atmospheric modeling of radioactive material dispersion and health risk in fukushima daiichi nuclear...

Annals of Nuclear Energy 53 (2013) 197–201

Contents lists available at SciVerse ScienceDirect

Annals of Nuclear Energy

journal homepage: www.elsevier .com/locate /anucene

Atmospheric modeling of radioactive material dispersion and health risk inFukushima Daiichi nuclear power plants accident

Tae Ho Woo ⇑Department of Nuclear Engineering, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 151-742, Republic of Korea

a r t i c l e i n f o

Article history:Received 6 April 2012Received in revised form 30 August 2012Accepted 8 September 2012Available online 28 November 2012

Keywords:Radioactive material dispersionRadioactive concentrationsNuclear power plantsAccident

0306-4549/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.anucene.2012.09.003

⇑ Tel.: +82 2 880 8337; fax: +82 2 889 2688.E-mail address: [email protected]

a b s t r a c t

The radioactive material dispersion is investigated in terms of the radioactive concentrations. The riskof the radioactive hazard material is important with respect to the public health. The prevailing wester-lies region is modeled for the dynamical consequences, whereby the Fukushima nuclear disaster inJapan is modeled. The multiplications effects of the wind values and plume concentrations areobtained. Monte Carlo calculations are performed for wind speed and direction. In Seoul and Pusan,Korea, the Cs-137 has the highest value among the chemical radioactive materials Cs-137, I-131, andSr-90. The time for highest concentration is shown to be around 48th hour in Seoul and 12th hourin Pusan. Cesium has the highest value in both cities, and iodine has the lowest value in both cities.The wind is assumed to determine the direction of movement. Therefore, the real values are believedto be lower than the calculated results. This modeling could be used for other industrial accident casesin chemical plants.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

After nuclear power plants (NPPs) accident in Fukushima,Japan, the radioactive fallout has dispersed throughout the world.Especially, the adjacent countries had a tendency to be damagedeasily. With respect to the Fukushima NPPs accident, Korea, whichis the west part of the Japan, is modeled in this study. The prevail-ing westerlies region is not affected directly by the atmosphericpollutant according to the atmospheric consideration. In the realsituation, the convection of the thermal air and global air circula-tion could affect the western part of the accident region. In fact,in the Fukushima accident, the radioactive contaminated falloutshad reached Korea by some rains. The air stream of the radioactivefallouts had come in from the southern regions by the air circula-tions. So, the air stream has too many uncertainties to predict ex-actly. It is calculated by random number generations in MonteCarlo method for the wind directions. The computational simula-tion of the north and south wind could give the preparations forthe possible fallout for the NPPs accident site in real situations.The Monte Carlo simulation can give us the numerical values ofthe nuclear fallout possibility, where the random number is usedof the decision of the radioactive material quantities and the winddirections. The final purpose of the study is to increase the reliabil-ity of the safety for the national standard in the nuclear accident.

ll rights reserved.

In literature review, Periànez studied a numerical three dimen-sional model to simulate the transport of Cs and Pu by the RhoneRiver plume (Periànez, 2004). That is, this model solves thehydrodynamic equations, including baroclinic terms (that accountfor density variations) and a turbulence model, the suspendedmatter equations, including several particle classes simulta-neously, settling, deposition and erosion of the sediment, andthe radionuclide dispersion equations. In addition, the investiga-tion was carried out to reveal the impact of solar radiation onpollutant dispersion in different urban street layouts using com-putational fluid dynamics (CFD) technique (Xie et al., 2005). Forsimulating the quantitative effects of regional biomass alterna-tives for energetic purpose (BfE) on air pollutant emissions, adynamical model was developed and applied for the Eu RegionAustrian–Hungarian cross-border area. The dynamic simulationprogram Vensim was used to build an overall regional modelwith economic, social and environmental sectors (Szarka et al.,2008). In addition, the National Atmospheric Release AdvisoryCenter (NARAC) has been served as a national resource for theUnited States, providing tools and services to quickly predictthe environmental contamination and health effects caused byairborne radionuclides, and to provide scientifically based guid-ance to emergency managers for the protection of human life.The NARAC was developed for the capabilities to respond todifferent types of release events (Bradley, 2007). The 2nd sectionexplains the method of the study. The 3rd section describes re-sults of the study. There are some conclusions in the 4th section.

http://dx.doi.org/10.1016/j.anucene.2012.09.003

mailto:[email protected]

http://dx.doi.org/10.1016/j.anucene.2012.09.003

http://www.sciencedirect.com/science/journal/03064549

http://www.elsevier.com/locate/anucene

JAPAN

KOREA

Pusan

~1170 kmSeoul Fukushima

~1030 km

Fig. 1. Distance between Fukushima site and Seoul/Pusan of Korea. This shows thedistance between Fukushima site and Seoul/Pusan of Korea where the possiblewind is described as the reddish arrow lines. That is, the air is coming in Korea bynorthwest and southwest directions. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

Table 1Comparisons for mean speed (m/s) and most frequency direction from 2001 to 2010.

Seoul Pusan

2010 25, WNW 33, NNE2009 24, WNW 34, NE2008 24, WNW 32, NE2007 24, WNW 33, NNE2006 24, WNW 31, NNE2005 25, WNW 31, NNE2004 24, WNW 31, NNE2003 20, NE 32, NNE2002 21, W 39, ENE2001 18, W 36, ENE

Table 2Comparisons for mean speed (m/s) and most frequency direction of averaged values.

Seoul Pusan

Speed (m/s) 22.9 33.2Direction 0.4 (WNW) 0.26 (NE)

198 T.H. Woo / Annals of Nuclear Energy 53 (2013) 197–201

2. Method

For the simulations of the radioactive decay in the nuclear fall-outs, it is assumed the wind is incorporated with the radioactivedecay. The configurations are described for the atmospheric dis-persions. The Gaussian plume model is applied, which is used forthe pollution dispersion near source. Therefore, the radioactivefallouts flow to the modeled cities (Seoul and Pusan) simply by thismodel where the linear distance is assumed for the flying distance.In addition, the wind velocities are assumed by the measured val-ues of two cities. The Gaussian plume dispersion model is writtenas follows.

Cðx; y; zÞ ¼ Q2puryrz

e� y2

2r2y e

�ðzþHÞ2

2r2z þ e

�ðz�HÞ2

2r2z

!ð2:1Þ

where Q (g/s): pollution rate emission rate, u (m/s): average windspeed, ry (m): y direction plume standard deviation, rz (m): z direc-tion plume standard deviation, y (m): y position, z (m): z position, H(m): effective stack height.

This is the solution for the plume contaminant concentration ata point in space. For the z = 0, the concentration is shown at theground level as follows:

Cðx; y;0Þ ¼ Qpuryrz

e� y2

2r2y e� H2

2r2z ð2:2Þ

Also, at y = 0, the concentration is shown at the ground levelalong the plume centerline as follows:

Cðx;0;0Þ ¼ Qpuryrz

e� H2

2r2z ð2:3Þ

In addition, if the emission source is at the ground level,

Cðx;0;0Þ ¼ Qpuryrz

ð2:4Þ

Therefore, the concentration of the plume contamination in theground is proportional to the pollution rate emission rate overwind velocity. The Qo is the initial value of the radioactive decaymaterial, which is assumed as 100. So,

Q ¼ Qoe�kt ¼ Q oe� ln 2

t1=2t ð2:5Þ

Therefore,

C � Qu¼ Qoe�kt

u¼ Q oe

� ln 2t1=2

t

uð2:6Þ

Fig. 1 shows the simplified configuration of the map of thegeographic positions. The possible fallouts come from the northand south wind in Korea. There are measured wind speeds anddirections in Korea in Tables 1 and 2 (Korea MeteorologicalAdministration, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008,2009, 2010). The azimuthal wind is numbered which are shownin Fig. 2. The values are from 0.1 to 0.9, which are generated byrandom numbers. The highest value is 0.9 of south wind andthe lowest value is 0.1 of north wind. This value is multipliedto the concentration of the radioactive material for the windadjustment. If the wind goes directly from Fukushima to Seoul,the time for reaching to Seoul, Korea is shown as follows:

1;170;000 m=22:9 ðm=sÞ ¼ 51; 091:703 s ¼ 14:192 h ð2:7Þ

This is assumed that this wind flows directly from Fukushima toSeoul. For the case of Pusan:

1;030;000 m=33:2 ðm=sÞ ¼ 31;024:096 s ¼ 8:618 h ð2:8Þ

The equation is changed as follows:

C � Qu¼ Q oe�kt

u¼ Q oe

� ln 2t1=2ð51;091:703 sÞ

ð22:9 m=sÞ ð2:9Þ

Then, for the case for Cs-137, T1/2 = 30.17 years.

C � Qu¼ Q oe�kt

u¼ Q oe�

ln 2ð30:17 yearÞð51;091:703 sÞ

ð22:9 m=sÞ ð2:10Þ

In the case of I-131, T1/2 is 8.0197 days. Also, in the case of Sr-90,T1/2 is 28.8 years.

As it is seen in the Eq. (2.10), the concentration depends on thehalf-life of each nuclide. It is very difficult to decide the source termin the accident. The failure of the protection systems in the NPPs

(a)

(b)

0.1

0.5

0.9

0.5

0.3

0.7

0.3

0.7

0.2

0.4

0.6

0.80.8

0.6

0.4

0.2

N

E

S

W

NE

SE

NW

SW

NNE

ENE

ESE

SSESSW

WSW

WNW

NNW

Fig. 2. Directions of wind: (a) azimuthal angle, and (b) numeric value. This meansthe directions of winds by numbering. The highest value is 0.9 of south wind andthe lowest value is 0.1 of north wind. This is used to adjust the atmosphericconcentration.

T.H. Woo / Annals of Nuclear Energy 53 (2013) 197–201 199

initiates the dispersion of radioactive fallouts. It is impossible to findout the exact released material. Although the final protection ofcontainment is exploded, there are several structures in the sys-tems, which can still make a role of the radiation protections. In thisstudy, the source term Qo is assumed as 100 in all cases as it is saidabove. Hence, the calculations depend on the decay time excludingof the source term variations. If one uses the source term model, thecalculations could be changed. There are some reviews of the mod-eling of sources. Cervone et al. used the Second-order Closure Inte-grated Puff (SCIPUFF) where the Lagrangian puff dispersion modelusing Gaussian puffs is used (Cervone and Franzese, 2012). The sim-ulation is performed in every 1 h. The concentration C at each sam-pler i and time t from all the releases are calculated as,

Cit ¼ Ri

1t þ Ri2t þ � � � þ Ri

nt ð2:11Þ

where RIxt is the concentration for release x measured at t at location

i.So, the optimization is done that the vector w minimizes the er-

ror between the simulated and observed values at all locations andall time steps.

E ¼X

i;t

ðw1 � Ri1t þw2 � Ri

2t þ � � � þwn � Rint � Ci

otÞ ð2:12Þ

where CIot is the observed concentration at t and location i.

In addition, the Kosovic et al. showed the probabilistic approachwith Bayesian inference and optimization using Genetic algorithm(Kosovic et al., 2012). In this work, the flexible framework isachieved by the probabilistic study and single best solution is ob-tained by the Genetic algorithm. Also, Rodriguez et al. worked onadjoint approach to the source term estimation (Rodriguez et al.,2012). The utilizing of information on uncertainty in observationsis done to be the cost function visualization/scaling and the sourceterm estimation (STE) variable uncertainty relations. The uncer-tainty mapping is performed by the physics-based method (uncer-tainties are inputs to constrain adjoint) and the ensemble-basedmethod (adjoint to define the initial uncertainty).

For the calculations with the generated random numbers, theVensim code system is used. Vensim is used for developing, analyz-ing, and packaging high quality dynamic models (Vensim, 2009).The modeling is constructed graphically or in a text editor. Thefeatures include dynamic functions, subscripting (arrays), Monte–Carlo sensitivity analysis, optimization, data handling, applicationinterfaces, and more (Vensim, 2009).

3. Results

The objective of the study is to find out radioisotope concentra-tions in Seoul and Pusan, Korea. The wind speed is assumed as con-stant from the source point to the interested cities. Fig. 3 shows theradioactive concentration for the hazard materials by time. Thesimulations are obtained by the Monte Carlo calculations of windspeed and direction. In Seoul, the Cs-137 has the highest valueamong the chemical radioactive materials of Cs-137, I-131, andSr-90. Also, the highest value is in Cs-137 of Pusan. The time forhighest concentration is shown around 48th hour in Seoul and12th hour in Pusan. Cesium has the highest value than the otherelements in both cities. Furthermore, Iodine has the lowest valuein both cities. The wind is assumed as the directed movement.So, the real value is lower than the results. Table 3 shows the radio-active concentration in the time of the reached cities. The initialvalue of fallout is assumed as 100. In the quantification, the valuesare dimensionless. This mean that the uncertainty of the atmo-spheric dispersion could be expresses by the dimensionless num-ber where the relative value can give us the significance of theradioactive hazard. The comparative values are shown in the calcu-lations. The diluted values by atmospheric dispersions are shownin the figures. The other cities could be calculated by the modifica-tion of the distance from the accident site.

4. Concluding remarks

Following the worst nuclear disaster in Fukushima, Japan, it wasvery difficult to predict of the radioactive material dispersion.Therefore, the random sampling of the Mont Carlo method couldshow the significance of the radioactive material dispersion in-stead of using the exact values of the radiation concentration, be-cause the exact estimation is impossible due to the uncertaintiesof the atmospheric movements. The mathematical model for theradioactive concentration is investigated for the radiation disper-sion chemistry. I-131 has comparatively a higher value. So, theradiological hazard against thyroid is very dangerous, because thisradioisotope accumulated in the organ. In the case of NPPs acci-dent, usually, the iodine radioisotope is distributed to the commu-nities. In the case of the Three Mile Island NPPs accident in theUnited States, the government distributed the iodine containingpills to the community. In the Fukushima case, the Japanese gov-ernment distributed among the residents of the accident site thepills. This study can show the dynamical importance of the hazardradioisotope material by simulations. Although there are some

Atmospheric Concentration Atmospheric Concentration



2,000

1,500

1,000

500

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)

Atmospheric Concentration : Current

1,000

750

500

250

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)


2,000

1,500

1,000

500

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)


800

600

400

200

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)


800

600

400

200

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)


800

600

400

200

0

0 10 20 30 40 50 60 70 80 90 100

Time (Hour)


(a)

(c) (d)

(b)

(e) (f)Fig. 3. Radioactive concentration for isotopes: (a) Cs-137 (Seoul), (b) I-131 (Seoul), (c) Sr-90 (Seoul), (d) Cs-137 (Pusan), (e) I-131 (Pusan), (f) Sr-90 (Pusan). This shows theradioactive concentration for isotopes Cs-137, I-131, and Sr-90 in Seoul and Pusan.

200 T.H. Woo / Annals of Nuclear Energy 53 (2013) 197–201

assumptions as the Gaussian plume model, the linear distance, andthe wind direction–velocity, this work focuses on how to treat theuncertain situation of the radioactive fallout concentrations. Thismodeling could give the importance of the radioactive materialdamages to the public dynamically. The time is one factor of radi-

ation protections which include distance and shielding. Hence, thesignificance of the time based simulation is shown as the environ-mental radiation matter where the shielding is nearly impossibleto the huge public numbers. There are several significant resultsof the study as follows;

Table 3Comparisons for radioactive values.

Seoul Pusan

Cs-137 6.69729 1.14922I-131 6.36823 1.11260Sr-90 6.69728 1.14922

T.H. Woo / Annals of Nuclear Energy 53 (2013) 197–201 201

� The numerical values of the radioactive concentrations areshown as the dynamical values.

� The prevailing westerlies region of the nuclear accident isassumed to take the directly flown fallouts from east.

� Possible nuclear accident is simulated in the aspect of theatmospheric contaminations.

� The dangerous situation of the hazard radioisotope could beinformed to the public.

� In a further study, some other variables can be considered inthe modeling.

� The public health condition could be enhanced by the inves-tigation of atmospheric radioactive hazard situation.

This modeling is applicable to other industry cases like thechemical plant accidents. For example, the toxic material couldbe simulated by the meteorological quantifications where the localdata should be considered for the reliability of the prediction. Themeasurement of the concentration in the chemical toxic elementscould be compared with the estimated simulations. There is thenext step of the simulation work where the mathematical calcula-tion can give the public the outline of the contaminated area. Then,the air flows of the chemical compounds are detected by theinstrumentations. Then, the corrections of the air flow modelingare performed by the computational procedures. Hence, the dan-gerous material can be controlled by the theoretical and measure-ment combinations.

Considering the HYSPLIT (HYbrid Single-Particle LagrangianIntegrated Trajectory) model, this is a complete system for com-puting simple air parcel trajectories to complex dispersion anddeposition simulations (Air Resources Laboratory, 2012). The ini-tial development was a result of a joint effort between NationalOceanic and Atmospheric Administration (NOAA) and Australia’sBureau of Meteorology. However, there is a limitation of the simu-lation of the HYSLIPT in which the modeling is done mainly on thepath simulations. It is weak for the simulation of the space disper-sion modeling. Especially, the radiation dispersions have manyuncertainties in the atmospheric space. The wide area orientedsimulations are needed in the NPPs accident. Therefore, the studiedmodeling in this paper is easy to control the designed assumptions.

In the radioactive material estimation, the prediction of the airflow is extremely important, because there are several particularcharacteristics in radioactive material. That is, the radioactive par-ticle can move very fast by air flow. So, the energy of radionuclide

is one of critical point of the movement. The time is proportional tothe severity of the radioactive material due to the half-life of thematerial. So, the distance between the interested area and the acci-dent site can affect to the safety in radiation hazard. Following thedistance, the radiation defense method could be decided. Forexample, the artificial rain could be one of possible method to di-lute the concentrations of the radioactive material dispersions.Hence, this study can give the basic information against the radia-tion fallout damage in the possible NPPs accident.

References

Air Resources Laboratory (ARL), 2012. HYSPLIT – Hybrid Single Particle LagrangianIntegrated Trajectory Model. National Oceanic and Atmospheric Administration(NOAA), Washington, DC.

Bradley, M.M., 2007. NARAC: an emergency response resource for predicting theatmospheric dispersion and assessing the consequences of airborneradionuclides. J. Environ. Radioactiv. 96, 116–121.

Cervone, G., Franzese, P., 2012. Source term estimation for the 2011 Fukushimanuclear accident. International Workshop on Source Term Estimation (STE)Methods for Estimating the Atmospheric Radiation Release from the FukushimaDaiichi Nuclear Power Plant, February 22–23, 2012, NCAR, Boulder, Colorado.

Korea Meteorological Administration, 2001. Annual Climatological Report, Seoul,Korea.










Kosovic, B., Young, G., Schmehl, K.J., Truesdell, D., Haupt, S.E., Annunzio, A.,Rodriguez, L., Aines, R.D., Belles, R.D., Dyer, K.M., Hanley, W.G., Johannesson, G.,Larsen, S.C., Mirin, A.A., Nitao, J.J., Sugiyama, G.A., Vogt, P.J., Chow, F.K.,Lundquist, J.K., Monache, L.D., 2012. Survey of evolutionary and probabilisticapproaches for source term estimation. International Workshop on Source TermEstimation (STE) Methods for Estimating the Atmospheric Radiation Releasefrom the Fukushima Daiichi Nuclear Power Plant, February 22–23, 2012, NCAR,Boulder, Colorado.

Periànez, 2004. The dispersion of 137Cs and 239,240Pu in the Rhone River plume: anumerical model. J. Environ. Radioactiv. 77, 301–324.

Rodriguez, L.M., Vandenberghe, F., Bieringer, P.E., Hurst, J., Weil, J., 2012. An adjointapproach for the Estimation of source terms for atmospheric releases.International Workshop on Source Term Estimation (STE) Methods forEstimating the Atmospheric Radiation Release from the Fukushima DaiichiNuclear Power Plant, February 22–23, 2012, NCAR, Boulder, Colorado.

Szarka, N., Kakucs, O., Wolfbauer, J., Bezama, A., 2008. Atmospheric emissionsmodeling of energetic biomass alternatives using system dynamics approach.Atmos. Environ. 42, 403–414.

Vensim, 2009. Vensim Simulation Software, Ventana Systems, Inc.Xie, X., Huang, Z., Wang, J., Xie, Z., 2005. The impact of solar radiation and street

layout on pollutant dispersion in street canyon. Build. Environ. 40, 201–212.

atmospheric modeling of radioactive material dispersion and health risk in fukushima daiichi nuclear...

Documents