Germany f Environmental Protection Department, Norwegian Radiation Protection Authority, P.O. Box 55, N-1332 Østera˚s, Norway g Centre for the Study of Environmental Change, Lancaster University, Lancaster LA1 4YT, UK Received 19 November 2003; received in revised form 30 April 2004; accepted 7 May 2004 Available online 23 May 2005
Abstract A spatially implemented model designed to assist the identification of optimal countermeasure strategies for radioactively contaminated regions is described. Collective and individual ingestion doses for people within the affected area are estimated together with collective exported ingestion dose. A range of countermeasures are incorporated within the model, and environmental restrictions have been included as appropriate. The model evaluates * Corresponding author. Tel.: C44 115 951 6253; fax: C44 115 951 3251. E-mail address: [email protected] (N.M.J. Crout). 0265-931X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvrad.2004.05.021
384
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
the effectiveness of a given combination of countermeasures through a cost function which balances the benefit obtained through the reduction in dose with the cost of implementation. The optimal countermeasure strategy is the combination of individual countermeasures (and when and where they are implemented) which gives the lowest value of the cost function. The model outputs should not be considered as definitive solutions, rather as interactive inputs to the decision making process. As a demonstration the model has been applied to a hypothetical scenario in Cumbria (UK). This scenario considered a published nuclear power plant accident scenario with a total deposition of 1.7 ! 1014, 1.2 ! 1013, 2.8 ! 1010 and 5.3 ! 109 Bq for Cs-137, Sr-90, Pu-239/240 and Am-241, respectively. The model predicts that if no remediation measures were implemented the resulting collective dose would be approximately 36 000 person-Sv (predominantly from 137Cs) over a 10-year period post-deposition. The optimal countermeasure strategy is predicted to avert approximately 33 000 person-Sv at a cost of approximately £160 million. The optimal strategy comprises a mixture of ploughing, AFCF (ammonium-ferric hexacyano-ferrate) administration, potassium fertiliser application, clean feeding of livestock and food restrictions. The model recommends specific areas within the contaminated area and time periods where these measures should be implemented. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Model; Restoration; Optimisation; Countermeasures
1. Introduction Following a nuclear accident, rural environments may be contaminated for many years. To protect the population, effective restoration strategies must be developed. To do this, decision makers must consider many factors, such as the radioecological, environmental, economic, and social conditions in the affected area. However, nuclear emergency planning often focuses on short-term (dayseweeks) responses and does not consider longer term (monthseyears) restoration strategies. Ideally, these mid-long term strategies should be considered during the early phase after an incident, as decisions made then may have consequences for the future management of the affected area. There are a number of alternative countermeasures that could be applied following a release and these vary in their effectiveness, mode of operation and cost. Moreover, environmental, agricultural, social and economic conditions can vary across an affected region. Therefore, selecting the optimal combination of countermeasures and determining when and where they are best applied is a complex task. The combination of Geographical Information Systems (GIS), radioecological models, and information about the effects of various radiological countermeasures presents the possibility of computationally assessing the consequences of restoration strategies. The aim of such a model should be to help select actions that reduce doses from radiation to ‘as low as reasonably achievable’ (ICRP, 1973) and produce a net benefit (i.e. the difference between the monetary value of the averted dose and the costs incurred by the action (ICRP, 1988)), while recognising that in reality decision
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
385
makers will have to consider a wider range of social impacts of both contamination and the countermeasure strategy. In this paper we describe a model which is designed to assist decision makers in identifying optimal agricultural countermeasure strategies under user-defined conditions. To illustrate the use of the model we present results of a case study application to a hypothetical scenario in Cumbria (NW England).
2. Model overview The approach described here is fully described by Cox and Crout (2003) (available from www.strategy-ec.org.uk/output/outputs.htm) and is summarised here. The overall spatial representation of radionuclide transfer is an evolution of the model for radiocaesium described by Gillett et al. (2001). The area under study is divided into a two dimensional array of grid squares so that the spatial variation in model inputs and outputs across the region can be represented using a raster-based GIS approach. Each grid square is associated with a record within a database, giving data for radionuclide deposition levels, soil characteristics, topography, agricultural production (animal, arable, and horticultural) and population density. Previously published plant uptake models for Cs, Sr, Pu and Am (Absalom et al., 2001; Mu¨ller and Pro¨hl, 1993) together with spatially attributed agricultural data are used to estimate the activity concentrations of 27 common food products. Soil exchangeable K, pH, percentage clay and percentage organic matter content are used to predict radiocaesium transfer from soils to crops (Absalom et al., 2001); other radionuclides are modelled assuming established transfer factors (Mu¨ller and Pro¨hl, 1993). These models are coupled with information on dietary habits and food sources to calculate the resulting ingestion doses of the population within the region simulated. The collective ingestion dose due to the consumption of food products exported from the region and consumed elsewhere is also calculated. To represent the diverse dietary preferences and sources (e.g. local, regional, imported) of the region’s population, representative individuals are simulated by Monte Carlo sampling attributes from appropriate distributions. For example, this approach can be used to create 1000 simulated individuals in each grid square with differing dietary preferences and sources. The collective dose for the grid square is calculated by scaling the dose for the simulated individuals to the population of the grid square (i.e. by a factor of number of simulated individuals/grid square population). This approach is used in an attempt to more accurately reflect the variations that exist between individuals. It allows the distributions in individual dose to be estimated and integrated to predict collective dose of the population, rather than using mean values (of diet etc.) or defined subsets such as critical groups. Another advantage is that attributes which lead to high dose can be identified under the conditions of a specific scenario. A range of countermeasures are considered within the model, these are: shallow ploughing (20e30 cm) of pastures; deep ploughing (approximately 45 cm) of pastures, edible crops, silage crops; skim and burial ploughing of pastures, edible crops, silage
386
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
crops (Roed et al., 1996); application of potassium fertilisers and/or lime to pastures, edible and silage crops; administration of AFCF to food producing animals (as concentrates to dairy cows, pigs and poultry and as boli to sheep and beef cattle); clean feeding of food producing animals; and dietary advice. Any combination of these countermeasures can be activated within each grid square and their combined effect on ingestion dose is simulated. Environmental constraints (physical and regulatory) have been included within the implementation of countermeasures as appropriate. In addition to the above countermeasures, the model assumes that restrictions on the sale of contaminated food will be triggered at the recommended Council Food Intervention Limits (CFILs, CEC (1989)). The countermeasures selected, and the extent of their implementation, can be user-defined or automatically adjusted to identify an optimal combination of countermeasures for a given scenario. The optimisation of countermeasure strategies is performed by the minimisation of a cost function (C, monetary units) which incorporates the countermeasure implementation costs (I, monetary units) together with the monetary value of the averted dose (A, person-Sv). CZI aA
ð1Þ 1
where a (monetary units person-Sv ) is a coefficient representing the monetary value of averted dose. The outputs from countermeasure optimisation are the spatial and temporal extent of each countermeasure’s implementation that constitutes the optimal balance between expenditure and effective dose reduction. Optimisation is undertaken using Powell’s conjugate vector method (Press et al., 1986). This makes successive line minimisations along the axis of each adjustable parameter (in this case the extent of an individual countermeasure’s implementation). These are used to define the direction for subsequent line minimisations along a composite axis for improved computational efficiency. The method is iterative, and can be numerically intensive, especially if large number of parameters (i.e. countermeasures) is optimised. Given the numerically intensive nature of countermeasure optimization, the recommended approach when using the model is to individually assess each countermeasure to identify those that are likely to be of benefit within the given scenario, prior to optimisation. This allows a reduction in the number of countermeasures included within the optimisation. For each individual countermeasure, the value of the cost function is assessed at regular intervals between the maximum and the minimum extent of countermeasure implementation. If the cost function is negative, or close to zero, at any point, the countermeasure may be of benefit in the scenario, and is included for optimisation.
3. Data sources for Cumbrian scenario A grid square resolution of 25 km2 (5 ! 5 km) was used giving a total of 271 grid squares for the study area (the county of Cumbria).
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
387
A simple deposition pattern was created using a source term derived from Kelly and Clarke (1982) who described potential releases from a pressurised water reactor degraded core accident (Fig. 1). This provides a realistic, large-scale accident. However, it should be noted that this is deliberately hypothetical as this type of reactor site is not present in the study area. The total deposition for the radionuclides considered were 1.7 ! 1014, 1.2 ! 1013, 2.8 ! 1010 and 5.3 ! 109 Bq for Cs-137, Sr90, Pu-239/240 and Am-241, respectively. The spatial distribution of deposition was assumed to be the same for each radionuclide. Deposition was assumed to occur on the 1st May (i.e. in the early stages of the principal growing season). Soil properties (pH, exchangeable K, % clay and % organic matter) were derived from block kriging interpolation of measured soil properties from the Geochemical Atlas of England and Wales (McGrath and Loveland, 1992). The Geochemical Atlas data were sampled at the centre of 5 ! 5 km grid squares for England and Wales (5648 samples); semi-variogram models were fitted to soil properties for England and Wales and used to interpolate at a 5 ! 5 km resolution. The region has extensive upland areas characterised by soils with high organic matter contents and low nutrient status. To estimate agricultural production and practices (e.g. tilled areas), spatially attributed land cover data (Fuller et al., 1994) were combined with county level
Fig. 1. Assumed spatial distribution of radiocaesium deposition (Bq m2) across Cumbria (grid square resolution is 5 ! 5 km).
388
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
production statistics (MAFF, 1996), advised stocking rates and regional crop yield data (Nix, 2002). The predominant agriculture in Cumbria is animal production. Spatial data for population density in Cumbria was taken from the national census undertaken in 1991 (http://www.census.ac.uk/cdu/software/ surpop/). Countermeasure effectiveness together with resources required for implementation, wastes produced and potential environmental restrictions were taken from the countermeasure datasheets described by Howard et al. (2002) and Nisbet et al. (2004). As the practical implementation of most countermeasures considered is similar to routine agricultural practices (i.e. ploughing, fertilisation, animal feeding, spreading of milk to pasture) the cost of a contractor supplying these services (the service cost) was used; these values are collated for the UK annually (Nix, 2002). Current year ‘farm gate’ values of produce and stock were assumed to compensate farmers for food bans; service costs for animal housing and supplementary feeding, and pasture fertilisation and reseeding were used to compensate for loss of grazing as a consequence of ploughing. Where appropriate, the cost of waste disposal (based on the lowest cost option for the local situation) was included within the implementation cost of each countermeasure (e.g. food restrictions all have associated waste costs). Restrictions on the use of farm machinery for countermeasure implementation were imposed where the average slope in a given grid square was greater than 16 (22 grid squares) and deep ploughing was not allowed where the soil depth was less than 50 cm (6 grid square). Cumbria contains a national park, includes a number of environmentally protected (by law) areas and many farms participate within agri-environment schemes. The laws/agreements associated within these areas and schemes dictate that under normal circumstances many of the countermeasures could not be used. In this exercise, we have assumed that the ploughing of, or fertiliser application to, pastures within such areas would not be allowed; this affects countermeasure implementation in 45% of grid squares. The dietary preferences of the general UK population were assumed to be typical of the Cumbrian population, this is characterised by a high intake of dairy and meat products, and a low intake of fruit and vegetables (MAFF, 1990). Food is assumed to be derived from three geographical areas, ‘locally’ (the grid square), the region (production weighted mean activity concentration for the study area), or from outside the study area (uncontaminated). Most of the food products consumed by residents of Cumbria are assumed to be produced outside of the region, as the majority of people in the UK obtain their groceries from supermarkets, who source their products very widely. However, 25% of people were assumed to grow some of their own fruit and vegetables (Stewart et al., 1990), but, on average, this only constituted 10% of their total intake of this food group. For the other food product groups, the majority of the produce consumed was assumed to be derived from outside of the region, with small fractions being sourced from within the region. One percent of the population consumes a large proportion of meat and dairy products (w50 and w80%, respectively) from within the grid square where they live.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
389
The value of a in Eq. (1) was estimated using the approach of Lochard et al. (1996), i.e. g di aZabase di Od0 d0 aZabase
di %d0
where abase (£ person-Sv1) is the value of a when the individuals dose (di) is less than a threshold dose (d0) and g is an aversion coefficient. The value used for abase was 20 000 £ person-Sv1 as recommended for use in the UK (NRPB, 1993). The aversion coefficient was taken from an assessment of the value of the averted dose for public exposure in Hungary (Katona et al., 2003). The threshold dose of 1 mSv yr1 is the dose limit recommended by the ICRP for additional doses resulting from practices (ICRP, 1990). Although radioactively contaminated land does not constitute ‘‘a practice’’, many national authorities use values intended for practices as targets in remediation situations (Linsley, 2002). Above this level of individual dose, the cost of a unit of collective dose increases according to the value of the aversion coefficient. We should emphasise that although we are using standard values of a its estimation is difficult, especially if social and economic factors are to be taken into account. 4. Results The model was used to investigate 3 remediation options: ‘‘Do nothing’’. No countermeasures were implemented. ‘‘Food restrictions’’. Food restrictions were implemented at the recommended CFILs. No other countermeasures were implemented. ‘‘Optimal’’. The implementations of selected countermeasures were optimised as outlined earlier. Food restrictions were also implemented at the recommended CFILs as above. In each case, simulations were performed for a period of 0e10 years after the deposition event. This choice of period was based on the need to simulate the period during which most of the ingestion dose is accumulated balanced against computational requirements. 4.1. ‘‘Do nothing’’ The majority of the dose arising from food products produced within Cumbria is exported outside the region (Table 1). The relatively small local collective dose from ingestion is due to significant sourcing of the local population’s diet from outside of the region, where food products are assumed to be uncontaminated.
390
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Table 1 Collective doses and overall costs for the ‘‘Do nothing’’, and ‘‘food restrictions’’ and ‘‘optimal’’ management options over a period of 10 years post-deposition
Local collective ingestion dose (person-Sv) Exported collective ingestion dose (person-Sv) Averted dose (person-Sv) Implementation cost (million £) Cost function (million £)
Do nothing
Food restrictions
Optimal
245 35 873 e e e
96.2 3156 32 866 2298 C1630
39.5 2887 33 192 159 528
The majority of people receive doses of less than 7.5 mSv over the 10-year study period (Fig. 2). The highest individual doses over this time are approximately 75 mSv. The distinguishing features of people who receive this level of dose are that their rate of milk consumption is above average (213 L yr1 compared with a mean of 177 L yr1), and, more importantly, most (88%) of the milk that they consume is locally produced (i.e. from within their own grid square). The food products responsible for the majority of the local collective ingestion dose are potatoes, lamb, cow’s milk, cereals, fruit, and beef (Fig. 3). The contributions from potatoes and fruit are large, primarily because they are food products that are often home-grown. Cow milk also contributes significantly to the local collective ingestion dose as although the proportion of people sourcing their milk locally, or regionally, is comparatively low, the mean consumption rate is the highest of any food product. The large contribution of lamb and beef is attributable to the high proportion of these products produced in areas with high radiocaesium bioavailability (i.e. upland areas with nutrient poor, highly organic soil).
Fig. 2. Distribution of individual doses received by the population over a period of 10 years postdeposition.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Fig. 3. Contribution of
137
90
Cs and
391
Sr to the local collective ingestion dose by food product.
The exported collective ingestion dose is predominantly due to lamb (Fig. 4). This is due to the high production of lamb in Cumbria and high long-term bioavailability of radiocaesium in upland areas where the grazing of sheep is the dominant agricultural practice.
Fig. 4. Contribution of
137
Cs and
90
Sr to the exported collective ingestion dose by food product.
392
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Radiocaesium is the greatest contributor to dose (approximately 50 times that of radiostrontium, other nuclides are negligible). This is due to a combination of its high relative deposition and long-term bioavailability in Cumbrian agricultural systems, and comparatively low transfer of the other radionuclides to meat (the production of which is a predominant farming practice in the area). The annual collective dose falls with time, approximately 40% of the total collective dose occurs during the first year after deposition; years 2 and 10 contribute approximately 10 and 5%, respectively. 4.2. ‘‘Food restrictions’’ Implementation of CFILs is predicted to cost approximately £2000 million, and avert approximately 33 000 person-Sv over the 10-year study period (Table 1). Most of the averted dose would be due to a reduction in the exported collective ingestion dose. This would be reduced by a factor of approximately 12, compared with ‘‘do nothing’’, whilst the local collective ingestion dose would be reduced by a factor of approximately 2. The maximum individual dose is reduced from approximately 75 mSv to !30 mSv (Fig. 2). The value of the cost function is positive (Table 1) as the resources required outweigh the monetary value of the benefit (i.e. the averted dose). According to ICRP philosophy these measures are therefore unjustified (ICRP, 1988). The majority of the implementation cost is due to the long-term restriction of lamb and beef food products (Table 2), because, as outlined earlier, a significant proportion of lamb and beef in Cumbria is produced on land susceptible to longterm radiocaesium bioavailability. It is predicted that, after 10 years, the radiocaesium activity concentration of lamb would still be above the recommended CFIL in nearly 50% of the grid squares in Cumbria, whilst restrictions on the entry of beef into the food chain would still be required in approximately 40% of the grid squares. Of course it is unlikely that normal agricultural practice would continue in these areas for such an extended period (i.e. market pressures or regulation would probably force changes) so the associated predicted costs are probably an overestimate, although some form of long-term compensation would probably be required. However, predicting such changes is impracticable. Table 2 Cost of food restrictions by food product and waste volumes generated, under the ‘Food Restrictions’ option over a period of 10 years post-deposition Food product
Cost of restriction (million £)
Waste generated
Lamb Beef Cow’s milk Cereals Cow cheese Fruit Potatoes
1500 728 19 8.7 15.2 0.5 0.5
5.2 ! 106 animals 1.7 ! 105 animals 9.4 ! 107 L 2.8 ! 108 kg 7.5 ! 107 L milk 2.7 ! 106 kg 2.4 ! 107 kg
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
393
4.3. ‘‘Optimal’’ The countermeasures suggested for implementation by countermeasure optimisation, their associated costs, and extent are shown in Table 3. The focus of the optimal countermeasure combination is to reduce the activity concentrations in sheep and beef cows, which are major contributors to both the local and total collective ingestion dose. This is achieved by skim and burial ploughing of pasture in those grid squares where it is permitted and practical, and by the administration of clean feed and AFCF in the grid squares where ploughing is restricted (Fig. 5). We should emphasise that the optimal countermeasure strategy requires the various countermeasures to be implemented for different time periods at different locations within the simulated region. The temporal variation is illustrated in Fig. 6 which shows the proportion of the regions grid squares in which AFCF (to diary cows, beef cattle, and sheep) and clean feeding (to beef cattle and sheep) are implemented as a function of time. During the first 4e5 months after deposition the extent of these countermeasures broadly increases. This is because the animal’s activity concentrations fall sufficiently for these measures to be able to reduce activity concentrations below CFILs and consequently the requirement for food bans restrictions reduces. After this period there is a slow but gradual decline in the extent of the countermeasure implementation, although large parts of the region still remain affected 10 years after deposition. The reduction in local collective ingestion dose is 40% greater than by the imposition of CFILs alone, whilst the exported collected dose is reduced by a further 10% (Table 1). Moreover, the cost of implementing the optimal countermeasure strategy is !10% of using only food restrictions. The difference between the costs of ‘‘Food restrictions’’ and ‘‘Optimal’’ is largely due to the reduced amount of lamb and beef products that require restriction and disposal over the 10-year period. The effect of the optimal countermeasure combination on the distribution of individual doses received by the Cumbrian population is to further reduce the number of people receiving the highest doses (Fig. 2). Table 3 Implementation cost and, where applicable, extent of implemented countermeasures under ‘‘optimal’’ management over a period of 10 years post-deposition Countermeasure implemented
Cost (million £)
Logistical extent/waste produced
Skim and burial plough (Pasture) K fertiliser (Pasture) Food restrictions
41.1 34.1 36.0
Clean feed (Sheep) AFCF (Sheep) Deep plough (Silage crops) Clean feed (Beef cows) AFCF (Beef cows) AFCF (Dairy cows) Skim and burial plough (Edible crops)
20.7 11.9 8.0 2.0 2.0 1.4 1.4
116 000 ha 125 000 ha 48 000 t milk 64 000 animals 291 000 ha edible crops 2.5 million sheep 2.9 million sheep 133 000 ha 38 000 beef cows 314 000 cows 111 000 cow months 11 000 ha
394
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
Fig. 5. Spatial extent of skim and burial ploughing of pasture (left), clean feeding of sheep 12 months after deposition(centre) and AFCF administration to sheep 12 months after deposition (right). Dark areas indicate where the countermeasure is being applied.
5. Discussion Countermeasure optimisation suggests a set of countermeasures which are an improvement over the sole use of food restrictions; the radiation protection outcome is similar but is achieved at a much lower cost. The countermeasures selected are focussed on reducing the activity concentrations of lamb and beef in particular. These results suggest that ploughing is the agricultural countermeasure of first choice. It is relatively cheap and effective, and in the simulations presented here it is AFCF (Dairy cows) AFCF (Cattle) AFCF (Sheep) Clean feed (Cattle) Clean feed (Sheep)
70
% of Grid Squares
60 50 40 30 20 10
1
2
m 0 on m th on t 3 m hs on t 4 m hs on t 5 m hs o 6 nth m s o 7 nth m s o 8 nth m s o 9 nth m s 10 ont m hs 11 ont m hs 12 ont m hs on t 2 hs ye ar s 3 ye a 4 rs ye a 5 rs ye a 6 rs ye a 7 rs ye a 8 rs ye a 9 rs ye 10 ars ye ar s
0
Time after deposition Fig. 6. Temporal variation in the regional extent of five animal based countermeasures. The vertical axis is the percent of the region’s grid squares where the countermeasure is conducted. Note: The horizontal time axis is not linear, the first year is expanded to illustrate early changes in countermeasure implementation.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
395
selected wherever it is allowed. However, the model takes no account of the logistical requirements of widespread ploughing. Therefore, implementation time may be longer in reality than that assumed in the model. This may influence the effectiveness of the countermeasure. Also the spatial resolution of 5 ! 5 km may have led to an overestimate of the areas which can be ploughed. For example, within the complex topography of Cumbria there will be localised areas of high slope unsuitable for ploughing which are lost when a mean slope is calculated for the grid square as a whole. Where ploughing is not allowed AFCF administration and clean feeding are selected, generally in combination with one another. For example, AFCF can be used to reduce the amount of clean feeding required to lower an animal product to its CFIL; AFCF being administered for a period prior to clean feeding. This does not affect the radiological outcome but does reduce the cost. Much of the benefit, in terms of dose reduction, of the suggested countermeasure is for people living outside of the region, whose individual doses are likely to be negligible. In part this is because any agricultural countermeasure that is implemented will reduce more exported than local dose, simply because more food products are exported from the region than can be consumed locally. It would be possible to extend the analysis to consider external doses and corresponding relevant countermeasures. In this case, any benefits would obviously accrue directly to the population of the contaminated areas. Of course the model presented has many limitations. Of these, perhaps the most noteworthy is the exclusion of some key radionuclides, especially 131I. Although this was outside the scope of our work the model could be extended to consider 131I. Clearly, this isotope has considerable potential for short-term food contamination. In the example we have presented we could anticipate that it would increase the requirement for short-term food restrictions/countermeasures. However, such measures were already recommended for many areas within the contaminated region due to radiocaesium and radiostrontium contamination. Therefore, the inclusion of 131I would probably force adjustments to the optimal strategy rather than its complete revision. If such models were to be considered for operational use then clearly the inclusion of 131I would be required. The model application presented demonstrates the importance of relevant local information in planning effective remediation. Recommendations that a combination of ploughing, AFCF administration, clean feeding and food restrictions are the essential ingredients of a remediation strategy will not be a surprise to experts in the field. However, the model has the potential to identify the optimal combination of these measures, in both space and time, taking into account the implementation costs. This is a high-dimension problem which is very difficult to visualise in a form that makes it manageable without computational tools such as the model presented. Therefore, the model has the potential to form a useful part of the decision making process. It is not intended to provide definitive answers; rather, it allows decision makers to investigate the effects of various options, and to provide information about the likely consequences of countermeasure combinations. The model provides a starting point for the design of a restoration strategy by highlighting the major sources of dose to affected populations, together with recommendations on when,
396
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
where and in what combinations countermeasures should be employed. To our knowledge the approach described is the first to successfully implement a countermeasure strategy optimisation within spatially predictive models. Acknowledgements The STRATEGY project was conducted under a contract (FIKR-CT-2000-00018) within the research and training programme (Euratom) in the field of nuclear energy, and this support is gratefully acknowledged. The paper is the sole responsibility of the authors and does not reflect Community opinion, and the Community is not responsible for any use that might be made of data appearing in this publication. We also wish to acknowledge the contribution of all the STRATEGY project members and the comments of an anonymous reviewer which improved this paper considerably. References Absalom, J.P., Young, S.D., Crout, N.M.J., Sanchez, A., Wright, S.M., Smolders, E., Nisbet, A.F., Gillett, A.G., 2001. Predicting the transfer of radiocaesium from organic soils to plants using soil characteristics. J. Environ. Radioact. 52, 31e43. CEC, 1989. Council Regulation (Euratom) No. 2218/89 amending Regulation (Euratom) No. 3954/87 laying down maximum permitted levels of radioactive contamination of foodstuffs and feedingstuffs following a nuclear accident or any other case of radiological emergency. Official Journal of the European Communities L211/1. Cox, G.M., Crout, N.M.J., 2003. A methodology for the selection and optimisation of countermeasure strategies. Report for the STRATEGY project (CEC Contract N : FIKR-CT 2000e00018). University of Nottingham, UK. Gillett, A.G., Crout, N.M.J., Absalom, J.P., Wright, S.M., Young, S.D., Howard, B.J., Barnett, C.L., McGrath, S.P., Beresford, N.A., Voigt, G., 2001. Temporal and spatial prediction of radiocaesium transfer to food products. Rad. Environ. Biophys. 40, 227e235. Howard, B.J., Andersson, K.G., Beresford, N.A., Crout, N.M.J., Gil, J.M., Hunt, J., Liland, A., Nisbet, A., Oughton, D., Voigt, G., 2002. Sustainable restoration and long-term management of contaminated rural, urban and industrial ecosystems. Radioprotection Colloques 37 (C1), 1067e1072. Fuller, R.M., Groom, G.B., Jones, A.R., 1994. The land cover map of great britain: an automated classification of landsat thematic mapper data. Photogramm. Eng. Remote Sens. 60, 553e562. International Commission on Radiological Protection (ICRP), 1973. Implications of commission recommendations that doses be kept as low as readily achievable. ICRP 22. Pergamon, Oxford. International Commission on Radiological Protection (ICRP), 1988. Optimisation and decision-making in radiological protection. ICRP 55. Pergamon, Oxford. International Commission on Radiological Protection (ICRP), 1990. Recommendations of the ICRP. ICRP 60. Pergamon, Oxford. Katona, T., Kanya´r, B., Eged, K., Kis, Z., Ne´nyei, A´., Bodna´r, R., 2003. The monetary value of the averted dose for public exposure assessed by the willingness to pay. Health Phys. 84, 594e598. Kelly, G.N., Clarke, R.H., 1982. An assessment of the radiological consequences of releases from degraded core accidents for Sizewell PWR. NRPB-R137. National Radiological Protection Board, Didcot. Linsley, G., 2002. Summary of the IAEA Arlington symposium on the restoration of environments with radioactive residues. In: Radiation Legacy of the 20th Century: Environmental Restoration. IAEATECDOC-1280. International Atomic Energy Agency, Vienna, pp. 30e34. Lochard, J., Lefaure, C., Scheiber, C., Schneider, T., 1996. A model for the determination of monetary values of the man-sievert. J. Radiol. Prot. 16, 201e204.
G. Cox et al. / J. Environ. Radioactivity 83 (2005) 383e397
397
Ministry for Agriculture, Fisheries, and Food (MAFF), 1990. The dietary and nutritional survey of British adults. SS1241, HMSO, London. Ministry of Agriculture, Fisheries and Food (MAFF), 1996. The digest of agricultural census statistics, United Kingdom. The Stationary Office, London. ISBN 0 11 243039 2. McGrath, S.P., Loveland, P.J., 1992. Geochemical Atlas of England and Wales. Blackie Academic & Professional, London. Mu¨ller, H., Pro¨hl, G., 1993. Ecosys-87: dynamic model for assessing radiological consequences of nuclear accidents. Health Phys. 64, 232e252. Nisbet, A.F., Mercer, J.A., Hesketh, N., Liland, A., Thørring, H., Bergan, T., Beresford, N.A., Howard, B.J., Hunt, J., Oughton, D., 2004. Datasheets on countermeasures and waste disposal options for the management of food production systems contaminated following a nuclear accident. NRPBW58. National Radiological Protection Board, Didcot. Nix, J., 2002. Farm Management Pocketbook, thirty-second ed. Imperial College at Wye, Melton Mowbray. ISBN 0-9541201-0-8. National Radiological Protection Board (NRPB), 1993. Occupational, Public, and Medical Exposure. Documents of the NRPB 4 (2). NRPB, Didcot. Press, W.H., Flannery, B.P., Teukolsky, S.A., Vettering, W.T., 1986. Numerical Recipes. Cambridge University Press, Cambridge. Roed, J., Andersson, K.G., Prip, H., 1996. The skim and burial plough: a new implement for reclamation of radioactively contaminated land. J. Environ. Radioact. 33 (2), 117e128. Stewart, T.H., Fulker, M.J., Jones, S.R., 1990. A survey of habits of people living close to the Sellafield nuclear reprocessing plant. J. Radiol. Prot. 10, 115e122.