207
Fukushimaprefectureexceededthislimitandsomeneighboring
prefecturessuchasMiyagi,Tochigi,andIbarakiarepartiallyclose
to the limit under our upper bound estimate (Movie S4) and,
therefore,local-scaleexceedanceislikelygiventhestrongspatial
variability of f 137Cs deposition. For those three prefectures,
detailed soil samplingis recommendedinthenearfuture.Esti-
matedandobservedcontaminationsinthewesternpartsofJapan
were not as serious, even though someprefectures were likely
effected to some extent(Fig.3,MovieS4,andTableS4).Con-
centrations in these areas are below 25Bqkg−1, which is far
below the threshold for farming. However, we strongly recom-
mend eachprefecturetoquicklycarry outsomesupplementary
soilsamplingsatcitylevelstovalidateourestimateseven ifthe
concentrations arelow.
TherelativelylowcontaminationlevelsoverwesternJapancan
bewellexplainedby theJapanesetopography.Theeasternand
northeastern partsofJapanaresurrounded bymountainranges
suchastheKanto,Echigo,andOhwumountainranges(Fig.S6)
(25), which, to a large extent, sheltered the northwestern and
westernpartsofJapanfromthedispersionofradioactivemateri-
al.Itisworthnoting,however,thatrelativelyhighercontamina-
tion levels can be seen over the Hida, Chugoku, and Shikoku
mountainranges(Fig.3,Fig.S6,andMovieS4),probablydueto
orographicenhancementofprecipitationand,thus,wetdeposi-
tionof137Cs.InHokkaido,to thenorthofJapan’smainisland,
bothloweraltitudeand higheraltitudessuchastheYubariand
Hidakamountain ranges areeffected by137Cs deposition,par-
tially dueto direct transport fromtheFukushima NPP via the
PacificOceanasshowninMoviesS1andS2andalsoassimulated
by another atmospheric transportmodel(12).
Weestimatethatatotal ofmorethan5.6and1.0PBq
137
Cs
were deposited over Japan and the surrounding ocean (130–
150°Eand30–46°N),andtheJapanIslandsinthisdomainonly,
respectively (Fig. 2A).Althoughtheestimatefor thelarger do-
mainisquiteuncertainbecauseitisconstrainedonlybymeasure-
ments in Japan, these numbers areconsistentwith a suspected
total releaseofabout12PBq
137
Cs (2).Mostofthedeposition
occurredoverthePacificOcean,yetsoilconcentrationsof137Cs
areabove100Bqkg−1 overlargeareasofeasternJapan(Fig.3).
Accordingtoourresults,foodproductionin easternFukushima
prefectureislikelyseverelyimpairedbythe137Csloadsofmore
than2;500Bqkg−1(upperlimitoffarming)andalsopartiallyim-
pactedinneighboringprovincessuchasIwate,Miyagi,Yamagata,
Niigata,Tochigi,Ibaraki,andChiba,wherevalues ofmorethan
250Bqkg−1 cannotbe excluded (Fig.3 andMovie S4).Notice
alsothatourestimatesarebasedonatransportmodeldrivenwith
meteorologicalanalysisdatafromaglobalmodel.Suchamodel
cannotfully capture all complexities oftheregional wind field
overJapanand,inparticular,doesnotresolvethehighspatiotem-
poralvariabilityofprecipitation.Therefore,weexpectthetruesoil
contaminationacrossJapantobeconsiderablymorevariablethan
inourestimate.Eveninregionswherewefindrelativelylowsoil
contaminationlevels,hotspotswithhighconcentrations(e.g.,due
to convective rain fall, orographic enhancement of rainfall, or
fine-grainsoilflowbyrainwaterontheground)maybepossible.
Incontrast,relativelyclean patchesmayalso bepresentinareas
with high overall contamination levels. Despite these shortcom-
ings,weexpectourresultstobeusefulforregulatorymeasuresand
for guidingmonitoringactivities towardareaswith expectedhigh
137Csburdens.Wehopethisstudywillcontributetounderstanding
thecontaminationissuein Japan.
MaterialsandMethods
Observations of Cesium-137 Deposition and Concentration in Soil in each
Prefecture.FromMarch18,MEXThasbeenobservingdailyradioactivitylevels
indepositioninmost of theprefecture(17).Theexactcoordinates of the
samplinglocationswereindividuallyaccessiblethroughourcontactstoMEXT
(TableS2).ThedepositiondatabetweenMarch18and19werenotusedin
ourestimatebecauseofnodepositionsatobservatoriesfromthemodeled
DRmapsasmentionedinthemaintext.Insomeprefectures,dataweremiss-
ingorunavailable[Miyagi,March18–April19(completelynoobservations);
Yamagata,March 29–April3;Fukushima,March 18–March26andApril4;
Gifu,March24,25,27,28,and30;Nara,March18–21andApril15–18;Oita,
March22–26).
FLEXPART and Estimated 137Cs Deposition. FLEXPART (9) is a Lagrangian
particledispersionmodel simulatingtransport, diffusion,dryand wet de-
position,andradioactivedecay of radioactivematerialssuch as131I,137Cs,
and133Xe(Seehttp://transport.nilu.no forfurtherdetailson FLEXPART).In
thisstudy,continuousemissionfromtheFukushimaDaiichiNPPwasassumed
after1800hourscoordinateduniversaltime(UTC) on March11,2011.The
simulationendedat0000hoursUTConApril20.FLEXPARTwasforcedwith
theEuropeanCenterforMedium-RangeWeatherForecasts(ECMWF)opera-
tionalanalysisdatawithaglobalresolutionof1°×1°and0.18°×0.18°for
120–168°Eand25–50°N.Theoutputhadaresolutionof0.2°×0.2°andwas
recordedevery3h(SIText).
For each day, we first normalized the modeled daily accumulated
depositionineachgridcellwiththemaximumaccumulateddepositionvalue
forthemodeldomain,hereaftercalleddailydepositionratio(DDR)maps:
DDR
ðx;yÞ
¼
1
FPD
max
i
∑
T
i¼1
FPD
ðx;yÞ
i
;
[1]
whereFPD
ðx;yÞi
isthethree-hourlymodeleddepositioningridcellðx;yÞand
FPD
max
isthemaximumdaily deposition valuefound in theentire model
domain.Tisthenumberofmodeloutputtimestepsperday(T¼8).Daily
griddeddepositionvaluesof137CswereestimatedbyscalingtheDDRmap
withavailabledailyobserved137CsdepositionsineachprefecturebyMEXT
(17) bythefollowingequation:
Depo
ðx;yÞ
¼
DDR
ðx;yÞ
N
∑
N
i¼1
Depo
iðObs:Þ
DDR
iðObs:LocÞ
;
[2]
whereDepo
ðx;yÞ
istheestimateddailytotal137Csdepositioningridcellðx;yÞ,
Depo
iðObs:Þ
istheobserved137Csdepositionatlocationi (TableS2),N≤47
isthenumberofavailablecountsonacertaindayinJapan’s47prefectures,
DDR
iðObs:Loc:Þ
is theDDR
ðx;yÞ
in the grid point where
137
Cs deposition was
observed,andDDR
ðx;yÞ
istheDDRingridcellðx;yÞ.Onlythecaseswithboth
theobserveddepositionandtheDDR
iðObs:Loc:Þ
notequaltozeroateachob-
servatory location n were used for counting g N on each day. Because the
Depo
iðObs:Þ
to DDR
iðObs:Loc:Þ
scaling factor in n Eq. 2becomes infinite when
thesimulatedDDRvalueisclosetozerobutdepositionisactuallyobserved,
aminimumpositiveDDRvalue,DRT,needstobeusedtoderivethescaling.
SeveralDRTsof0.001,0.005,0.007,0.01,0.05,and0.1forDDR
iðObs:Loc:Þ
within
thesimulation domainon each day wereusedto avoidabnormally high
Depo
ðx;yÞ
valuesduetodividing bysmallvalues(SIText).If DDR
iðObs:Loc:Þ
at
acertaingridpointwaslessthanaDRTvalue,DDR
iðObs:Loc:Þ
wassettothe
DRTvalue.
Forcomputingtotal137CsdepositionbetweenMarch20andApril19,we
correctedallvaluestoApril19usingahalf-lifeof137Csof30.1y(4).Thesum
ofallthedailyobservedorestimated137Csdepositionsisthetotal137Csde-
position(Fig.2andFig.S4A).
Observations on
137
Cs ConcentrationsinSoilandGrass.Forcomparisonwith
our estimates,measurementsof
137
Csconcentrations in soil or grasswere
used(SITextandTableS1).Meantransferfactorofsoil-to-grassof0.13,which
wasobtainedfromtheobservationsinJapanesesoilandgrass,wasusedto
convertgrasscontaminationtosoilequivalentcontamination(grasscontam-
inationdividedbythetransferfactor)(23).Thetimesandlocationsofthose
samplingsvaried.Tocoverthetimeperiodofourstudy(March20–April19),
wealsousedsomesoilsamplesfromlaterdates,butwedidnotuseanydata
afterMay19.Noticealsothatthesoilsampleswerealsoeffectedby137Cs
deposition before March 20(SI Text).Someobservatories measured total
cesiumconcentrationincludingboth137Csand134Cs.Inthatcase,weassumed
thathalfofthetotalCswas137Cs.
Toconvert the137Cs depositioninto soil concentration,soil depthand
densityinformationareneeded.However,itiscurrentlydifficulttoobtain
thisinformation acrossall of Japan. Thereisan empiricalrelationshipon
theratiobetween137Csconcentrationand deposition from0to5cmsoil,
paddysoil,andfieldsoilsamples(22)(Fig.S5).Weconsideredthemeanvalue
oftheratioasCCof5315kgm−2 reflectingthe5-cmdepthsoilinforma-
tion and its density. Our estimated CC valueis closeto the CC valueof
65kgm
−2
assumed by MEXT (26) with 5-cm soil and a soil density of
Yasunarietal.
PNAS ∣ December r 6, 2011 ∣ vol.108 ∣ no.49 ∣ 19533
ENVIRONMENTAL
SCIENCES
SEECOMMENTARY
