- Email: [email protected]
Probabilistic indicators of atmospheric transport for regional monitoring and emergency preparedness systems
Environment International 29 (2004) 1063 – 1069 www.elsevier.com/locate/envint
Probabilistic indicators of atmospheric transport for regional monitoring and emergency preparedness systems Alexander Mahura a,b,*, Alexander Baklanov a a
b
Danish Meteorological Institute, Lyngbyvej 100, DK-2100 Copenhagen, Denmark Institute of the Northern Environmental Problems, Kola Science Center, Apatity, RUS-184200, Russia Received 25 November 2002; accepted 7 April 2003
Abstract In this paper, following a methodology developed within the ‘‘Arctic Risk’’ Project of the Nordic Arctic Research Programme, several probabilistic indicators to evaluate the risk site possible impact on the geographical regions, territories, countries, counties, cities, etc., due to atmospheric transport from the risk site region were suggested. These indicators—maximum possible impact zone, maximum reaching distance, and typical transport time—were constructed by applying statistical methods and using a dataset of isentropic trajectories originated over the selected nuclear risk site (Ignalina nuclear power plant, Lithuania) during 1991 – 1996. For this site, the areas enclosed by isolines of the maximum possible impact zone and maximum reaching distance indicators are equal to 42 104 and 703 104 km2, respectively. The maximum possible impact zone’s boundaries are more extended in the southeast sector from the site and include, in particular, Latvia, Lithuania, Belarus, and several western regions of Russia. The maximum reaching distance’s boundaries are twice more extended in the eastern direction from the site (reaching the Caspian Sea) compared with the western direction. The typical transport time to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, and western border regions of Russia, and central aquatoria of the Baltic Sea. Detailed analysis of temporal patterns for these indicators showed importance of the seasonal variability. D 2003 Elsevier Science Ltd. All rights reserved. Keywords: Atmospheric transport; Emergency preparedness; Monitoring; Impact zone
1. Introduction For planning regional systems of monitoring and emergency preparedness, it is important to know the relative geographical distribution of the risk levels from various risk sources or sites, and in particular, the possible atmospheric transport time from the sources of concern toward the regions of interest. These transport times of major interest include the typical time of transport and the fastest transport from the site of concern toward remote areas. These regions of interest could include geographical territories, countries, administrative units, cities, etc. Among the sources of
* Corresponding author. Danish Meteorological Institute, Lyngbyvej 100, DK-2100 Copenhagen, Denmark. Tel.: +45-3915-7441; fax: +45-3915-7460. E-mail address: [email protected] (A. Mahura). 0160-4120/$ - see front matter D 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0160-4120(03)00100-4
concern could be possible danger from nuclear, chemical, biological sources, etc. In this study, the Ignalina nuclear power plant (NPP) (55.5jN vs. 26jE; Lithuania), which is the oldest Chernobyl-type reactor, was selected as a risk site (RS), although other risk sites of biological or chemical concern could be likewise chosen. Information about transport time can help regional authorities and decision makers to plan more effectively the system of operational monitoring and emergency preparedness (i.e., to know when different regions, counties, administrative units, etc., should be ready for countermeasures after an accident/event has occurred at risk sites). In this paper, the probabilistic method to evaluate the RS possible impact, from the probabilistic point of view, due to atmospheric transport from the site on the surrounding and remote territories is used for such purposes. This method is based on application of: (1) trajectory modelling of atmospheric transport from the site for a multiyear period, and (2)
1064
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
probabilistic analysis of calculated trajectory dataset to build several useful indicators reflecting the RS possible impact. The impact is presented by a set of probabilistic indicators: maximum possible impact zone, maximum reaching distance, and typical transport time.
For NPP, in the case of an accidental release, the most probable release heights would be within the boundary layer of the atmosphere, i.e. within several hundred meters above the ground. Therefore, at the next step, only those trajectories originating within this layer (17,500) from all trajectories (280,000) were selected for further statistical analysis.
2. Methodology
2.2. Construction of indicators of risk site possible impact
In this paper, we will present and discuss some methodological aspects and interpretation of results based on the application of the probabilistic analysis of the NRS possible impact suggested by AR-NARP (2001 –2003), Baklanov and Mahura (2001), Mahura (2001), Baklanov et al. (2002b,c) and Mahura and Baklanov (2002).
The statistical analysis of trajectories representing the atmospheric transport from the RS region is one of several ways to estimate, from a probabilistic point of view, the RS possible impact on geographical regions, territories, countries, sites, etc. For this purpose, a large dataset of forward isentropic trajectories that passed over different geographical regions was computed for the risk site. Each calculated trajectory contains information about longitude, latitude, altitude, pressure, temperature, relative humidity, etc., for every 12-h interval. In further discussion, the focus is only (1) on the trajectory’s latitude and longitude at each time step of 12 h (representing atmospheric transport of air parcels), and (2) trajectories that originated over the NPP region within the boundary layer (up to 1500 m) during 1991– 1996. In addition to the airflow and fast transport probability fields described by Mahura et al. (2001, 2002), Baklanov et al. (2002a) and Mahura (2002) for the Kola NPP (Murmansk Region, Russia) and RS in the Russian Far East, and by Mahura et al. (2003) and Baklanov et al. (2003) for the Kursk nuclear submarine’s lifting and transportation operation (Barents Sea vs. Kola Peninsula, Russia), three new probabilistic indicators of the RS possible impact can be introduced. There are several approaches to construct probability fields based on trajectory modelling results (as shown by Baklanov and Mahura, 2001; Mahura, 2001; Mahura and Baklanov, 2002). For all these approaches, initially, a new gridded domain with the site located in the centre of domain on the intersection between grid lines should be constructed. The number and resolution of grid cells depend on a resolution of original meteorological fields used for trajectory calculation purposes as well as the possibility of taking into account the remotest geographical regions from the RS location, which the airflow might hypothetically reach within 5 days of atmospheric transport. The first indicator—maximum possible impact zone (MPIZ)—shows boundaries of territories with the highest probability of being reached by trajectories during the first day of atmospheric transport from the site. To construct the MPIZ indicator, all endpoints of 12-h intervals of calculated trajectories only during the first day (i.e., at 12 and 24 h) of transport were counted in the cells of a new rectangular gridded domain. Then, among all these cells, the cell in which the absolute maximum of intersections took place was identified as the ‘‘absolute maximum cell’’ (AMC). The number of intersections in other cells adjacent to AMC was
2.1. Trajectory modelling for risk sites Trajectory models are useful tools to evaluate airflow patterns within meteorological systems on various scales. These models use different kind of approaches to simulate temporal and spatial air parcel movement. In this study, an isentropic approach to model trajectories originating over the Ignalina NPP region has been selected. Although it uses the assumption of adiabatically moving air parcels and neglects various physical effects, it is still a useful research tool for such kind of studies. It should also be noted that some uncertainties in trajectory models and, hence, calculated trajectories are related to the interpolation of the original meteorological data, assumptions of vertical transport, numerical scheme errors, selection of initial points for trajectory calculation, errors of forecasted and analyzed fields, etc. In this study, the original National Center for Environmental Prediction (NCEP) gridded wind fields were interpolated to potential temperature levels (or isentropic surfaces) for a period of 6 years (1991 –1996). Then, the wind fields on isentropic surfaces were used to calculate (based on a technique described by Merrill et al., 1985) the 5-day forward isentropic trajectories originating over the RS region. The gridded domain selected for this study is located between 20jN – 85jN and 60jW – 127.5jE. The trajectories (in a total of more than 280,000) were computed twice per day (at 00:00 and 12:00 Universal Coordinated Time (UTC)) at different potential temperature levels (in total 16). More details concerning the trajectory calculation for the nuclear risk sites can be found in Baklanov and Mahura (2001) and Mahura (2001). The quality of trajectory calculation is highly dependent on the original quality of the NCEP fields (2.5 2.5j latitude vs. longitude), and it may not adequately reflect a contribution of frontal passages and local terrain phenomena. However, the trajectory errors arising from a single calculation might be smoothed over in further statistical analysis due to the large number of trajectories computed for the multiyear dataset.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
compared, and then assigned additional cells, which had less than 10% of difference between cells. Therefore, this new ‘‘region of maximums’’ represents the area (or region) with the highest probability of possible impact (AHPPI) from the risk site. Assuming a value of 100% for this region, the rest could be re-calculated as a percentage of AHPPI. For the MPIZ indicator, it was assumed that MPIZ includes all cells with an isoline of more than 90% of AHPPI, and then, an isoline of MPIZ was drawn through these cells. The second indicator—maximum reaching distance (MRD)—shows the farthest boundaries on a geographical map that might be reached during the first day by at least one trajectory originating over the RS location. To construct the MRD indicator, all endpoints of 12-h intervals of calculated trajectories during the first day (i.e., at 12 and 24 h) of atmospheric transport were used. An isoline of MRD was drawn through the grid cells of the rectangular grid domain where, at least, one trajectory intersected with the grid cell’s boundaries. It should be noted that although the likelihood that an air parcel will reach these boundaries is low, it is still a possible case of the most remote atmospheric transport from the risk site. The third indicator—typical transport time (TTT) field—shows (1) how long it will take for an air parcel to reach a particular geographical region during transport from the RS location, and (2) what territories would be at the highest risk during the first few days of contaminated cloud transport after an accident/event occurred at the risk site. To construct the TTT fields, the following procedure was applied. First, a new polar grid domain having 1260 grid
1065
cells—36 sectors (10j each) and 35 grid cells (2j each) along each sector line—with the risk site in the centre of domain was built. Second, for a particular term (for example, 12 h), the geographical coordinates of trajectories (latitude and longitude) were converted into the polar coordinates: radius and angle. Third, in each sector, a number of trajectory intersections was counted in each grid cell, and then compared among the grid cells along the sector line to find a cell with the absolute maximum of trajectory intersections (similarly called the AMC). For simplicity, if several AMC cells were identified along the sector line, then only the one closest to the risk site was selected. Finally, the TTT isoline for a particular term was drawn through AMCs in each of the 36 sectors (coordinates of the AMC centres were again converted from the polar into geographical coordinates). A similar procedure was applied for each term of 0.5, 1, 1.5 days, and so on.
3. Results and discussion In emergency response systems for accidental releases of radioactivity, an estimation of the typical transport time, boundaries of possible maximal contamination, and geographically farthest territories reachable by a contaminated cloud during a limited time atmospheric transport from the risk sites to/over a particular geographical territory, region, country, city, etc., is one of the important input parameters for the decision-making process. In this section, based on the example of the Ignalina NPP, the indicators, which could
Fig. 1. Annual boundaries of the maximum possible impact zone (—INP_MPIZ—) and maximum reaching distance (—INP_MRD—) indicators for the Ignalina NPP (INP).
1066
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
allow one to perform such an estimation and which could provide useful input data for the planning and decisionmaking processes, are presented and evaluated. 3.1. Maximum possible impact zone The MPIZ indicator is an important integral characteristic of the RS possible impact during the first day of atmospheric transport after an accident/event. For the Ignalina NPP, the annual boundary of the MPIZ indicator is shown in Fig. 1. It should be noted that, although MPIZ is concentrated in the RS vicinity, the configuration of the MPIZ boundary (or isoline) depends strongly on the dominating transport patterns during the first day of atmospheric transport from the site, and this isoline is extended along the main flow direction. For studies on the long-term consequences of routine discharges or accidental releases at RS, it is important to evaluate not only the geographical boundary of MPIZ, but
also area of MPIZ. To estimate such area, a simple approach could be used. First, a figure inside the MPIZ isoline is approximated by a set of triangles, where each triangle has a top at the RS coordinates and a side on the MPIZ isoline. Second, the areas of all triangles are calculated, and the total area under the MPIZ isoline was simply equal to a sum of areas of all triangles. For the Ignalina NPP, the annual area of MPIZ is equal to 42 104 km2. Its boundaries are more extended in the southeast sector from the site (between 24 –34jE and 53– 57jN), and include the eastern territories of Latvia and Lithuania, central and northern regions of Belarus, as well as, partially, territories of several Russian regions: Pskov, Bryansk, Smolensk, and Tver. Analysis of the MPIZ seasonal variability showed dependence on the wind velocities on the synoptic- and meso-scales throughout the year. On a seasonal scale (Fig. 2), the MPIZ area reaches its highest rate during winter (57 104 km2) and the lowest during fall (32 104 km2). During winter, the MPIZ boundary is more
Fig. 2. Seasonal boundaries of the maximum possible impact zone (—INP_MPIZ—) and maximum reaching distance (—INP_MRD—) indicators for the Ignalina NPP (INP) during (a) spring, (b) summer, (c) fall, and (d) winter.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
extended in the eastern direction from the site passing over 35jE and reaching the borders of Ukraine and Estonia. In all seasons, except spring (when MPIZ is more extended to the west of the site), the MPIZ boundaries are more pronounced in the southeast sector from the site. 3.2. Maximum reaching distance The maximum reaching distance (MRD) indicator, similar to the maximum possible impact zone indicator, is an important characteristic of the RS possible impact during the first day of atmospheric transport following an accident/ event. The boundary of the MRD indicator for the Ignalina NPP is shown in Fig. 1. There is a possibility that contaminated air parcels may reach the remotest geographical regions after the hypothetical accidental release. If one considers a large dataset of trajectories, the distance to these remote region boundaries could be in thousands of kilometres. The boundary for the MRD indicator depicted in Fig. 1 reflects such possibility. They could be interpreted as possible boundaries of the RS impact during the first day of atmospheric transport after the hypothetical accident/event. Applying a similar estimation of areas enclosed by the MRD isolines (as for the MPIZ), it was found that for the Ignalina NPP, the annual average MRD area is equal to 703 104 km2, with a maximum level in winter (843 104 km2) and a minimum level in summer (540 104 km2). The annual MRD boundary is extended significantly (almost twice) in the eastern direction from the site (passing over the 50jE longitude and almost reaching the Caspian Sea)
1067
compared with the western direction. On a seasonal scale (Fig. 2), during winter, the MRD boundaries are extended farther to the southeast as well as they pass over the Kola Peninsula compared with other seasons. During all seasons, except summer, the MRD western boundaries are extended farther to the west than Denmark. The annual MRD area is significantly larger (almost 17 times) than the MPIZ area, and the latter is also included in the MRD area. The dominating direction of atmospheric transport is less underlined by the MRD indicator compared with MPIZ. Because a construction of the MRD isolines includes cells of gridded domain with the lowest probabilities of atmospheric transport through these cells (i.e., at least one trajectory reached a cell), it is not possible to explain such temporal variability through the synoptical scale features alone (i.e., there is a contribution of the global scale patterns, as well as the fact that there might be anomalies in the transport patterns from the site from year to year). Additionally, it should be noted that simultaneous plotting of both—MPIZ and MRD—indicators allows one to estimate the order of difference (1) between the boundaries of possibly highly impacted territories and the farthest reaching boundaries during atmospheric transport from the site, and (2) between the highest (MPIZ) and the lowest (MRD) probabilities of the RS possible impact on the geographical territories of interest. 3.3. Typical transport time The typical transport time (TTT) fields show (1) the time it will take for an air parcel to typically reach a particular
Fig. 3. Annual typical transport time fields at 1 ( – 1d – ) and 2 ( – 2d – ) days of atmospheric transport for the Ignalina NPP (INP).
1068
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
geographical region from the risk site location, and (2) the territories that would be at highest risk during the first few days of radionuclide cloud transport after an accident at the risk site. On an annual scale, for INP (Fig. 3), the typical transport time to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, and the western border regions of Russia, and central aquatoria of the Baltic Sea. On the second day, the atmospheric transport from the site is relatively faster (i.e., isoline for 2 days is more—almost three times—extended in the eastern direction compared with the western direction). The typical transport time to reach territories of the Czech Republic, Germany, Denmark, Kazakhstan, and Romania is 2 days. During this time, the atmospheric transport could typically occur over the Ukraine, Poland, Central and Northwest Russia, Moldova, as well as Sweden and Finland adjacent to the Baltic Sea. The difference in distance between firstand second-day isolines is similar in both southern and northern directions, but it is 2 –3 times higher in the eastern direction compared with the western direction. On a seasonal scale, analysis of the TTT fields by Mahura and Baklanov (2002) showed some peculiarities. During winter, the TTT field is significantly shifted in the northeast sector from the site, almost reaching 40jE by the end of the first day, and 50jE by the end of the second day. During summer, the TTT field is almost twice more extended in the north – south direction compared with the east – west direction. Detailed analysis of the TTT fields for the nuclear risk site allows the identification of geographical regions and territories of the neighbouring countries, which might be potentially reached by atmospheric flows and might be impacted by radioactive pollution during a selected period of time, if an accident/event occurred at the risk site. This information could be used to forecast the arrival of the contaminated cloud to a certain territory, and to calculate, if necessary, doses to the population, and hence, allowing authorities to plan countermeasures, including informing and, in a critical scenario, evacuating the local population. It should be noted that construction of the TTT fields for later terms (more than 3 days) is complicated due to significant airflow propagation from the RS location, and hence, the later term isolines are not concentrated around the site and are less representative.
4. Conclusion In this paper, following a methodology developed within the ‘‘Arctic Risk’’ Project (AR-NARP, 2001 –2003), several probabilistic indicators to evaluate the risk site (RS) possible impact on geographical regions, territories, countries, etc., due to atmospheric transport from the RS region are
suggested. These indicators—maximum reaching distance, maximum possible impact zone, and typical transport time—were constructed by applying statistical methods and using a dataset of isentropic trajectories originated over the Ignalina NPP (selected as a risk site for this study) region during 1991 –1996. The first indicator is maximum possible impact zone (MPIZ) and it outlines the boundaries of territories with the highest probability of being reached by trajectories during the first day of atmospheric transport from the RS location. The second indicator is maximum reaching distance (MRD) and it depicts the farthest boundaries on the geographical map that might be reached during the first day by, at least, one trajectory originating from the RS location. The third indicator is typical transport time (TTT) field and it shows (1) the time it will take for an air parcel to reach a particular geographical region during transport from the risk site location, and (2) what territories would be at the highest risk during the first few days of the contaminated cloud atmospheric transport following an accident/event at the risk site. For the Ignalina NPP, the annual areas enclosed by the MPIZ and MRD isolines are equal to 42 10 4 and 703 104 km2, respectively. The MPIZ boundaries are more extended in the southeast sector from the site and include, in particular, Latvia, Lithuania, Belarus, and several western regions of Russia. The MRD boundaries are twice extended in the eastern direction from the site (reaching the Caspian Sea) compared with the western direction. The annual TTT to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, western border regions of Russia, and central aquatoria of the Baltic Sea. Detailed analysis of temporal patterns for these indicators showed the importance of seasonal variability. These indicators are additional information, which is an important factor for planning of operational forecast of radioactive situation after accident/event based on dispersion modelling for emergency preparedness. Furthermore, they are important input data for studies on the social and economic consequences of the RS impact on population and environment for neighbouring countries as well as for the multidisciplinary risk and vulnerability analysis, and probabilistic assessment of the meso-, regional-, and long-range transport of radionuclides. These suggested indicators are important characteristics of the RS impact prediction reflecting only the atmospheric transport, and they focus on optimization of the regional monitoring and emergency preparedness systems. In order to evaluate possible consequences for population and environment on local and regional scales, a series of appropriate models can be applied, for example, MACCS, COSYMA, ECOSYS, ECOMARC, and others. Such evaluation of consequences might be simplified if only territories, first of all, within the boundaries of the MPIZ indicator are considered with respect to selected RS.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
It should be also noted that modelling the selection of terms—every 12 h—depends on the temporal (interval of 12 h) and horizontal (2.5j 2.5j latitude vs. longitude) resolution of meteorological data. However, the linear interpolation of all indicators could be done for finer time intervals (for example: 1, 3, 6 h, and so on); it would be better to use a finer resolution meteorological data for modelling purposes. Any other model and meteorological dataset of finer resolution could be used too, for example: within the ‘‘Arctic Risk’’ Project (AR-NARP, 2001 – 2003) the ECMWF and DMI-HIRLAM datasets and 3D DMI trajectory model (Sørensen et al., 1994) are used for indicators of the RS possible impact based on the dispersion modelling approach.
Acknowledgements The authors are grateful to Drs. Leif Laursen and Jens Havskov Sørensen (Danish Meteorological Institute, Copenhagen), Ronny Bergman (Swedish Defence Research Authority, Umea˚), Olga Rigina (Geographical Institute, University of Copenhagen), Boris Segersta˚hl (Thule Institute of University of Oulu, Finland), John Merrill (University of Rhode Island, Graduate School of Oceanography, Narragansett, RI, USA), Frank L.Parker (Vanderbilt University, Nashville, TN, USA), Keith Compton (International Institute for Applied Systems Analysis, Laxenburg, Austria), Sven Nielsen (Risø National Laboratory, Denmark), and Steen C. Hoe (Danish Emergency Management Agency, DEMA) for collaboration, discussions and constructive comments. The authors are also thankful to an anonymous reviewer for critical comments and constructive suggestions. The computer facilities at the Danish Meteorological Institute (DMI) and National Center for Atmospheric Research (NCAR, Boulder, CO, USA) have been used extensively in the study. The NCAR meteorological data were used as input in the trajectory modelling. Thanks are also due to the computer consulting services at DMI and NCAR, especially, to Jess Jørgensen (HIRLAM group, DMI), Peter Teisner (Computer Services, DMI), and Ginger Caldwell (Scientific Computing Division, NCAR) for the collaboration, computer assistance, and advice. Financial support of this study included the grants of the Nordic Arctic Research Programme (NARP), Nordisk Forskerutdanningsakademi (NorFA), US Department of Energy (US DOE).
1069
References AR-NARP. On-going Project Atmospheric Transport Pathways, Vulnerability and Possible Accidental Consequences from the Nuclear Risk Sites in the European Arctic (Arctic Risk) of the NARP: Nordic Arctic Research Programme. WWW site: http://www.dmi.dk/f + u/luft/arcticrisk/main.html; 2001 – 2003. Baklanov A, Mahura A. Atmospheric Transport Pathways, Vulnerability and Possible Accidental Consequences from Nuclear Risk Studies: Methodology for Probabilistic Atmospheric Studies. Danish Meteorological Institute Scientific Report, 01-09, ISBN: 87-7478-450-1; 2001. 43 pp. Baklanov A, Mahura A, Jaffe D, Thaning L, Bergman R, Andres R. Atmospheric transport patterns and possible consequences for the European North after a nuclear accident. J Environ Radioact 2002a;60(1 – 2): 1 – 26. Baklanov A, Rigina O, Mahura A. Nuclear risk and vulnerability in the Arctic: new method for multidisciplinary assessments. In: Strand P, Jølle ˚ , editors. Proceedings of the 5th International Conference on T, Sand A Environmental Radioactivity in the Arctic and Antarctic, St. Petersburg, Russia, 16 – 20 June. Oslo, Norway: NRPA; 2002b. p. 397 – 400. Baklanov A, Mahura A, Rigina O, Sørensen JH. Nuclear risk and vulnerability: approach for multidisciplinary assessments. In: Strand P, Børretzen P, Jølle T, editors. Proceedings of the International Conference on Radioactivity in the Environment, Principality of Monaco, 2 – 5 September. Oslo, Norway: NRPA; 2002c. p. 87 – 92. Baklanov A, Mahura A, Sørensen JH. Methodology for evaluation of consequences of possible accidental atmospheric releases of hazardous matter: ‘‘Kursk’’ submarine study. Atmos Chem Phys Disc 2003;3:1 – 42. Mahura A. Probabilistic assessment of atmospheric transport patterns from nuclear risk sites. PhD Phys. and Math. thesis, Russian State Hydrometeorological University/Kola Science Center. Russian Academy of Sciences; 2001. 172 pp. Mahura A. Assessment of Impact of Russian Nuclear Fleet Operations on Russian Far Eastern Coastal Regions. IIASA interim report IR-02-004; January 2002. 76 pp. Mahura A, Baklanov A. Probabilistic analysis of atmospheric transport patterns from nuclear risk sites in Euro-Arctic Region. Danish Meteorological Institute Scientific Report, 02-15. ISBN: 87-7478-469-2; 2002. 87 pp. Mahura A, Andres R, Jaffe D. Atmospheric transport patterns from the Kola nuclear reactors. CERUM Northern Studies, vol. 24. Sweden: Umea˚ University; 2001. 33 pp. Mahura A, Baklanov A, Rigina O, Parker FL. Statistical analysis of atmospheric transport from the nuclear risk sites in the Arctic Region. In: ˚ , editors. Proceedings of the 5th International Strand P, Jølle T, Sand A Conference on Environmental Radioactivity in the Arctic and Antarctic, St. Petersburg, Russia, 16 – 20 June. NRPA; 2002. p. 119 – 23. Mahura A, Baklanov A, Sørensen JH. Methodology for evaluation of possible consequences of accidental atmospheric releases of hazardous matter. Radiat Prot Dosim 2003;103(2):131 – 9. Merrill J, Bleck R, Boudra DB. Techniques of Lagrangian trajectory analysis in isentropic coordinates. Mon Weather Rev 1985;114:571 – 81. Sørensen JH, Laursen L, Rasmussen A. Use of DMI-HIRLAM for operational dispersion calculations. In: Gryning SE, Millan MM, editors. Air Pollution Modelling and Its Application, vol. X. New York: Plenum Press; 1994. p. 373 – 81.
Probabilistic indicators of atmospheric transport for regional monitoring and emergency preparedness systems Alexander Mahura a,b,*, Alexander Baklanov a a
b
Danish Meteorological Institute, Lyngbyvej 100, DK-2100 Copenhagen, Denmark Institute of the Northern Environmental Problems, Kola Science Center, Apatity, RUS-184200, Russia Received 25 November 2002; accepted 7 April 2003
Abstract In this paper, following a methodology developed within the ‘‘Arctic Risk’’ Project of the Nordic Arctic Research Programme, several probabilistic indicators to evaluate the risk site possible impact on the geographical regions, territories, countries, counties, cities, etc., due to atmospheric transport from the risk site region were suggested. These indicators—maximum possible impact zone, maximum reaching distance, and typical transport time—were constructed by applying statistical methods and using a dataset of isentropic trajectories originated over the selected nuclear risk site (Ignalina nuclear power plant, Lithuania) during 1991 – 1996. For this site, the areas enclosed by isolines of the maximum possible impact zone and maximum reaching distance indicators are equal to 42 104 and 703 104 km2, respectively. The maximum possible impact zone’s boundaries are more extended in the southeast sector from the site and include, in particular, Latvia, Lithuania, Belarus, and several western regions of Russia. The maximum reaching distance’s boundaries are twice more extended in the eastern direction from the site (reaching the Caspian Sea) compared with the western direction. The typical transport time to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, and western border regions of Russia, and central aquatoria of the Baltic Sea. Detailed analysis of temporal patterns for these indicators showed importance of the seasonal variability. D 2003 Elsevier Science Ltd. All rights reserved. Keywords: Atmospheric transport; Emergency preparedness; Monitoring; Impact zone
1. Introduction For planning regional systems of monitoring and emergency preparedness, it is important to know the relative geographical distribution of the risk levels from various risk sources or sites, and in particular, the possible atmospheric transport time from the sources of concern toward the regions of interest. These transport times of major interest include the typical time of transport and the fastest transport from the site of concern toward remote areas. These regions of interest could include geographical territories, countries, administrative units, cities, etc. Among the sources of
* Corresponding author. Danish Meteorological Institute, Lyngbyvej 100, DK-2100 Copenhagen, Denmark. Tel.: +45-3915-7441; fax: +45-3915-7460. E-mail address: [email protected] (A. Mahura). 0160-4120/$ - see front matter D 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0160-4120(03)00100-4
concern could be possible danger from nuclear, chemical, biological sources, etc. In this study, the Ignalina nuclear power plant (NPP) (55.5jN vs. 26jE; Lithuania), which is the oldest Chernobyl-type reactor, was selected as a risk site (RS), although other risk sites of biological or chemical concern could be likewise chosen. Information about transport time can help regional authorities and decision makers to plan more effectively the system of operational monitoring and emergency preparedness (i.e., to know when different regions, counties, administrative units, etc., should be ready for countermeasures after an accident/event has occurred at risk sites). In this paper, the probabilistic method to evaluate the RS possible impact, from the probabilistic point of view, due to atmospheric transport from the site on the surrounding and remote territories is used for such purposes. This method is based on application of: (1) trajectory modelling of atmospheric transport from the site for a multiyear period, and (2)
1064
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
probabilistic analysis of calculated trajectory dataset to build several useful indicators reflecting the RS possible impact. The impact is presented by a set of probabilistic indicators: maximum possible impact zone, maximum reaching distance, and typical transport time.
For NPP, in the case of an accidental release, the most probable release heights would be within the boundary layer of the atmosphere, i.e. within several hundred meters above the ground. Therefore, at the next step, only those trajectories originating within this layer (17,500) from all trajectories (280,000) were selected for further statistical analysis.
2. Methodology
2.2. Construction of indicators of risk site possible impact
In this paper, we will present and discuss some methodological aspects and interpretation of results based on the application of the probabilistic analysis of the NRS possible impact suggested by AR-NARP (2001 –2003), Baklanov and Mahura (2001), Mahura (2001), Baklanov et al. (2002b,c) and Mahura and Baklanov (2002).
The statistical analysis of trajectories representing the atmospheric transport from the RS region is one of several ways to estimate, from a probabilistic point of view, the RS possible impact on geographical regions, territories, countries, sites, etc. For this purpose, a large dataset of forward isentropic trajectories that passed over different geographical regions was computed for the risk site. Each calculated trajectory contains information about longitude, latitude, altitude, pressure, temperature, relative humidity, etc., for every 12-h interval. In further discussion, the focus is only (1) on the trajectory’s latitude and longitude at each time step of 12 h (representing atmospheric transport of air parcels), and (2) trajectories that originated over the NPP region within the boundary layer (up to 1500 m) during 1991– 1996. In addition to the airflow and fast transport probability fields described by Mahura et al. (2001, 2002), Baklanov et al. (2002a) and Mahura (2002) for the Kola NPP (Murmansk Region, Russia) and RS in the Russian Far East, and by Mahura et al. (2003) and Baklanov et al. (2003) for the Kursk nuclear submarine’s lifting and transportation operation (Barents Sea vs. Kola Peninsula, Russia), three new probabilistic indicators of the RS possible impact can be introduced. There are several approaches to construct probability fields based on trajectory modelling results (as shown by Baklanov and Mahura, 2001; Mahura, 2001; Mahura and Baklanov, 2002). For all these approaches, initially, a new gridded domain with the site located in the centre of domain on the intersection between grid lines should be constructed. The number and resolution of grid cells depend on a resolution of original meteorological fields used for trajectory calculation purposes as well as the possibility of taking into account the remotest geographical regions from the RS location, which the airflow might hypothetically reach within 5 days of atmospheric transport. The first indicator—maximum possible impact zone (MPIZ)—shows boundaries of territories with the highest probability of being reached by trajectories during the first day of atmospheric transport from the site. To construct the MPIZ indicator, all endpoints of 12-h intervals of calculated trajectories only during the first day (i.e., at 12 and 24 h) of transport were counted in the cells of a new rectangular gridded domain. Then, among all these cells, the cell in which the absolute maximum of intersections took place was identified as the ‘‘absolute maximum cell’’ (AMC). The number of intersections in other cells adjacent to AMC was
2.1. Trajectory modelling for risk sites Trajectory models are useful tools to evaluate airflow patterns within meteorological systems on various scales. These models use different kind of approaches to simulate temporal and spatial air parcel movement. In this study, an isentropic approach to model trajectories originating over the Ignalina NPP region has been selected. Although it uses the assumption of adiabatically moving air parcels and neglects various physical effects, it is still a useful research tool for such kind of studies. It should also be noted that some uncertainties in trajectory models and, hence, calculated trajectories are related to the interpolation of the original meteorological data, assumptions of vertical transport, numerical scheme errors, selection of initial points for trajectory calculation, errors of forecasted and analyzed fields, etc. In this study, the original National Center for Environmental Prediction (NCEP) gridded wind fields were interpolated to potential temperature levels (or isentropic surfaces) for a period of 6 years (1991 –1996). Then, the wind fields on isentropic surfaces were used to calculate (based on a technique described by Merrill et al., 1985) the 5-day forward isentropic trajectories originating over the RS region. The gridded domain selected for this study is located between 20jN – 85jN and 60jW – 127.5jE. The trajectories (in a total of more than 280,000) were computed twice per day (at 00:00 and 12:00 Universal Coordinated Time (UTC)) at different potential temperature levels (in total 16). More details concerning the trajectory calculation for the nuclear risk sites can be found in Baklanov and Mahura (2001) and Mahura (2001). The quality of trajectory calculation is highly dependent on the original quality of the NCEP fields (2.5 2.5j latitude vs. longitude), and it may not adequately reflect a contribution of frontal passages and local terrain phenomena. However, the trajectory errors arising from a single calculation might be smoothed over in further statistical analysis due to the large number of trajectories computed for the multiyear dataset.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
compared, and then assigned additional cells, which had less than 10% of difference between cells. Therefore, this new ‘‘region of maximums’’ represents the area (or region) with the highest probability of possible impact (AHPPI) from the risk site. Assuming a value of 100% for this region, the rest could be re-calculated as a percentage of AHPPI. For the MPIZ indicator, it was assumed that MPIZ includes all cells with an isoline of more than 90% of AHPPI, and then, an isoline of MPIZ was drawn through these cells. The second indicator—maximum reaching distance (MRD)—shows the farthest boundaries on a geographical map that might be reached during the first day by at least one trajectory originating over the RS location. To construct the MRD indicator, all endpoints of 12-h intervals of calculated trajectories during the first day (i.e., at 12 and 24 h) of atmospheric transport were used. An isoline of MRD was drawn through the grid cells of the rectangular grid domain where, at least, one trajectory intersected with the grid cell’s boundaries. It should be noted that although the likelihood that an air parcel will reach these boundaries is low, it is still a possible case of the most remote atmospheric transport from the risk site. The third indicator—typical transport time (TTT) field—shows (1) how long it will take for an air parcel to reach a particular geographical region during transport from the RS location, and (2) what territories would be at the highest risk during the first few days of contaminated cloud transport after an accident/event occurred at the risk site. To construct the TTT fields, the following procedure was applied. First, a new polar grid domain having 1260 grid
1065
cells—36 sectors (10j each) and 35 grid cells (2j each) along each sector line—with the risk site in the centre of domain was built. Second, for a particular term (for example, 12 h), the geographical coordinates of trajectories (latitude and longitude) were converted into the polar coordinates: radius and angle. Third, in each sector, a number of trajectory intersections was counted in each grid cell, and then compared among the grid cells along the sector line to find a cell with the absolute maximum of trajectory intersections (similarly called the AMC). For simplicity, if several AMC cells were identified along the sector line, then only the one closest to the risk site was selected. Finally, the TTT isoline for a particular term was drawn through AMCs in each of the 36 sectors (coordinates of the AMC centres were again converted from the polar into geographical coordinates). A similar procedure was applied for each term of 0.5, 1, 1.5 days, and so on.
3. Results and discussion In emergency response systems for accidental releases of radioactivity, an estimation of the typical transport time, boundaries of possible maximal contamination, and geographically farthest territories reachable by a contaminated cloud during a limited time atmospheric transport from the risk sites to/over a particular geographical territory, region, country, city, etc., is one of the important input parameters for the decision-making process. In this section, based on the example of the Ignalina NPP, the indicators, which could
Fig. 1. Annual boundaries of the maximum possible impact zone (—INP_MPIZ—) and maximum reaching distance (—INP_MRD—) indicators for the Ignalina NPP (INP).
1066
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
allow one to perform such an estimation and which could provide useful input data for the planning and decisionmaking processes, are presented and evaluated. 3.1. Maximum possible impact zone The MPIZ indicator is an important integral characteristic of the RS possible impact during the first day of atmospheric transport after an accident/event. For the Ignalina NPP, the annual boundary of the MPIZ indicator is shown in Fig. 1. It should be noted that, although MPIZ is concentrated in the RS vicinity, the configuration of the MPIZ boundary (or isoline) depends strongly on the dominating transport patterns during the first day of atmospheric transport from the site, and this isoline is extended along the main flow direction. For studies on the long-term consequences of routine discharges or accidental releases at RS, it is important to evaluate not only the geographical boundary of MPIZ, but
also area of MPIZ. To estimate such area, a simple approach could be used. First, a figure inside the MPIZ isoline is approximated by a set of triangles, where each triangle has a top at the RS coordinates and a side on the MPIZ isoline. Second, the areas of all triangles are calculated, and the total area under the MPIZ isoline was simply equal to a sum of areas of all triangles. For the Ignalina NPP, the annual area of MPIZ is equal to 42 104 km2. Its boundaries are more extended in the southeast sector from the site (between 24 –34jE and 53– 57jN), and include the eastern territories of Latvia and Lithuania, central and northern regions of Belarus, as well as, partially, territories of several Russian regions: Pskov, Bryansk, Smolensk, and Tver. Analysis of the MPIZ seasonal variability showed dependence on the wind velocities on the synoptic- and meso-scales throughout the year. On a seasonal scale (Fig. 2), the MPIZ area reaches its highest rate during winter (57 104 km2) and the lowest during fall (32 104 km2). During winter, the MPIZ boundary is more
Fig. 2. Seasonal boundaries of the maximum possible impact zone (—INP_MPIZ—) and maximum reaching distance (—INP_MRD—) indicators for the Ignalina NPP (INP) during (a) spring, (b) summer, (c) fall, and (d) winter.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
extended in the eastern direction from the site passing over 35jE and reaching the borders of Ukraine and Estonia. In all seasons, except spring (when MPIZ is more extended to the west of the site), the MPIZ boundaries are more pronounced in the southeast sector from the site. 3.2. Maximum reaching distance The maximum reaching distance (MRD) indicator, similar to the maximum possible impact zone indicator, is an important characteristic of the RS possible impact during the first day of atmospheric transport following an accident/ event. The boundary of the MRD indicator for the Ignalina NPP is shown in Fig. 1. There is a possibility that contaminated air parcels may reach the remotest geographical regions after the hypothetical accidental release. If one considers a large dataset of trajectories, the distance to these remote region boundaries could be in thousands of kilometres. The boundary for the MRD indicator depicted in Fig. 1 reflects such possibility. They could be interpreted as possible boundaries of the RS impact during the first day of atmospheric transport after the hypothetical accident/event. Applying a similar estimation of areas enclosed by the MRD isolines (as for the MPIZ), it was found that for the Ignalina NPP, the annual average MRD area is equal to 703 104 km2, with a maximum level in winter (843 104 km2) and a minimum level in summer (540 104 km2). The annual MRD boundary is extended significantly (almost twice) in the eastern direction from the site (passing over the 50jE longitude and almost reaching the Caspian Sea)
1067
compared with the western direction. On a seasonal scale (Fig. 2), during winter, the MRD boundaries are extended farther to the southeast as well as they pass over the Kola Peninsula compared with other seasons. During all seasons, except summer, the MRD western boundaries are extended farther to the west than Denmark. The annual MRD area is significantly larger (almost 17 times) than the MPIZ area, and the latter is also included in the MRD area. The dominating direction of atmospheric transport is less underlined by the MRD indicator compared with MPIZ. Because a construction of the MRD isolines includes cells of gridded domain with the lowest probabilities of atmospheric transport through these cells (i.e., at least one trajectory reached a cell), it is not possible to explain such temporal variability through the synoptical scale features alone (i.e., there is a contribution of the global scale patterns, as well as the fact that there might be anomalies in the transport patterns from the site from year to year). Additionally, it should be noted that simultaneous plotting of both—MPIZ and MRD—indicators allows one to estimate the order of difference (1) between the boundaries of possibly highly impacted territories and the farthest reaching boundaries during atmospheric transport from the site, and (2) between the highest (MPIZ) and the lowest (MRD) probabilities of the RS possible impact on the geographical territories of interest. 3.3. Typical transport time The typical transport time (TTT) fields show (1) the time it will take for an air parcel to typically reach a particular
Fig. 3. Annual typical transport time fields at 1 ( – 1d – ) and 2 ( – 2d – ) days of atmospheric transport for the Ignalina NPP (INP).
1068
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
geographical region from the risk site location, and (2) the territories that would be at highest risk during the first few days of radionuclide cloud transport after an accident at the risk site. On an annual scale, for INP (Fig. 3), the typical transport time to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, and the western border regions of Russia, and central aquatoria of the Baltic Sea. On the second day, the atmospheric transport from the site is relatively faster (i.e., isoline for 2 days is more—almost three times—extended in the eastern direction compared with the western direction). The typical transport time to reach territories of the Czech Republic, Germany, Denmark, Kazakhstan, and Romania is 2 days. During this time, the atmospheric transport could typically occur over the Ukraine, Poland, Central and Northwest Russia, Moldova, as well as Sweden and Finland adjacent to the Baltic Sea. The difference in distance between firstand second-day isolines is similar in both southern and northern directions, but it is 2 –3 times higher in the eastern direction compared with the western direction. On a seasonal scale, analysis of the TTT fields by Mahura and Baklanov (2002) showed some peculiarities. During winter, the TTT field is significantly shifted in the northeast sector from the site, almost reaching 40jE by the end of the first day, and 50jE by the end of the second day. During summer, the TTT field is almost twice more extended in the north – south direction compared with the east – west direction. Detailed analysis of the TTT fields for the nuclear risk site allows the identification of geographical regions and territories of the neighbouring countries, which might be potentially reached by atmospheric flows and might be impacted by radioactive pollution during a selected period of time, if an accident/event occurred at the risk site. This information could be used to forecast the arrival of the contaminated cloud to a certain territory, and to calculate, if necessary, doses to the population, and hence, allowing authorities to plan countermeasures, including informing and, in a critical scenario, evacuating the local population. It should be noted that construction of the TTT fields for later terms (more than 3 days) is complicated due to significant airflow propagation from the RS location, and hence, the later term isolines are not concentrated around the site and are less representative.
4. Conclusion In this paper, following a methodology developed within the ‘‘Arctic Risk’’ Project (AR-NARP, 2001 –2003), several probabilistic indicators to evaluate the risk site (RS) possible impact on geographical regions, territories, countries, etc., due to atmospheric transport from the RS region are
suggested. These indicators—maximum reaching distance, maximum possible impact zone, and typical transport time—were constructed by applying statistical methods and using a dataset of isentropic trajectories originated over the Ignalina NPP (selected as a risk site for this study) region during 1991 –1996. The first indicator is maximum possible impact zone (MPIZ) and it outlines the boundaries of territories with the highest probability of being reached by trajectories during the first day of atmospheric transport from the RS location. The second indicator is maximum reaching distance (MRD) and it depicts the farthest boundaries on the geographical map that might be reached during the first day by, at least, one trajectory originating from the RS location. The third indicator is typical transport time (TTT) field and it shows (1) the time it will take for an air parcel to reach a particular geographical region during transport from the risk site location, and (2) what territories would be at the highest risk during the first few days of the contaminated cloud atmospheric transport following an accident/event at the risk site. For the Ignalina NPP, the annual areas enclosed by the MPIZ and MRD isolines are equal to 42 10 4 and 703 104 km2, respectively. The MPIZ boundaries are more extended in the southeast sector from the site and include, in particular, Latvia, Lithuania, Belarus, and several western regions of Russia. The MRD boundaries are twice extended in the eastern direction from the site (reaching the Caspian Sea) compared with the western direction. The annual TTT to reach the southern territories of Sweden and Finland, northern regions of Ukraine, and northeast of Poland is 1 day. During this time, the atmospheric transport could typically occur over the Baltic States, Belarus, western border regions of Russia, and central aquatoria of the Baltic Sea. Detailed analysis of temporal patterns for these indicators showed the importance of seasonal variability. These indicators are additional information, which is an important factor for planning of operational forecast of radioactive situation after accident/event based on dispersion modelling for emergency preparedness. Furthermore, they are important input data for studies on the social and economic consequences of the RS impact on population and environment for neighbouring countries as well as for the multidisciplinary risk and vulnerability analysis, and probabilistic assessment of the meso-, regional-, and long-range transport of radionuclides. These suggested indicators are important characteristics of the RS impact prediction reflecting only the atmospheric transport, and they focus on optimization of the regional monitoring and emergency preparedness systems. In order to evaluate possible consequences for population and environment on local and regional scales, a series of appropriate models can be applied, for example, MACCS, COSYMA, ECOSYS, ECOMARC, and others. Such evaluation of consequences might be simplified if only territories, first of all, within the boundaries of the MPIZ indicator are considered with respect to selected RS.
A. Mahura, A. Baklanov / Environment International 29 (2004) 1063–1069
It should be also noted that modelling the selection of terms—every 12 h—depends on the temporal (interval of 12 h) and horizontal (2.5j 2.5j latitude vs. longitude) resolution of meteorological data. However, the linear interpolation of all indicators could be done for finer time intervals (for example: 1, 3, 6 h, and so on); it would be better to use a finer resolution meteorological data for modelling purposes. Any other model and meteorological dataset of finer resolution could be used too, for example: within the ‘‘Arctic Risk’’ Project (AR-NARP, 2001 – 2003) the ECMWF and DMI-HIRLAM datasets and 3D DMI trajectory model (Sørensen et al., 1994) are used for indicators of the RS possible impact based on the dispersion modelling approach.
Acknowledgements The authors are grateful to Drs. Leif Laursen and Jens Havskov Sørensen (Danish Meteorological Institute, Copenhagen), Ronny Bergman (Swedish Defence Research Authority, Umea˚), Olga Rigina (Geographical Institute, University of Copenhagen), Boris Segersta˚hl (Thule Institute of University of Oulu, Finland), John Merrill (University of Rhode Island, Graduate School of Oceanography, Narragansett, RI, USA), Frank L.Parker (Vanderbilt University, Nashville, TN, USA), Keith Compton (International Institute for Applied Systems Analysis, Laxenburg, Austria), Sven Nielsen (Risø National Laboratory, Denmark), and Steen C. Hoe (Danish Emergency Management Agency, DEMA) for collaboration, discussions and constructive comments. The authors are also thankful to an anonymous reviewer for critical comments and constructive suggestions. The computer facilities at the Danish Meteorological Institute (DMI) and National Center for Atmospheric Research (NCAR, Boulder, CO, USA) have been used extensively in the study. The NCAR meteorological data were used as input in the trajectory modelling. Thanks are also due to the computer consulting services at DMI and NCAR, especially, to Jess Jørgensen (HIRLAM group, DMI), Peter Teisner (Computer Services, DMI), and Ginger Caldwell (Scientific Computing Division, NCAR) for the collaboration, computer assistance, and advice. Financial support of this study included the grants of the Nordic Arctic Research Programme (NARP), Nordisk Forskerutdanningsakademi (NorFA), US Department of Energy (US DOE).
1069
References AR-NARP. On-going Project Atmospheric Transport Pathways, Vulnerability and Possible Accidental Consequences from the Nuclear Risk Sites in the European Arctic (Arctic Risk) of the NARP: Nordic Arctic Research Programme. WWW site: http://www.dmi.dk/f + u/luft/arcticrisk/main.html; 2001 – 2003. Baklanov A, Mahura A. Atmospheric Transport Pathways, Vulnerability and Possible Accidental Consequences from Nuclear Risk Studies: Methodology for Probabilistic Atmospheric Studies. Danish Meteorological Institute Scientific Report, 01-09, ISBN: 87-7478-450-1; 2001. 43 pp. Baklanov A, Mahura A, Jaffe D, Thaning L, Bergman R, Andres R. Atmospheric transport patterns and possible consequences for the European North after a nuclear accident. J Environ Radioact 2002a;60(1 – 2): 1 – 26. Baklanov A, Rigina O, Mahura A. Nuclear risk and vulnerability in the Arctic: new method for multidisciplinary assessments. In: Strand P, Jølle ˚ , editors. Proceedings of the 5th International Conference on T, Sand A Environmental Radioactivity in the Arctic and Antarctic, St. Petersburg, Russia, 16 – 20 June. Oslo, Norway: NRPA; 2002b. p. 397 – 400. Baklanov A, Mahura A, Rigina O, Sørensen JH. Nuclear risk and vulnerability: approach for multidisciplinary assessments. In: Strand P, Børretzen P, Jølle T, editors. Proceedings of the International Conference on Radioactivity in the Environment, Principality of Monaco, 2 – 5 September. Oslo, Norway: NRPA; 2002c. p. 87 – 92. Baklanov A, Mahura A, Sørensen JH. Methodology for evaluation of consequences of possible accidental atmospheric releases of hazardous matter: ‘‘Kursk’’ submarine study. Atmos Chem Phys Disc 2003;3:1 – 42. Mahura A. Probabilistic assessment of atmospheric transport patterns from nuclear risk sites. PhD Phys. and Math. thesis, Russian State Hydrometeorological University/Kola Science Center. Russian Academy of Sciences; 2001. 172 pp. Mahura A. Assessment of Impact of Russian Nuclear Fleet Operations on Russian Far Eastern Coastal Regions. IIASA interim report IR-02-004; January 2002. 76 pp. Mahura A, Baklanov A. Probabilistic analysis of atmospheric transport patterns from nuclear risk sites in Euro-Arctic Region. Danish Meteorological Institute Scientific Report, 02-15. ISBN: 87-7478-469-2; 2002. 87 pp. Mahura A, Andres R, Jaffe D. Atmospheric transport patterns from the Kola nuclear reactors. CERUM Northern Studies, vol. 24. Sweden: Umea˚ University; 2001. 33 pp. Mahura A, Baklanov A, Rigina O, Parker FL. Statistical analysis of atmospheric transport from the nuclear risk sites in the Arctic Region. In: ˚ , editors. Proceedings of the 5th International Strand P, Jølle T, Sand A Conference on Environmental Radioactivity in the Arctic and Antarctic, St. Petersburg, Russia, 16 – 20 June. NRPA; 2002. p. 119 – 23. Mahura A, Baklanov A, Sørensen JH. Methodology for evaluation of possible consequences of accidental atmospheric releases of hazardous matter. Radiat Prot Dosim 2003;103(2):131 – 9. Merrill J, Bleck R, Boudra DB. Techniques of Lagrangian trajectory analysis in isentropic coordinates. Mon Weather Rev 1985;114:571 – 81. Sørensen JH, Laursen L, Rasmussen A. Use of DMI-HIRLAM for operational dispersion calculations. In: Gryning SE, Millan MM, editors. Air Pollution Modelling and Its Application, vol. X. New York: Plenum Press; 1994. p. 373 – 81.