GIS/EM4 - A GIS-framework for modeling environmental fate of Chernobyl-derived radiocesium

A GIS-framework for modeling environmental fate of Chernobyl-derived radiocesium

GIS/EM4 No. 130

Marcel van der Perk
Andrew G. Gillett
Jiske R. Burema

Abstract

The Chernobyl accident in Ukraine, April 1986, has resulted of soil contamination by radiocesium over vast areas in Europe, particularly in Ukraine, Belarus, and Russia. To assess the environmental fate of radiocesium released by the Chernobyl accident, a flexible, user-friendly GIS-framework has been developed, which comprises several GIS-embedded environmental models. These models enable to simulate radiocesium transfer into human food chains, external radiation exposures to humans, and radiocesium redistribution within catchments. Besides the models, the GIS-framework includes various environmental spatial data sets of areas in Ukraine, Belarus, and Russia, to feed the models, including digital elevation models and maps of soil contamination, soil type, and land use. The overall aim of the GIS-framework is to provide local scientists a modeling tool that enables the identification of vulnerable areas in terms of enhanced radionuclide transfer into food chains and/or the presence of 'critical population groups' that suffer enhanced internal and/or external exposure to radionuclides. In addition, it permits the user to evaluate the economic feasibility of short rotation coppice as an alternative land use for energy production in highly contaminated areas. Therefore, it provides information to support decisions about where to implement countermeasures and where to restore contaminated land most effectively.

Keywords

Chernobyl, radioecological modeling, radiocesium, decision support systems

Introduction

The accident in the Chernobyl nuclear power plant in Ukraine, 26 April 1986, has resulted in deposition of various radionuclides over vast areas in Europe. In contaminated areas of Ukraine, Belarus and Russia, the resulting soil contamination by radiocesium (Cs-137) may still be a source of enhanced people's radiation exposure, both external and internal through ingestion of contaminated food products. Although initial deposition patterns are very important at the national scale, at the local scale the importance of other factors that governs the fate of radiocesium often becomes manifest. At the local scale, the initial patterns of soil contamination by radiocesium have been affected by physical decay and radiocesium redistribution due to soil erosion and deposition processes. In the time passed since the accident, most radiocesium has been irreversibly adsorbed to soil clay minerals. However, in soils with very low clay content, for instance in sandy podzolic and peat bog soils, radiocesium in soil can still be available for uptake and so be transferred into food chains. To predict the transfer of radiocesium into human foodchains it is therefore essential to allow for the spatial and temporal variation of the factors governing radiocesium contamination and uptake.
As part of various EC-funded projects, a flexible, generic GIS-framework has been developed to model the environmental fate of radiocesium that was released by the Chernobyl accident, accounting for the spatial and temporal variation of soil contamination, the different pathways for radiocesium, and the geo-chemical, hydrological, and biological processes involved (Van der Perk et al. 1998, Van der Perk et al. 2000b). The overall aim of this GIS-framework is to identify vulnerable areas in terms of enhanced external radiation exposure and/or radionuclide transfer into food chains and to provide local scientists a modeling tool to design and evaluate management strategies for radioactively contaminated land. The framework comprises data-sets of selected contaminated areas in Ukraine, Belarus, and Russia and various GIS-embedded environmental models, which can be accessed through a user-friendly HTML-based interface.

Structure of the GIS-framework

The core of the GIS-framework is a chain of GIS-embedded environmental models to calculate radiocesium transfer through parts of the human food chain (see figure 1). The first step in the modeling chain is to simulate radiocesium transfer from soil to the most important agricultural and semi-natural food products based on maps of soil type, land use and soil contamination. Subsequently, radiocesium intake by humans is calculated using the previously calculated food product contamination and diet information. In addition, external exposure is calculated based on soil contamination by radiocesium and residence time in various environments. Besides models to calculate radiocesium transfer through the food chain and external doses, models to calculate radiocesium redistribution within catchments and to evaluate the economic feasibility of short rotation coppice are being developed and implemented in the GIS-framework.

Figure 1. Schematic overview of the chain of environmental models in the GIS-framework.

The GIS-framework also comprises generic spatial data sets from 48 areas in Ukraine, Belarus, and Russia (figure 2). The scales of these areas range from the entire contaminated zone of Ukraine, Belarus, and Russia (about 30000 km2), the district level (about 1000 km2), to the farm level (about 50 km2). The spatial databases include generic maps of soil contamination by radiocesium, soil type, soil texture, physical and chemical soil properties (clay content, organic matter content, bulk density, pH, soil exchangable potassium (only at the district and farm levels), land use, crop type, farms (only at the district level), and settlements. Each available map is accompanied with a meta-data document in which all steps made in converting from the original digital data (as they were made available) to the final PCRaster map (map and unit conversions, projection, reclassification) have been summarized.

Figure 2. Areas in Ukraine, Belarus, and Russia for which spatial environmental data are included in the GIS-framework (green). The Klintsy and Novozybkov districts, Russia, consist of 20 and 18 farms, respectively.

A user-friendly interface has been designed based on HTML documents, which permits the user to select the model area, adjust model input parameters, run models, display model inputs and outputs, and browse detailed descriptions of available data and models accounting for data sources, model concepts, and default parameter values. In this manner, the model results for various scenarios can be compared and evaluated. Figure 3 shows an example of the HTML interface, a model dialog box and display of a model outcome.

Figure 3. Example of the HTML interface, a model dialog box of the steady state transfer factor model and PCRaster display of the calculated milk contamination.

Outline of the spatial models

The models implemented in the GIS-framework have been largely based on existing models. All models have been implemented in the PCRaster spatio-temporal modeling language (Wesseling et al. 1996) and have been made generic so they can be applied to either any of the existing data sets or any other suitable user-defined spatial data set. The sections below briefly discuss the main features of the models. For a more detailed technical description of the models we refer to Van der Perk (2000b) and the other literature mentioned.

Radiocesium transfer model

Two types of models are available for predicting the soil to food product transfer: a simple steady state transfer factor model and a semi-mechanistic dynamic soil-plant transfer model. The first model adopts a transfer factor, i.e. the ratio between food product contamination (Bq/kg) and soil contamination (kBq/m2), which depends on soil texture (sand - clay - loam - peat) and time. Radiocesium contamination of six agricultural and semi-natural food products (milk, potatoes, grain, meat, mushrooms, and berries) can be predicted simultaneously. Soil contamination by radiocesium is corrected for physical decay (half-life = 30 y). The decline in transfer factor values through time is predicted using a double exponential equation using respective half-lives of 10 years and 1 year.
The second model adopts a semi-mechanistic model that predicts radiocesium transfer uptake as function of physical-chemical soil characteristics (clay content, exchangeable soil potassium) (Absalom et al. 1999) and of time using a time step of one month. This model predicts radiocesium contamination of ten agricultural products (pasture grass, hay, grain, straw, potatoes, silage (maize), leafy vegetation, cows’ milk, beef, and pork). Plant activity concentration (Bq/kg) is calculated from the the radiocesium activity concentration in soil solution using a concentration factor (CF, dm3/kg). This CF is related to soil solution potassium concentration, which is predicted as function of exchangeable potassium and soil clay content. The soil clay content is also used to predict the so-called radiocesium interception potential (RIP) which is the product of the radiocesium distribution coefficient (Kd, dm3/kg) and soil solution potassium concentration. The RIP and the soil solution potassium concentration can be used to estimate the soil Kd value and Kd may then be used to estimate soil solution radiocesium activity concentration. The model described above is applicable to agricultural soils with low to moderate organic matter contents. For organic soils, a second version of the model has been developed. The major adaptations involve the partitioning of sorbed potassium and radiocesium between organic and inorganic soil fractions. To predict activity concentration of animal products (cows’ milk, beef, and pork) a conventional equilibrium transfer factor approach is used, so the daily radiocesium intake is the product of the animals daily intake of pasture grass and fodder products, which changes over the seasons, and the activity concentrations of the animal food products.
The choice of which model to use depends upon the scale and quality of the input data: the more complex semi-mechanistic transfer model requires the availability of detailed information of various soil attributes. Where such detailed information is not available, the simpler transfer factor model can be employed.

Radiocesium intake model

The radiocesium intake model, including diet and origin of food product, calculates individual annually averaged daily radiocesium intake through foodstuffs (Bq/d) and the accompanying individual internal dose (mSv/y) based on basis a diet of an individual living in a given settlement. The average contamination of the ingested food product is either calculated as an area average (model area, user-defined farm or field) or a weighted average adopting a spatial function in which the weight factor decreases by a factor 2 with every user-defined half-weight distance from the settlement of residence. The daily radiocesium intake by an individual person via a food product is the product of the daily food product consumption and the radiocesium contamination of the food product. The accompanying effective internal dose is calculated by multiplying the calculated daily radiocesium intake by a dose conversion factor.

External dose model

The external dose model calculates individual external doses from various environments from soil contamination by radiocesium using a dose rate factor, defined as the ratio between external dose rate (mSv/y) and soil contamination by radiocesium (kBq/m2). The external dose is calculated as a weighted average for residence in five different environments, namely home settlement, work settlement, grassland fields, arable fields, and forests. Using a weighting factor as function of distance from the home settlement is optional. A correction factor for each environment has been introduced to account for, for example, housing conditions, reduction of doses as a result of countermeasures (e.g. decontamination of settlement areas) or increased radiation doses in forests.

Radiocesium redistribution model

Prediction of the changes in soil contamination by radiocesium at two time scales, namely at the event scale (hours) and the long-term scale (years). The basic input for both models is a map of initial soil contamination by radiocesium, a digital elevation model (DEM), and maps of soil type and land use.
The event-based radiocesium redistribution model has been based on the existing LISEM soil erosion model (De Roo et al. 1996). The LISEM model accounts for rainfall, interception, surface storage, infiltration, overland flow, detachment by rainfall and throughfall, detachment by overland flow, and transport capacity of overland flow. Radiocesium exchange processes between the dissolved and adsorbed phase in both top soil and overland flow have been incorporated based on a distribution coefficient (Kd) approach. Figure 4 shows the structure of the model. At the event time scale, it can be assumed that Kd is constant for the different sediment types and equilibrium is reached instantaneously. The Kd is estimated based on sediment type and time since initial deposition. Radiocesium exchange processes between the top soil and runoff water are modeled assuming an active layer of 5 mm. The initial radiocesium contamination values usually expressed as Bq/m2 are converted to activity concentrations in this active layer (Bq/kg) using soil bulk density and a standardized depth distribution of radiocesium that depends on time as a result of vertical advective and diffusive transport and ploughing. Besides the changes in soil radiocesium inventories, the model yields also rates of both dissolved and particulate radiocesium transport from the model area.

Figure 4. Schematic overview of the event-based radiocesium redistribution model.

The long-term radiocesium transport model is a simplified spatial version of a soil erosion and deposition model presented by Govers et al. (1993) and accounts for rill-erosion and transport capacity. The model adopts a time step of 1 year. The predicted soil erosion and deposition rates are used as input for a radiocesium mass-balance, for which the radiocesium activity concentration of the eroded sediment equals the activity concentration of the active layer (see above) and the activity concentration of the deposited sediment equals a weighted average of the activity concentration of eroded sediment from the upstream grid cells. Each time step, the radiocesium inventory in the top 25 cm of the soil profile is redistributed to simulate homogenization due to ploughing.

Short rotation coppice model

The short rotation coppice (SRC) model consists of three modules: 1) calculation of the potential area for SRC production, 2) evaluation of the profitability of SRC production, and 3) evaluation of the profitability of the conversion of SRC to energy. An area is considered to be potential for SRC production if the initial land use is suitable for SRC production and the radiocesium contamination of soil, SRC, and the ash of the SRC produced by energy conversion do not exceed pre-defined values. The contamination of the SRC and its ash is calculated using transfer factors and the loss on ignition of the SRC. The average external dose rate received by a person during SRC production is also estimated to evaluate whether the calculated potential areas are safe to work in. In potential areas, the yield per hectare is estimated based on soil texture and soil moisture conditions. Subsequently, the cashflow of the SRC production is calculated by subtracting the production cost from the gains from selling the yield to a energy conversion plant. The SRC production is profitable if the cashflow is greater then a user-defined minimal cashflow.
At the scale of the entire contaminated area of Ukraine, Belarus, and Russia, a window operation is adopted to calculate if the amount of SRC that is produced around a location is enough to support a conversion plant. At the farm scale, it is assumed that all SRC is produced within the farm. For each grid cell with sufficient profitable SRC production in its surroundings, the cashflow of the conversion plant is calculated using the price of energy, the energy production costs, and total energy production. If the average radiocesium contamination of the SRC ash exceeds an intervention limit, the extra costs for ash disposal is included in the energy production costs. If the cashflow is greater then a minimum cashflow, energy production is considered to be potentially profitable.

An example analysis

Although it is not the aim of this paper to present case studies for all different areas and models available in the GIS-framework, the output of the steady state transfer factor model and radiocesium intake model is illustrated here for the entire contaminated area of Ukraine, Belarus, and Russia at a resolution of 1 km2. For calibration, validation and application of the foodchain models we refer to Van der Perk et al. (2000a, 2000b). Since the radiocesium redistribution model and the short coppice rotation are currently being developed, modeling results have not been available yet, but will be published in separate papers in due time.
In this example, we calculated radiocesium ingestion via private consumption of mushrooms for a person living in Gomel town, Belarus. First the radiocesium contamination of mushrooms was predicted using the steady state transfer factor model and, subsequently, the daily radiocesium intake and accompanying internal dose from mushroom consumption was calculated using the transfer factor model. We used the default transfer factor values of the models (Van der Perk 1999) and a half-weight distance for private mushroom collection of 8000 m.
Figure 5 shows part of the spatial output from this example calculation, namely the mushroom contamination by radiocesium and the origin of ingested radiocesium from 0.002 kg day-1 consumption of privately collected mushrooms. The total ingestion rate of radiocesium via mushrooms amounts to 34.5 Bq day-1 and the internal dose rate amounts to 0.16 mSv/y. Although it was assumed that most mushrooms were collected in the neighborhood of Gomel town, the ingested radiocesium largely originates from the most contaminated area north-east of Gomel town. The area south-west of Gomel town contributes very little to the radiocesium ingestion, whilst the pattern of mushroom collection from this area was assumed to be the same as for the area north-east of the town. Figure 5b obviously indicates which areas should be avoided for collection of foodstuffs.


a)	b)
Figure 5. Examples of output from the GIS-embbedded models: a) Mushroom contamination by radiocesium (Bq/kg) from the steady state transfer facter model; b) Spatial origin of radiocesium ingestion (Bq/ha/day) from the radiocesium intake model. The arrow indicates the location of Gomel town.

Concluding remarks

The GIS-framework presented in this paper has resulted in the implementation of a range of straightforward and widely accepted radioecological models into a spatio-temporal modeling environment and, accordingly, has assisted to focus on the variation of the environmental transfer of radiocesium in both space and time. It was developed to provide a PC-based modeling tool for interactive data analysis and modeling to identify the key pathways of radiocesium exposure. Although the framework does not give direct solutions for the problems of enhanced radiation exposure, it provides a means whereby the implications of land use changes, radiocesium redistribution, human diet and behavior on radiocesium ingestion and exposure may be assessed and the location, extent, and spatial context of problems can be identified. In this way, it helps to design optimal management strategies for radioactively contaminated areas and to make decisions about the application of most appropriate countermeasures to reduce radiation exposure to population groups tailored to local and even individual circumstances.
The spatial data sets of the selected model areas and the models that simulate the soil-plant-human transfer of radiocesium have recently been issued on CD-ROM as an environmental decision support system (Van der Perk et al. 1999). This system is already able to satisfy many of the needs of local scientists and the flexibility and allows an easy implementation of new models, such as the radiocesium redistribution model and short rotation coppice model.

Acknowledgements

The development of the GIS-framework was funded by several projects funded by the European Commission DGXII, namely RESTORE (EC Contract No. FI4P-CT95-0021c), STRESS (EC Contract No. ERBIC15-CT96-0215), PHYTOR (EC Contract No. ERBIC15-CT98-0213), and SPARTACUS (EC Contract No. ERBIC15-CT98-0215). The authors acknowledge all partners who have contributed to the development of the framework.

References

Absalom JP, Young SD, Crout NMJ, Nisbet AF, Woodman RFM, Smolders E , Gillett AG 1999. Predicting soil to plant transfer of radiocesium using soil characteristics. Environmental Science and Technology 33 (8): 1218-1223.

De Roo APJ, Wesseling CG, Ritsema CJ 1996. LISEM: a single event physically-based hydrologic and soil erosion model for drainage basins. I: Theory, input and output. Hydrological Processes 10 (8): 1119-1126.

Govers G, Quine TA, Walling DE 1993. The effect of water erosion and tillage movement on hillslope profile development: a comparison of field obsevations and model results. In: Wicherek S, editor, Farm land erosion in temparate plains environment and hills. Amsterdam: Elsevier. p. 285-300.

Van der Perk M, Burrough PA, Voigt G 1998. GIS-based modelling to identify regions of Ukraine, Belarus, and Russia affected by residues of the Chernobyl nuclear power plant accident. Journal of Hazardous Materials 61: 85-90.

Van der Perk M, Burema JR, Gillett AG, De Jong K, Van der Meer MB, Wesseling CG 1999. RESTORE Environmental Decision Support System. CD-ROM; Deliverable for the RESTORE project (EC Contract No. FI4 CT95 0021c). Utrecht: Utrecht University. Available from the author.

Van der Perk M, Lev T, Gillett AG, Absalom JP, Burrough PA, Crout NMJ, Garger EK, Semiochkina N, Stephanishin YN, Voigt G 2000a. Spatial modelling of transfer of long-lived radionuclides from soil to agricultural products in the Chernigov region, Ukraine. Ecological Modelling 128 (1): 35-50.

Van der Perk M, Burema JR, Burrough PA, Gillett AG, Van der Meer MB 2000b A GIS-based environmental decision support system to assess the transfer of long-lived radiocaesium through food chains in areas contaminated by the Chernobyl accident. International Journal of Geographical Information Science (in press).

Wesseling CG, Karssenberg D, Burrough PA, Van Deursen WPA 1996. Integrating dynamic environmental models in GIS: the development of a dynamic modelling language. Transaction in GIS 1: 40-48.

Authors

Marcel van der Perk, Lecturer/Researcher,Utrecht Centre for Environmental Dynamics, Faculty of Geographical Sciences, Utrecht University, P.O. Box 80115, 3508 TC Utrecht, The Netherlands. Email: m.vanderperk@geog.uu.nl, Tel: +31-30-2535565, Fax: +31-30-2531145, URL: http://globis.geog.uu.nl/Users/Perk/

Andrew G. Gillett, Researcher, Environmental Sciences Division, School of Biological Sciences, University of Nottingham, Sutton-Bonington Campus, Loughborough, LE12 5RD, Umited Kingdom. Email: andy.gillett@nottingham.ac.uk, Tel: +44-115-9516259, Fax: +44-115-9516261

Jiske R. Burema, Researcher,Utrecht Centre for Environmental Dynamics, Faculty of Geographical Sciences, Utrecht University, P.O. Box 80115, 3508 TC Utrecht, The Netherlands. Email: j.burema@geog.uu.nl, Tel: +31-30-2535565, Fax: +31-30-2531145