Paper
DOSES FROM EXTERNAL IRRADIATION TO MARSHALL
ISLANDERS FROM BIKINI AND ENEWETAK NUCLEAR
WEAPONS TESTS
André Bouville,* Harold L. Beck,† and Steven L. Simon*
Castle series, all of which were high yield. The measurement data used for those estimates of exposure were
collected by two methods: (1) stationary, ground-level
continuous reading Geiger-Müller type instruments with
paper strip chart recording mechanisms, and (2) aerial
surveys using fixed wing aircraft that carried scintillometer instruments. Most of the atolls of the Marshall
Islands, including all that had populations of significant size, were monitored in the aerial radiological
surveys in 1954 (Breslin and Cassidy 1955). The range
of estimated cumulative exposures from the Castle
series reported by Breslin and Cassidy (1955) covered
approximately five orders of magnitude, similar to the
range of 137Cs concentrations measured in the environment of the Marshall Islands by the Nationwide
Radiological Study conducted approximately 40 y
later (Simon and Graham 1994, 1997). The USAECplaced instrument on Rongerik Atoll was responsible
for alerting the U.S. military weather observers on
Rongerik to high levels of early fallout, leading to
their evacuation and to the evacuation of Marshallese
from Rongelap, Ailinginae, and Utrik following the
test Bravo in 1954 (Eisenbud 1987; Simon 2000).
Other than atoll-specific values for the external
exposure (reported in Roentgens or R) published in the
USAEC reports (Eisenbud 1953; Breslin and Cassidy
1955), and later estimates of external dose by Lessard et
al. (1985) for Rongelap and Utrik, few, if any, external
dose estimates have been reported for Marshallese. One
significant source of information on nuclear testing in the
Marshall Islands, a special issue of Health Physics
(Simon and Vetter 1997), was largely concerned with
land contamination, resettlement issues, and assessments
of doses received decades after nuclear tests were conducted in the Marshall Islands. Until the publication of
this paper, no systematic effort had been made to
estimate the annual doses from external irradiation,
received from 1948 to 1970, from all tests at all
inhabited atolls.
Abstract—Annual doses from external irradiation resulting
from exposure to fallout from the 65 atmospheric nuclear
weapons tests conducted in the Marshall Islands at Bikini and
Enewetak between 1946 and 1958 have been estimated for the
first time for Marshallese living on all inhabited atolls. All tests
that deposited fallout on any of the 23 inhabited atolls or
separate reef islands have been considered. The methodology
used to estimate the radiation doses at the inhabited atolls is
based on test- and location-specific radiation survey data,
deposition density estimates of 137Cs, and fallout times-ofarrival provided in a companion paper (Beck et al.), combined
with information on the radionuclide composition of the fallout
at various times after each test. These estimates of doses from
external irradiation have been combined with corresponding
estimates of doses from internal irradiation, given in a companion paper (Simon et al.), to assess the cancer risks among
the Marshallese population (Land et al.) resulting from exposure to radiation from the nuclear weapons tests.
Health Phys. 99(2):143–156; 2010
Key words:
137
Cs; dose, external; fallout; Marshall Islands
INTRODUCTION
DESPITE NUMEROUS efforts to monitor the Marshall Islands
for radioactivity during the United States Pacific nuclear
testing program and afterwards, there has been relatively
little effort towards estimating radiation doses to all
Marshallese exposed to the fallout from the testing. The
United States Atomic Energy Commission (USAEC)
issued a report on radiological surveys following Operation Ivy of 1952 (Eisenbud 1953) and Operation Castle
of 1954 (Breslin and Cassidy 1955). The latter report
estimated cumulative exposures from the tests of the
* Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892;
†
New York, NY 10014 (retired from U.S. DOE).
For correspondence contact: Steven L. Simon, National Cancer
Institute, National Institutes of Health, 6120 Executive Blvd., Bethesda, MD 20892, or email at ssimon@mail.nih.gov.
(Manuscript accepted 5 March 2010)
0017-9078/10/0
Copyright © 2010 Health Physics Society
DOI: 10.1097/HP.0b013e3181dc521d
143
144
Health Physics
The method of estimating external doses can be
based on either historical data of ground-level exposure
rates, or, alternatively, on data on the deposition density
of particular radionuclides contained in the fallout, such
as 137Cs, combined with information on the ratio of the
nuclide activity at the time of fallout to the exposure rate
at that time. Crude exposure estimates can also be made
from retrospective estimates of 137Cs or 90Sr inventories
measured in soil samples, provided one can estimate the
relative contributions from each of the tests to the total
measured inventory. The reliability of dose estimates is
dependent, however, on having reliable estimates of the
time of transport of the fallout from the detonation point
to the receptor point. Those data, called the fallout
“time-of-arrival” (TOA, measured in h), can considerably affect the dose estimates for locations relatively
close to the detonation point (i.e., within a few hours
transit time for the fallout). In a separate paper, Beck et
al. (2010) describe available post-test data on measured
exposure rates and provide estimates of both TOA and
ground deposition densities (Bq m⫺2) of 137Cs based on
those and other types of data. 137Cs deposition density
estimates were developed for each of 32 atolls and
separate reef islands of the Marshall Islands from each of
the 20 tests that took place at Bikini or Enewetak that
resulted in measurable fallout on the atolls.‡ Estimates of
fallout TOA were also developed for those tests and
atolls so that estimates of dose from external irradiation
could be reconstructed using either method. The names,
dates, and yields of the 20 tests that deposited fallout on
any of the inhabited atolls or separate reef islands, other
than the test site atolls themselves, are provided in Simon
et al. (2010a, Table 1).
In a companion paper (Simon et al. 2010b), the
doses from internal irradiation also are estimated for all
the tests and atoll populations that are considered in this
paper. The risks of cancer resulting from the doses
arising from exposure to radioactive fallout from regional nuclear testing in the Marshall Islands, taking into
consideration age and atoll of residence at the time of the
tests, are assessed in another companion paper (Land
et al. 2010).
MATERIALS AND METHODS
The doses from external irradiation were estimated
in three basic steps:
‡
The reader will note that this work does not attempt to quantify
the deposition on the test site atolls (Bikini and Enewetak). Not only
was the contamination on the islands of those atolls very heterogeneous, but they were monitored extensively for many years and those
data are reported elsewhere. Moreover, those atolls were never
inhabited during the testing years.
August 2010, Volume 99, Number 2
1. estimation of the outdoor exposure rates at 12 h at
each atoll after each test and of the temporal variation
of the exposure rate after each test;
2. estimation of the total external exposures from fallout
from TOA to infinity, obtained by integrating the
estimated exposure rates over time assuming continuous residence on the atoll (with corrections for
relocated populations); and
3. estimation of whole-body and organ doses by applying conversion factors from outdoor exposure to
tissue dose.
Estimation of the outdoor exposure rates
The outdoor exposures at each atoll following each
test have been assessed in one of two ways depending on
whether measured exposure rates were available for the
times and locations of interest. If historical data on
exposure rate were available, the data were assessed and
a best estimate of the island- or atoll-average exposure
rate at 12 h post detonation (termed Ė12) was made.
Because the quality of the exposure-rate measurements
varied by test and location, expert judgments were often
used to determine the appropriate weighting of measurements of varying quality. As discussed in Beck et al.
(2010), many of the reported measurements were made
before all the fallout from a test was deposited, while
other measurements were obtained many weeks after the
test when the exposure rate had been attenuated due to
weathering of the fallout by rainfall or human activities.
Of course, neither would have been as preferable as
high-quality ground-level exposure-rate measurements
made soon after deposition was complete.
If no reliable exposure-rate data were available to
estimate Ė12 directly, then the dose estimation method
used was that developed by the Off-Site Radiation
Exposure Review Project (ORERP) for estimating external whole-body and organ doses from fallout originating
at the Nevada Test Site (NTS) (Hicks 1982). That
method relates the 137Cs deposition densities and fallout
TOA values to Ė12 using ratios of 137Cs to Ė12 for a
range of times developed specifically for some of the
tests considered in this paper (Hicks 1984). The types of
data provided by Hicks (1981, 1982, 1984) are: (i)
calculated exposure rates from all radionuclides in the
fallout debris relative to a reference exposure rate of 1
mR h⫺1 at H⫹12 (12 h post detonation), at 31 times after
detonation, ranging from 1 h to 50 y, and (ii) related
radionuclide ground deposition densities, expressed in
Ci m⫺2, for more than 60 of the most important fission
and activation products (the number varies from one test
to another). Activities of fission products per unit of
Doses from external irradiation ● A. BOUVILLE
exposure rate were calculated from classified and declassified data available to Hicks on the amount of fissionable nuclides in the device and the measured fission
neutron spectra. The “zero time” activation product
values were the results of measurements made by aircraft
surveillance within 1 to 4 h post detonation. Hicks made
assumptions regarding fractionation effects from which
he developed his tables for unfractionated debris (designated as R/V ⫽ 1, where R stands for refractory radionuclides and V for volatile radionuclides), as well as for
debris with 50 and 90% of the refractory elements
removed (designated as R/V ⫽ 0.5 and R/V ⫽ 0.1,
respectively). As described in Beck et al. (2010), we
modified Hicks’ calculated activity ratios for unfractionated fallout (R/V ⫽ 1) to estimate the activity ratios for
various degrees of fractionation. For all tests except the
Bravo test, available data support our assumption of an
R/V ratio of 0.5 at all atolls. In contrast, however, there
were some significant variations in the degree of fractionation for Bravo fallout at some atolls: 0.7 for Likiep,
0.9 for Mejit Island, 1.3 for Ailinginae, 1.4 for Rongelap,
1.5 for Rongerik, and 0.5 at all other inhabited locations.
The high fractionation conditions (R/V ⬎ 1) for test
Bravo at atolls close to the Bikini Atoll test site reflect
the preferential deposition of large particles at early times
of arrival, in which the activity of refractory radionuclides is greater than that of volatile radionuclides.
Hicks calculated nuclide composition as a function
of time for six thermonuclear tests in the 1954 Castle
series (Mike, Bravo, Romeo, Yankee, Zuni, and Tewa);
the data from the other 14 thermonuclear tests that
deposited fallout in the Marshall Islands are still classified. As described below, Hicks’ data were used in two
different ways in our calculations according to the
information that was available for each test and location:
1. If the exposure rate was measured or inferred at any
time after the test, then only information on the
temporal variation of the exposure rate was required
to correct exposure-rate measurements made at different times to H⫹12, and, as described later in this
paper, to integrate the estimated exposure rates to
obtain total exposure. This is the method that was
generally used for the atolls and tests where exposure
rates were measured by airplane surveys or ground
surveys conducted soon after the test. In our method,
corrections were also made for the gradual decrease of
radionuclide activities in the upper layers of soil
resulting from environmental loss processes (termed
“weathering effects” in this paper), which are not
taken into account in Hicks’ calculations. Those
corrections, described later in this paper, are trivial for
ET AL.
145
the first week or month after the test, but are substantial when calculations of exposure rate are made for
years or decades after the test.
2. If the exposure rate had not been measured, but rather
the 137Cs deposition density was estimated for a given
test i and at an atoll j, then Ė12(i, j) was estimated
from Hicks’ predicted ratios of 137Cs to Ė12, modified
to account for our best estimate of fractionation. Eqn
(1) presents the form of this calculation:
Ė12(i, j) ⫽
A(i, j)
,
ND(i, j)
(1)
where A(i, j), in Bq m⫺2, is the 137Cs activity deposited
per unit area of ground at atoll j after test i (Beck et al.
2010), and ND(i, j) is the normalized 137Cs deposition
density, expressed in Bq m⫺2 per mR h⫺1 at H⫹12,
and inferred from the work of Hicks (1981, 1984) for
the selected value of R/V for test i at atoll j.
The method described above would be appropriate if
the 137Cs deposition density was measured within about
one month after the test and if it could be unequivocally
assumed to have been a result of fallout from that test.
However, as a rule, 137Cs was measured in soil many
years later in the 1970’s and the 1990’s. In that case, we
first decay-corrected the measurements of 137Cs deposition back to the time of the testing in order to obtain a
preliminary estimate of Ė12 for further refinement.
In practice, as described by Beck et al. (2010), both
methods were used to estimate both Ė12 and 137Cs
deposition, often in an iterative manner in order to
obtain: (1) credible fallout patterns over the territory of
the Marshall Islands; (2) reasonable sets of Ė12 and
fallout TOA values; and (3) in some cases, estimates of
fractionation.
As shown by Beck et al. (2010), the ratio of 137Cs to
Ė12 decreases as the degree of fractionation increases,
from 31.8 Bq m⫺2 per mR h⫺1 at H⫹12 for R/V ⫽ 0.5 to
7.8 Bq m⫺2 per mR h⫺1 at H⫹12 for R/V ⫽ 1.5. As
previously indicated in this paper, the fallout from Bravo
at some of the more northern atolls was enriched in
refractory nuclides (i.e., R/V ⬎ 1) resulting in a reduced
ratio of 137Cs to Ė12 relative to fallout deposited at
further distances from the test site where typical values of
R/V were 0.5. Although the dependence of the 137Cs to
Ė12 ratio on fractionation is substantial, it had only a
minor impact on the exposure-rate estimates made in this
study since actual exposure-rate measurements were
available for most of the atolls impacted by fractionated
Bravo fallout. Thus, in practice, only the Hicks’ data for
R/V ⫽ 0.5, typical of fallout at distant sites from a
detonation point (Hicks 1982), were used to estimate Ė12
146
Health Physics
from 137Cs deposition density estimates. However, as
described in Beck et al. (2010), the assumed degree of
fractionation was very important for estimating 137Cs
deposition density from exposure-rate measurements at
some atolls heavily impacted by Bravo.
August 2010, Volume 99, Number 2
Table 1. Fitted parameter values of an and n for use in eqn (2) to
describe the variation of the exposure rate with time after detonation according to Hicks’ (1981, 1984) data for fractionated debris
(R/V ⫽ 0.5) for Bravo (thermonuclear tests) and Tesla (nonthermonuclear tests). The values of an are normalized to an
exposure rate of 1 mR h⫺1 at H⫹12.
Thermonuclear tests
Estimation of the total exposure from fallout
In order to estimate the total exposure from fallout
from an estimate of exposure rate at any specific time, we
used the temporal variations of exposure rate given by
Hicks (1981, 1984) in a manner described below.
First, we developed analytic expressions of the
temporal variation of the normalized exposure rate for
both Bravo and for a non-thermonuclear test (Tesla) that
was conducted at the NTS for the purposes of deriving
the exposure over any interval of time (post-detonation)
from the data provided by Hicks (1981, 1984). The Hicks
exposure-rate data, which are relative to an exposure rate
of 1 mR h⫺1 at H⫹12, but do not take weathering effects
into account, were fit to 10-component exponential
functions such that a mathematical integration could be
easily accomplished. The form for the fitted functions of
the exposure rate was:
Non-thermonuclear tests
Component of
exponential
an
(mR h⫺1)
n
(h⫺1)
an
(mR h⫺1)
n
(h⫺1)
1
2
3
4
5
6
7
8
9
10
9.30 ⫻ 101
3.35 ⫻ 101
1.65 ⫻ 100
5.00 ⫻ 100
1.85 ⫻ 100
3.50 ⫻ 10⫺1
9.58 ⫻ 10⫺2
1.38 ⫻ 10⫺2
1.40 ⫻ 10⫺3
7.37 ⫻ 10⫺6
2.25 ⫻ 100
8.30 ⫻ 10⫺1
8.30 ⫻ 10⫺1
3.88 ⫻ 10⫺1
9.67 ⫻ 10⫺2
2.28 ⫻ 10⫺2
5.83 ⫻ 10⫺3
1.43 ⫻ 10⫺3
3.05 ⫻ 10⫺4
2.66 ⫻ 10⫺6
1.02 ⫻ 102
3.26 ⫻ 101
1.00 ⫻ 10⫺8
1.68 ⫻ 100
9.57 ⫻ 10⫺1
3.04 ⫻ 10⫺1
8.08 ⫻ 10⫺2
8.75 ⫻ 10⫺3
9.28 ⫻ 10⫺6
2.38 ⫻ 10⫺6
1.86 ⫻ 100
6.44 ⫻ 10⫺1
6.44 ⫻ 10⫺1
1.34 ⫻ 10⫺1
8.99 ⫻ 10⫺2
2.03 ⫻ 10⫺2
4.35 ⫻ 10⫺3
7.58 ⫻ 10⫺4
4.05 ⫻ 10⫺6
1.00 ⫻ 10⫺8
10
Ė(t)/Ė12 ⫽
冘a e
n
⫺nt
,
(2)
n⫽1
where
t ⫽ the time elapsed since the time of the
detonation of the device (h);
Ė(t)/Ė12 ⫽ the ratio of the exposure rate at time t to the
exposure rate 12 h after detonation, expressed in mR h⫺1;
an ⫽ the coefficient to the nth exponential term;
and
n ⫽ the decay constant for the nth exponential
term (h⫺1).
The fitted regression values for an and n for Bravo
and Tesla are given in Table 1 for R/V ⫽ 0.5. As shown
in Fig. 1, exposure-rate data for six thermonuclear tests
(Hicks 1984) are highly similar. For that reason, we
concluded that the single set of regression parameters,
shown in Table 1, would be suitable for all 16 thermonuclear tests listed in Simon et al. (2010a, Table 1). The
regression parameters shown in Table 1 correspond to a
degree of fractionation (R/V) of 0.5, typical of fallout at
relatively large distances from the site of detonation
where most of the deposited activity was associated with
relatively small particles (⬍50 m diameter). We also fit
Hicks’ data for Bravo for R/V ⫽ 1.0 and used those
values for the higher fractionation ratios. As shown later,
the difference in decay rates between the R/V ⫽ 1.0 and
Fig. 1. Variation with time of the normalized exposure rates for six
thermonuclear tests and a non-thermonuclear test (Tesla) for a
fractionation level, R/V, of 0.5 (Hicks 1981, 1984).
R/V ⫽ 0.5 curves is small. We used the R/V ⫽ 1.0 decay
rate regression fit to calculate total exposure and Ė12
values for close-in distances and short TOAs of fallout
where we assumed R/V to be greater than 0.5. In the
absence of similar data for any non-thermonuclear tests
at Bikini or Enewetak, we concluded that the data
derived by Hicks (1981) for the Tesla test conducted at
the NTS would adequately reflect the decay rate and
nuclide composition of the four non-thermonuclear tests
(Simon et al. 2010a, Table 1) that deposited relatively
low levels of fallout in the Marshall Islands. As shown in
Fig. 1, the decay rate for Tesla is very similar to that for
the six thermonuclear tests.
We subsequently took into account the influence of
weathering on the temporal variation of the exposure
Doses from external irradiation ● A. BOUVILLE
rate. In Fig. 1 and Table 1, the gradual decrease of the
exposure rate caused by the migration of the deposited
activity into deeper layers of soil is not taken into
consideration. Most radionuclides penetrate into the soil
quite rapidly during the first year after deposition, but the
vertical distribution of activity tends to stabilize after the
first year. To properly account for the influence of
weathering with time, we developed a time-dependent
weathering correction factor, W(t⫺TOA) (Fig. 2). We
believe that this weathering correction, which is based on
the analysis of actual depth profiles of 137Cs and 90Sr
measured in the soil in the Marshall Islands in 1978 and
1991–1993, reasonably reflects the actual time variation
in exposure rates from Bikini/Enewetak fallout in the
years from 1948 –1970. Mean values of observed relaxation lengths in the Marshall Islands in 1978 were about
5–7 cm and were only slightly greater in 1993–1994
(with of course wide variations). However, as discussed
in Beck et al. (2010), 137Cs was known to have been lost
from the soil profile with an effective half life of about
12–20 y compared with a physical half life of 30 y. Thus,
after about 5 y, when 137Cs begins to account for most of
the external exposure rate, the weathering loss of 137Cs
accounts for most of the reduction in exposure rate.
The weathering correction factor, W(t ⫺ TOA), was
analytically implemented in our calculations of exposure
in one of two different ways, depending on the time after
deposition: in the year of deposition, a weathering rate,
w, of 0.00018 h⫺1, corresponding to a half-time of 5 mo
and reflecting the initial weathering correction shown in
Fig. 2, was added to each of the n values; while for
subsequent years, the exposure rates obtained using the
values presented in Table 1 were multiplied by the
ET AL.
147
time-dependent factors shown in Fig. 2, i.e., equal to 0.5
and 0.35 in years 1 and 2 following the test, respectively,
and decreasing gradually to a value of 0.1 in the
twentieth year after the test.
Our corrections for weathering impact the estimated
exposure rates only after a few weeks and, because of the
rapid decrease in fallout exposure rates with time shown
in Fig. 1, have only a minor effect on an individual’s
integrated exposure as shown in Table 2. The effect on
total exposure is greatest for large TOAs, corresponding
in general to relatively low fallout. However, as discussed above, weathering does have a significant effect
on the small annual doses from residual long-lived
activity.
The variation with time of the exposure rate, relative
to an exposure rate of 1 mR h⫺1 at H⫹12 h, is shown in
Fig. 1 and illustrates that there is little difference in the
rate of change of exposure from fallout from different
tests. Fig. 3 and Table 2 illustrate that the degree of
fractionation also has only a minor effect on the temporal
variations of exposure rate and mainly at very long times
after the detonation. At long times, the more volatile
nuclides, such as 106Ru and 137Cs, contribute a greater
fraction of the exposure compared to refractory nuclides.
As shown in Fig. 3, weathering also has only a minor
effect on exposure rate at early times, when the exposure
rate is high and, thus, as shown in Table 2, has only a
relatively small effect on the integrated exposure. However, as shown in Fig. 3, weathering does significantly
reduce the exposure rate at long times after deposition.
Given the expressions for the decay rate as a
function of time, modified for weathering, the exposure,
E, between any two times of interest, t1 and t2, is
determined by integrating the normalized exposure rate
(either measured or calculated) using eqn (3):
E(t 1, t 2, i, j) ⫽
冕
10
t2
[Ė12(i, j)
冘a e
n
⫺ nt
W(t ⫺ TOA)]dt
n⫽1
t1
10
Ė12(i, j) ⫻
冘 a [e
n
⫺(n ⫹ w)ti
n⫽1
⫽
( n⫹ w)
⫺ e ⫺(n ⫹ w)t2]
(3)
for the first year of exposure.
Fig. 2. Time-dependent correction factor used to take the weathering effect into account.
Estimation of the conversion factors from outdoor
exposure to tissue dose
In order to estimate whole-body or organ dose
from the integrated exposure, the following factors
were considered.
First, the exposure rates estimated above correspond
to outdoor conditions in the populated areas but do not
148
Health Physics
August 2010, Volume 99, Number 2
Table 2. Variation of the exposure (mR) with increasing TOA (h) and influence of the weathering effect. The exposure
rate is normalized to 1 mR h⫺1 at H⫹12 and the relative degree of fractionation (R/V) is assumed to be 0.5.
Exposure (mR)
Exposure (mR)
from TOA to:
Weathering?
(Y/N)
TOA ⫽
0h
TOA ⫽
4h
TOA ⫽
6h
TOA ⫽
12 h
TOA ⫽
22 h
TOA ⫽
40 h
TOA ⫽
68 h
TOA ⫽
162 h
1 wk
1 wk
1 mo
1 mo
1y
1y
10 y
10 y
70 y
70 y
Y
N
Y
N
Y
N
Y
N
Y
N
143.0
143.3
154.0
154.5
158.0
162.1
159.1
162.1
159.1
165.2
42.8
43.0
53.3
54.2
58.2
61.4
58.5
62.1
58.5
62.6
37.0
37.3
47.4
48.3
52.3
55.5
52.6
56.4
52.7
56.8
28.5
28.8
39.0
39.8
43.9
47.0
44.2
47.9
44.2
48.3
21.3
21.5
31.8
32.6
36.7
39.7
37.0
40.6
37.1
41.1
14.7
14.8
25.2
25.9
30.1
33.0
30.5
33.8
30.5
34.4
9.1
9.1
19.7
20.3
24.6
27.4
24.8
28.2
25.0
28.8
0.34
0.34
11.1
11.5
16.1
18.6
16.5
19.4
16.5
20.0
Fig. 3. Exposure rate as a function of time after detonation with
and without a weathering correction at two fractionation levels,
R/V ⫽ 0.5 and R/V ⫽ 1.0.
necessarily reflect the variation from one area of an
island to another or to indoor conditions. The potential
reduction of exposure due to shielding by building
materials when inside traditional Marshallese houses
would be small as suggested by measurements made
after Bravo that indicated that native housing did not
appear to substantially attenuate the fallout radiation
(Sharp and Chapman 1957; Conard et al. 1975). Contemporary measurements of outdoor exposure rates (Fig.
4), however, show substantial variation from one area of
any island to other areas. At downwind distances where
most atolls were located, fallout debris clouds were, for
the most part, larger than individual islands. For that
reason, we believe that fallout deposition was usually
relatively homogeneous over any given island. Therefore, during the first year after fallout, when over 97% of
the lifetime exposure occurred (Table 2), there was little
difference in the exposure rates from one area of any
island to another. Over time, however, exposure rates in
areas near the shore became lower compared to exposure
rates in the center of islands as a consequence of
weathering, human activity, and intermittent flooding
from storms. The exposure rates were also much lower in
subsequent years than during the first year after fallout.
In this work, we have assumed that our estimated outdoor
exposure rates, based on the original fallout levels, were
representative of the average conditions under which
people lived during the periods of maximum exposure,
but we recognize that this assumption may have resulted
in a very slight overestimation of the cumulative exposure as Marshallese spend much of their time in village
areas that are typically near the lagoon shore.
In order to calculate the organ and tissue doses from
the free-in-air exposure data, one must first convert
exposure to dose in air using a factor of 8.75 ⫻ 10⫺3 Gy
R⫺1. Then, a factor of 0.75 Gy Gy⫺1 was used to convert
from dose in air to dose in tissue or organ. This factor of
course varies with the energy of the radiation and the
orientation with respect to radiation incidence (NCRP
1999; Eckerman and Ryman 1993; ICRP 1996), as well
as with the organ and tissue that is considered and with
the anthropometric characteristics of the person. Because
there is little difference between the values of this
conversion factor for one organ to another for gamma-ray
energies of a few hundreds of keV that are typical for
fission products (Jacob et al. 1990; ICRP 1996), the same
value was used for all organs and tissues that were
considered in this study and also would be used if the
effective doses were to be calculated. The conversion
factor from dose in air to effective dose was taken as 0.75
Sv Gy⫺1 by the United Nations Scientific Committee on
the Effects of Atomic Radiation (UNSCEAR 1993) and
by the National Council on Radiation Protection and
Measurements (NCRP 1999) for adults exposed to fallout. The net conversion from exposure in air to tissue or
Doses from external irradiation ● A. BOUVILLE
ET AL.
149
Fig. 4. Relative exposure rates (arbitrary units) across Eniwetak Island, Rongerik Atoll, in 1978. Data derived from the
U.S. Department of Energy-sponsored aerial radiological survey of the Marshall Islands (Tipton and Miebaum 1981).
organ dose is thus about 8.75 ⫻ 10⫺3 (Gy R⫺1) ⫻ 0.75
(Gy Gy⫺1) ⫽ 6.6 ⫻ 10⫺3 (Gy R⫺1) for adults.
While the dose conversion factor for an actual
person depends on the age and sex of the person, or, more
precisely, her or his anthropometric characteristics, doses
in this study were estimated for representative persons,
defined as hypothetical individuals with anthropometric
characteristics that are typical of those of the people who
lived in the Marshall Islands in the 1950’s. Calculations
using anthropomorphic phantoms of different ages
(Jacob et al. 1990) indicate that body size, which is
generally correlated with age, results in slightly higher
doses for younger ages. Based on those calculations, we
adjusted our estimated doses for representative adults to
doses for younger (⬍3 y, including in utero) and older (3
through 14 y) children by multiplying the adult doses by
1.3 and 1.2, respectively.
RESULTS AND DISCUSSION
Doses from external irradiation were estimated for
the entire population of the Marshall Islands and for each
of the 20 tests that took place at Bikini or Enewetak that
resulted in measurable fallout on inhabited atolls of the
Marshall Islands (see Table 1 in Simon et al. 2010a). The
population of the Marshall Islands was classified into 26
population groups consisting of the permanent residents
of 23 atolls and islands, and of three population groups
that were evacuated or relocated (see Table 2 in Simon et
al. 2010a). With the exception of the populations of the
atolls that were evacuated following the test Bravo of
1954 or were relocated before the testing began (see
Simon et al. 2010a, Table 3), we have assumed that our
estimated doses pertained to representative persons from
each atoll, and that there was no movement of those
people from one atoll to another.
Estimated exposures
As shown in Fig. 1, the ground-level exposure rate
decreases very rapidly with time after detonation, by a
factor of more than 1,000 during the first 1,000 h (about
40 d). For that reason, the lifetime exposure varies
substantially with the fallout TOA, as was shown in
Table 2. The example TOA values that were chosen for
Table 2, with the exception of the extreme value of
TOA ⫽ 0, correspond to the range of values estimated
for the inhabited atolls after the various tests (Beck et al.
2010). As previously indicated, most of the external
exposure occurs within the first year following the
detonation. The influence of the weathering effect is
barely noticeable during the first month after the detonation and only plays a substantial role after one year. It
is, however, extremely important to take the weathering
150
Health Physics
effect into account when the only measurement available
is of 137Cs activity in soil sampled decades after the test.
The estimated outdoor exposures, from TOA to
infinity, are presented in Table 3 for each test and each
atoll or island, whether it was inhabited or not. For most
of the tests, exposures of less than 1 R were estimated at
all atolls and islands. Much higher exposures, ranging
from 5 to 500 R, were assessed for several atolls in the
northern part of Marshall Islands and for several tests of
the 1954 Castle series (Bravo, Romeo, Yankee, Koon,
and Union). When exposures are summed over all atolls
and islands, those five tests account for 99% of the total
exposure, with Bravo alone contributing 84%.
Estimated tissue and organ doses
Annual doses from external irradiation have been
estimated for representative persons of the 26 population
groups classified into three different age categories
(infants, children, and adults). The annual doses are
reported for the time period from 1948 to 1970. By 1970,
the doses had decreased to very low levels in comparison
to the peak observed in 1954. Since the doses are
estimated for representative persons who were assumed
to have remained on each atoll with movements between
atolls limited to the relocated and evacuated populations
(Simon et al. 2010a, Table 3), the doses from external
irradiation are proportional to the exposures calculated
using eqn (3), which are based on the environmental
radiation data (measurements or estimated values) available for each atoll and test. The doses reported for the
relocated populations include, where appropriate, contributions from exposures received before evacuation, during the period of resettlement, and following return to the
atoll of origin.
Estimated annual doses for adults, shown in Fig. 5,
were highest in 1954 and then decreased to values that
were, in 1970, less than one thousandth of the peak
values observed in 1954. The annual doses shown in Fig.
5 are for representative adults of four population groups
(Majuro residents, Kwajalein residents, Utrik community
members, and Rongelap Island community members)§
that represent a range of deposition densities, as well as
a range of exposures in four distinct areas.
In Table 4, the doses through 1970 resulting from
the Bravo test are compared, for each of the 26 population groups, to the corresponding doses from all the tests
of the Castle series conducted in 1954 and from all tests
§
Note to reader: As indicated in Simon et al. (2010a), we make
the distinction in this paper between “residents” of either Majuro and
Kwajalein and “community members” of Rongelap or Utrik. In the
former case, we are referring to anyone living on those atolls at the
time of fallout. In the latter case, we are referring to the entire group
of persons exposed on either Rongelap or Utrik and who were
members of the group relocated from those atolls.
August 2010, Volume 99, Number 2
listed in Simon et al. (2010a, Table 1). At every atoll in
the Marshall Islands, the Castle series was the predominant contributor to the total external dose. While Bravo
was responsible for most of the external dose for the
northern atolls, it was not the case for the mid-latitude
and southern atolls. For example, the proportions of the
external dose contributed by Bravo for the Rongelap
Island community, the Utrik community, Kwajalein residents, and Majuro residents were ⬎99%, 84%, 4.6%,
and 23%, respectively. In contrast, among the midlatitude atolls (Kwajalein and others), Yankee was the
most important test. The contributions from Yankee to
the external dose for the Rongelap Island community, the
Utrik community, Kwajalein residents, and Majuro residents were ⬍⬍1%, 4.5%, 39%, and 1.9%, respectively.
Among the southern atolls, the Romeo and Koon tests
were the most important contributors to external dose.
The contributions to the external dose from the combination of Romeo and Koon fallout for the Rongelap
Island community, the Utrik community, Kwajalein residents, and Majuro residents were 0.5%, 6.5%, 25%, and
61%, respectively.
The external doses we estimated for the adult populations of the Rongelap Island and Utrik communities from
Bravo are very similar to those estimated previously by
Lessard et al. (1985), but our estimated dose for the 18
persons from Rongelap Island who were exposed to Bravo
fallout on Ailinginae is about one half the dose estimated by
Lessard et al. The reason for the differing estimates for
exposures on Ailinginae appears to be due to different
estimates of TOA, 3 h for Lessard et al. (1985) compared to
4 h assumed in this study. As shown in Table 2, the integral
dose over the first few days is very sensitive to TOA,
particularly within the first day. The exact TOA for Bravo
fallout at Ailinginae was not measured directly but was
inferred from measurements at other atolls and, thus, is
uncertain.
Estimates of external doses to representative adults
from all tests are summarized in Table 5 according to
region of residence. For reference, the populations of
each atoll are given in Simon et al. 2010a (Table 2). As
shown, the estimated total external doses from 1948
through 1970 to the adult populations of the southern
atolls were all on the order of 5–22 mGy, and in the
mid-latitude region, 22–59 mGy. The doses to the populations of Rongelap Island community, Ailinginae, and
Utrik community were much higher, reflecting the heavy
fallout from Bravo, even though the populations were
relocated within a few days after the test (Simon et al.
2010a, Table 3). The dose shown for Rongerik in Table
4 (940 mGy) is the estimated dose from Bravo fallout
Table 3. Free-in-air exposure (mR), from TOA to infinity, by location and test.
Dog
Ailinginae
Ailinglaplap
Ailuk
Arno
Aur
Bikar
Ebon
Erikub
Jabat
Jaluit
Jemo Island
Kili Island
Knox
Kwajalein
Lae
Lib Island
Likiep
Majuro
Maloelap
Mejit Island
Mili
Namorik
Namu
Rongelap
Island
Rongerik
Taka
Taongi
Ujae
Ujelang
Utirik
Wotho
Wotje
290
1.6
1.4
2.9
2.4
2.6
1.5
2.1
0.0
2.2
0.80
0.77
2.0
710
0.75
0.76
1.4
2.4
2.5
1.9
2.0
2.6
1.4
280
120
0.0
33
0.0
0.10
14
0.0
8.6
0.0
0.0
35
0.0
0.0
5.6
1.8
0.0
0.0
0.0
0.34
23
0.0
0.0
0.02
96
160
0.95
14
0.45
13
1.7
340
2.1
79
53
1.2
4.8
200
53
41
3.4
Item
Mike
King
Bravo
Romeo
Koon
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
61
61
61
61
61
61
61
61
61
61
61
25
62
61
61
61
61
61
61
61
61
61
61
61
0.0
36
36
36
36
0.0
0.0
34
36
31
71
0.0
32
38
75
36
71
31
36
34
31
0.0
36
0.0
120,000
56
5,600
350
500
93,000
110
750
170
160
4,500
160
290
150
240
100
3,700
340
770
4,200
280
130
110
480,000
11,000
270
720
430
440
5,300
460
450
240
400
740
390
350
340
290
330
900
470
450
700
330
310
290
33,000
0.0
0.0
0.0
0.0
710
0.0
0.0
0.0
58
61
61
61
66
57
61
61
0.0
0.0
0.0
36
44
0.0
0.0
34
400,000
25,000
390
150
310
35,000
640
2,600
24,000
2,600
130
270
610
2,400
470
730
Union
Yankee
Nectar
Zuni
Flathead
6,700
180
250
23
510
340
540
20
77
24
5,000 1,700
64
19
370
53
370
23
110
16
290
62
120
16
320
38
480
83
36
28
410
53
180
69
440
39
320
22
440
390
310
36
150
4.2
550
27
29,000 12,000
2,000
31
1,600
87
120
8,500
67
120
31
41
710
20
20
1,200
220
340
710
29
92
1,500
20
19
92
6,800
24
180
0.0
0.0
0.0
0.0
19
0.0
180
7.6
0.0
57
0.0
20
380
380
0.0
0.0
0.0
0.0
0.0
57
310
7.1
40
0.0
75
0.0
0.0
0.0
0.0
0.0
0.0
0.0
75
0.0
0.0
28
80
0.0
110
0.0
0.0
0.0
0.0
0.0
0.0
33
3.9
140
0.0
52
210
1.9
0.0
26
200
160
0.0
100
0.0
170
150
150
0.0
74
100
7.5
0.0
100
140
3.4
19,000
2,100
31
29
140
2,400
290
430
5,800
1,200
36
150
370
1,200
320
710
0.0
0.0
1.8
350
310
0.0
170
0.0
900
0.0
0.0
53
57
45
800
0.0
170
0.0
0.0
180
38
8.6
150
7.5
4,600
370
29
29
360
940
150
28
Tewa
Cactus
Fir
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
35
10
76
3.9
23
43
23
37
12
16
73
8.2
0.94
37
0.83
2.9
36
9.0
30
91
0.90
8.1
7.5
190
0.0
0.0
0.0
0.0
360
0.0
0.0
0.0
0.0
0.0
0.0
0.0
20
0.0
0.0
0.0
190
69
490
0.0
28
170
77
50
Koa
Maple
Redwood
Cedar
Total
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
75
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
74
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
18
140,000
1,100
9,000
1,600
1,500
110,000
820
1,900
1,300
1,000
6,600
900
1,100
3,300
1,600
1,900
5,900
1,500
1,900
7,500
1,100
840
1,600
570,000
0.0
0.0
160
0.0
6.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
68
0.0
0.0
0.0
2.5
0.0
2.5
0.00
52
0.0
0.0
0.0
0.0
0.0
7.4
0.0
10
0.0
45
0.0
450,000
32,000
1,400
1,300
3,800
42,000
3,600
4,600
ET AL.
Yoke
Doses from external irradiation ● A. BOUVILLE
Atoll
151
152
Health Physics
Fig. 5. Estimated annual doses, in mGy, to adults of four
population groups.
Table 4. Estimates of external doses (mGy) received by adults
from the Bravo test, the entire Castle (1954) test series, and from
all tests (dose estimates rounded to two significant digits).
a
Atoll or population
group
Bravo
Castle
series
All tests
Ailinginaea
Ailinglaplap
Ailuk
Arno
Aur
Bikini communityb
Ebon
Enewetak communitya
Jaluit
Kwajalein
Lae
Lib Island
Likiep
Majuro
Maloelap
Mejit Island
Mili
Namorik
Namu
Rongelap control groupc
Rongelap Island communitya
Rongerikd
Ujae
Utrik communitya
Wotho
Wotje
460
0.37
37
2.3
3.3
1.1
0.71
2.1
1.1
1.0
1.6
0.7
25
2.2
5.1
27
1.8
0.70
0.73
8.4
1,600
940
1.0
110
4.3
17
470
5.3
57
9.3
7.7
5.0
4.8
14
4.8
15
7.8
11
37
8.7
11
47
6.4
4.4
9.0
17
1,600
—
6.4
130
13
30
470
6.9
59
10
9.9
14
5.3
25
6.6
22
10
12
39
9.8
12
49
7.0
5.5
11
22
1,600
—
8.6
130
23
31
Includes doses received while relocated (see Table 3 in Simon et al.
2010a).
b
Includes doses while on Kwajalein and Kili (see Table 3 in Simon et al.
2010a).
c
Includes doses while on Majuro and on Rongelap Island.
d
Dose to U.S. military personnel on Rongerik prior to evacuation (see
Table 3 in Simon et al. 2010a).
August 2010, Volume 99, Number 2
received by the U.S. military weather observers who
were stationed there and evacuated within 2 d of the
detonation. The Rongerik dose is based on only a few
survey meter measurements made after the evacuation by
a survey team but agrees very well with reported external
exposure measured by film badges worn by the personnel
(35–98 R) (Sharp and Chapman 1957), particularly
considering the considerable uncertainty in both sets of
measurements and the fact that some of the military
personnel were indoors at least part of the time.
Whole-body absorbed doses (mGy) from external
irradiation, cumulated over the time period from 1948
through 1970, for representative persons by birth year
(1930 to 1958), are presented in Table 6 for the Majuro
residents, the Kwajalein residents, the Utrik community,
and the Rongelap Island community. As noted, doses for
Utrik and Rongelap Island communities account for
relocations. For a given population, the cumulative doses
are greater for persons who were young at the beginning
of the testing period.
The radionuclides that contributed most to the dose
rate from external irradiation vary according to the time
elapsed since the detonation. These contributions can
readily be derived from the tables prepared by Hicks
(1984), as the relative exposure rates are provided for all
radionuclides for a range of times after detonation. As an
example, the changing proportions of the external dose
rate contributed by some of the most important contributing radionuclides to external exposure are shown in
Fig. 6 for the Bravo test and an assumed relative degree
of fractionation, R/V, of 0.5. In Fig. 6, 132Te is the most
important radionuclide within a few hours after the test,
but is replaced successively by 140Ba-140La, 95Zr, and
finally by 137Cs. Expressed in percentage of total exposure (averaged over a range of degrees of fractionation),
132
Te-132I accounts for about 25–30%, 140Ba-140La about
20%, 131I ⫹ 133I ⫹ 135I about 15–20%, and 95Zr-95Nb ⫹
97
Zr-97Nb about 10 –15%. The exact percentages at any
atoll and following any particular test also depend on
fractionation with greater relative contributions from
Zr-Nb isotopes for larger R/V values. Although 137Cs
and 106Ru contribute little to the total integral dose
from TOA to 1970, they contribute almost all the
annual dose after 5 y.
All together, the deposition densities of 63 of the
radionuclides listed in Simon et al. (2010a, Table 4) have
been estimated at each inhabited atoll or reef island
following each of the 20 tests. These radionuclides
combined contribute more than 95% of the external dose.
The proportions of the total exposure contributed from
the individual radionuclides shown in Fig. 6 are actually
Doses from external irradiation ● A. BOUVILLE
ET AL.
153
Table 5. Population-weighted average external dose to adults of four groups of atolls and/or communities. Grouping is
based on similar levels of deposition of total 137Cs (see Fig. 2 of Simon 2010a). Range in parentheses represents the
minimum and maximum total external dose within the group of atolls or communities. All values rounded to two
significant digits.
Atoll or population
group
Total external dose
through 1970 from
all tests (mGy)
Atolls
Southern latitude
Mid-latitude
Utrik community
Rongelap Island/Ailinginae/
Rongerik evacuees
All
Ailinglaplap, Arno, Aur, Ebon, Jaluit,
Kili Islanda, Lae, Lib Island, Majurob,
Maloelap, Mili, Namorik, Namu, Ujae
Ailuk, Kwajalein, Likiep, Mejit Island,
Ujelangc, Wotho, Wotje
Utrik and atoll of relocationd
Rongelap, Ailinginae, Rongerik, and
atolls of relocationd
All
8.8
34
Range of total external
doses among atolls
(mGy)
5.3−22
22−59
130
1,000
—
470−1,600
27
5.4−1,600
a
Primary residence location of Bikini community during test years.
Includes Majuro permanent residents and Rongelap control group.
Primary residence location of Enewetak community during testing years.
d
See Table 3 of Simon et al. (2010a) for atolls of relocation.
b
c
Table 6. Whole-body absorbed doses (mGy) from external irradiation cumulated from 1948 through 1970 for representative persons by birth year (1930 to 1958) (rounded to two significant
digits). Doses for Utrik and Rongelap Island communities account
for relocations.
Whole-body dose from external irradiation (mGy)
Birth
year
Majuro
residents
Kwajalein
residents
Utrik
community
Rongelap Island
community
⬍1931
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
9.8
9.8
9.8
9.8
9.8
9.8
9.8
9.8
9.8
10
10
12
12
12
12
12
12
12
12
12
12
12
13
12
4.2
0.78
0.47
0.14
0.09
22
22
22
22
22
23
23
23
23
23
23
26
26
26
26
26
26
27
23
20
21
21
22
21
8.8
2.1
1.3
0.41
0.23
130
130
130
130
130
130
130
130
130
130
150
150
150
150
150
150
150
150
150
150
150
150
160
160
45
3.1
2.7
2.3
1.4
1,600
1,600
1,600
1,600
1,600
1,600
1,600
1,600
1,600
1,600
1,600
1,900
1,900
1,900
1,900
1,900
1,900
1,900
1,900
1,900
1,900
1,900
2,100
2,100
470
13
13
12
9.2
slight overestimates since the derived proportions are
relative to only the 63 radionuclides considered.
External dose from 239⫹240Pu, the last radionuclide
listed in Table 4 of Simon et al. (2010a), has not been
Fig. 6. Relative contribution (%) of selected radionuclides to the
total exposure rate on the ground as a function of time (h) after the
detonation.
calculated as deposition estimates for that radionuclide
are not available for each test separately. More importantly, the corresponding external doses would have
been trivial.
As a basis for evaluating the magnitude of the
estimated external doses, the annual and total doses
reported for adults in Tables 4 to 6 and in Fig. 5 can be
compared with the external doses that Marshallese adults
typically received from natural background radiation or
with typical doses received by Americans who lived near
the NTS. The average annual external dose received by
Marshallese from natural sources is about 0.24 mGy,
154
Health Physics
primarily from cosmic radiation, since the concentrations
of 238U, 40K, and 232Th in the coral soils is very low
(Robison et al. 1997). This can be compared to the
highest annual dose received in Majuro from fallout of
8 mGy in 1954 and annual doses on the order of 3–5
Gy after testing in the Marshall Islands ended (Fig.
5). External doses from atmospheric tests conducted at
the NTS from 1951–1958 that were received by
Americans (in this example, outdoor workers who
lived in towns in Nevada and SW Utah) ranged from
about 0.03 to 40 mGy (Henderson and Smale 1990).
Because of shielding when indoors, the NTS doses
were smaller for persons who spent much of their time
indoors.
Uncertainty
Uncertainties in the total dose received by each
population group in each year from all tests in that
year were derived relying, primarily, on the uncertainty of available measurements of exposure rates and
of deposition densities of long-lived radionuclides. For
a given test i and a given atoll j, the external dose to
permanent residents of age a, Da in mGy, can be
expressed as:
D a(i, j) ⫽ Ė12(i, j) ⫻
冋
册 冉 冊 冉 冊
X(i, j)
D ad
Da
⫻
⫻
,
X
D
Ė12(i, j)
ad
(4)
where
Ė12(i, j) ⫽ the exposure rate at H⫹12 (mR h⫺1) following test i at atoll j;
X(i, j) ⫽ the lifetime exposure (mR) due to test i at
atoll j; and
Dad/X ⫽ the conversion factor from exposure to
dose for adults (mGy mR⫺1).
The uncertainties were assessed to be as follows:
Ė12: as discussed in Beck et al. (2010), an uncertainty estimate was assigned to each estimate of Ė12
as inferred from the available measurement data.
These uncertainties, expressed in terms of geometric
standard deviations (GSDs), range from 1.3 to 3.0,
depending on the availability, quality, and number
of measurements of exposure rates and long-lived
radionuclides at the atoll for the test under consideration; and
● X/Ė12: because the exposure, X, is delivered over a
number of years, at a rate that is relatively high
during the year of the test and much smaller during
the following years, the simplifying assumption was
made, for the purposes of the evaluation of the
●
August 2010, Volume 99, Number 2
uncertainties, that the exposure was delivered only
during the year of the test. During that year:
10
X/Ė12 ⫽
10
an ⫺(n⫻TOA)
an ⫺[n(EOY ⫺ H)]
⫺
,
e
e
n
n
n⫽1
冘冋
n⫽1
册 冘冋
册
(5)
where an and n, with n varying from 1 to 10, are the
parameters of the fit to Hicks’ calculated exposure rates
vs. time (Hicks 1981, 1984), TOA, in hours, is the
estimated time of arrival of fallout counted from the time of
the test, H, and (EOY⫺H) is the time elapsed between the
time of the test and the end of the year (EOY).
As previously indicated, the exposure, X, is very
sensitive to TOA (Table 2), while the uncertainty in the
values of an and n is assumed to be relatively minor
compared to the uncertainty due to TOA. Also, as shown
in Fig. 1, regression fit parameters vary little from one
test to another. For that reason, we assumed that TOA is
the parameter in eqn (5), which is uncertain to any
significant degree. In our simulations, the uncertainty
distribution for TOA for all atolls and all tests was taken
to be uniform between 0.8 and 1.2 times the nominal
values given in Table 6 of Beck et al. (2010):
Dad/X: its nominal value of 6.6 ⫻ 10⫺3 mGy mR⫺1 is
based on the calculations of Jacob et al. (1990) and on
the recommendations of ICRP (1996). The value of
Dad/X depends on the geometry of irradiation, on the
energy spectrum of the incident ␥-rays, and on the
tissue or organ that is considered. In our analysis,
the same nominal value is taken to apply to all organs
and tissues of the body. The uncertainty distribution of
Dad/X is taken to be uniform between 0.9 and 1.1 times
the nominal value and to mainly reflect differences
between the doses to various organs and tissues of the
body for exposures to ␥-rays of a few hundred keV
characteristic of fallout; and
● Da/Dad is the ratio of the external dose to children of
age a to adults. Its nominal value is 1.3 for young
children (less than 3 y of age) and 1.2 for older
children. Here, the uncertainty distribution, which is
assumed to be uniform between 0.9 and 1.1 times the
nominal value, reflects the relatively large range of
ages to which the nominal value applies.
●
The uncertainty estimates for individual tests were
derived via Monte Carlo simulation to obtain an estimate of
uncertainty for the total external dose received in each
calendar year from all tests in that year. Results are
presented in Table 7 in terms of GSD for representative
persons (both adults and children) of four communities
(Kwajalein residents, Majuro residents, the Rongelap Island
Doses from external irradiation ● A. BOUVILLE
ET AL.
155
Table 7. Derived uncertainties, expressed in terms of the geometric standard deviation (GSD), in the annual doses from
external irradiation for four representative communities of the Marshall Islands.
Annual external whole-body dose (mGy) and uncertainty
Majuro residents
Kwajalein
residents (south)
Utrik community
Rongelap Island
community
Year of
exposure
Year of
birth
Mean dose
GSD
Mean dose
GSD
Mean dose
GSD
Mean dose
GSD
1948
1929
1947
1929
1947
1929
1947
1929
1953
1929
1953
1929
1953
0.016
0.021
—
—
0.55
0.67
8.5
11
0.48
0.57
0.063
0.076
2.8
2.8
—
—
1.6
1.6
1.2
1.2
2.9
2.9
2.8
2.8
4.6
6.0
0.039
0.047
0.60
0.72
15
19
1.3
1.5
0.24
0.29
2.0
2.0
2.8
2.8
1.9
1.9
1.3
1.3
1.7
1.7
1.9
2.0
0.011
0.014
0.35
0.42
0.34
0.41
120
160
0.56
0.67
1.2
1.4
2.8
2.8
2.9
2.9
1.8
1.8
1.3
1.3
1.4
1.4
1.3
1.3
1.8
2.4
0.62
0.75
0.37
0.45
1,600
2,000
0.48
0.58
2.7
3.3
2.8
2.8
2.7
2.7
1.8
1.8
1.3
1.3
2.8
2.8
1.5
1.5
1951
1952
1954
1956
1958
community, and the Utrik community) that represent the
range of exposures across the Marshall Islands. The table
applies to all years when testing gave rise to measurable
fallout in the Marshall Islands. The derived GSDs, which
range from 1.2 to 2.9, vary among years and among atolls,
essentially depending on the uncertainties assigned to Ė12
for each test and location. There is, however, very little
dependence with age as the uncertainties for adults and for
children have the same numerical values within a few percent.
The external doses resulting from the tests detonated in
1954 were the largest of any year, regardless of the atoll or
island. In contrast, the uncertainties of doses from tests in
1954 are the smallest because they are based on relatively
good measurement data in comparison to other years when
the doses were low and primarily based on 137Cs deposition
estimates derived from interpolation of measurement data at
nearby atolls or, in some cases, meteorological modeling.
As a simplification for the purposes of estimating the risks
of radiation-induced cancers (see Land et al. 2010), the
uncertainties assigned to the annual doses from external
irradiation to members of each community were given the
same value for all years when testing took place. The GSDs
assigned were based on the derived GSD estimates in the
years in which the doses were most significant. Overall, the
GSDs were smallest in communities where the greatest
dose resulted from the 1954 tests and highest in communities with the lowest doses from the 1954 tests. The derived
GSDs were 1.2 for the Rongelap Island community, 1.5 for
the Utrik community, and 1.8 for the Kwajalein and Majuro
residents. Because the quality and availability of data were
roughly the same for atolls and islands within each of four
atoll groups (see Table 5), the GSDs were assumed to be the
same for all communities within each group.
These selected uncertainties apply to the annual
doses received during the years when testing with measurable fallout occurred. In later years, the uncertainty
would be larger as weathering and estimated loss of 137Cs
from the soil profile, which can vary from one area of the
island to another, could be substantial. However, the
annual doses in the years without tests are lower by a
factor of 100 or more than the doses received during the
years of the tests; their GSDs have not been individually
derived but assumed equal to those in the years assessed.
SUMMARY AND CONCLUSION
Annual doses from external irradiation resulting from
fallout from regional nuclear weapons testing have been
estimated, for the first time, for all tests that resulted in
measurable fallout and for all Marshallese alive at the time
of nuclear testing (1946 –1958), and at all 25 inhabited
atolls. The methodology used to estimate the doses is based
on test- and location-specific radiation survey data coupled with estimates of fallout TOAs at the inhabited atolls or
on deposition density estimates of 137Cs coupled with fallout
TOAs. Both types of data are discussed in a companion
paper (Beck et al. 2010). For every test, the major part of the
dose from external irradiation was received during the first
year following the detonation. The most important tests
with respect to external exposure were those of the Castle
series conducted in 1954. Bravo was most important to
the northern atolls, Yankee was more important to the
mid-latitude atolls (Kwajalein and others), and Romeo
and Koon were more important to the southern atolls
(Majuro and others).
The total external doses to the populations of all the
inhabited atolls from all tests at Bikini and Enewetak
varied over two orders of magnitude with the adult
residents of the southern atolls receiving relatively low
total external doses ranging from 5–22 mGy on average,
the adults at the mid-latitude atolls receiving external
doses of 22–59 mGy, while the residents of the northern
156
Health Physics
atolls most impacted by the Castle series and the Bravo
test received external doses in the hundreds to over 1,000
mGy, even though the populations of the three most
exposed communities (Rongelap Island, Ailinginae, and
Utrik) were evacuated shortly after the test.
Our estimates of doses from external irradiation have
been merged with corresponding estimates of doses from
internal irradiation, given in a companion paper (Simon et al.
2010b), to assess the cancer risks (Land et al. 2010) among the
Marshallese population as a consequence of exposure to
radioactive fallout from the nuclear weapons tests.
Acknowledgments—This work was supported by the Intra-Agency agreement between the National Institute of Allergy and Infectious Diseases and
the National Cancer Institute, NIAID agreement #Y2-Al-5077 and NCI
agreement #Y3-CO-5117. The authors are indebted to several individuals
whose analyses and research have made substantial contributions to this
work. They include William Robison for various data and publications on
measurements made in the Marshall Islands, Brian Moroz for meteorological analysis and graphic support, Robert Weinstock for computer programming support, Dunstana Melo for the review of the manuscript, as
well as other scientists who have added to our understanding of the
contamination and consequences in the Marshall Islands through their
scientific publications, many of which are cited here.
REFERENCES
Beck HL, Bouville A, Moroz BE, Simon SL. Fallout deposition
in the Marshall Islands from Bikini and Enewetak nuclear
weapons tests. Health Phys 99:124 –142; 2010.
Breslin AJ, Cassidy ME. Radioactive debris from Operation
Castle, Islands of the Mid-Pacific. New York: United States
Atomic Energy Commission, New York Operations Office;
Health and Safety Division report NYO-4623; 1955. [Available from National Technical Information Service.]
Conard RA, Knudsen KD, Dobyns BM. A twenty-year review
of medical findings in a Marshallese population accidentally
exposed to radioactive fallout. Upton, NY: Brookhaven
National Laboratory; Report BNL 50424; 1975.
Eckerman K, Ryman JC. Dose coefficients for external exposure
to radionuclides distributed in air, water and soil. Washington,
DC: U.S. EPA; Federal Guidance Report 12; 1993.
Eisenbud M. Radioactive debris from Operation Ivy. New York:
United States Atomic Energy Commission, New York Operations
Office; Health and Safety Division report NYO-4522; 1953.
Available from National Technical Information Service.
Eisenbud M. Monitoring distant fallout: the role of the Atomic
Energy Commission Health and Safety Laboratory during
the Pacific tests, with special attention to the events following Bravo. Health Phys 73:21–27; 1987.
Henderson RW, Smale RF. External exposure estimates for individuals near the Nevada Test Site. Health Phys 69:715–721; 1990.
Hicks HG. Results of calculations of external gamma radiation
exposure rates from fallout and the related radionuclide
compositions. Livermore, CA: Lawrence Livermore National Laboratory; Report UCRL-53152, Parts 1– 8; 1981.
Hicks HG. Calculation of the concentration of any radionuclide
deposited on the ground by offsite fallout from a nuclear
detonation. Health Phys 42:585– 600; 1982.
August 2010, Volume 99, Number 2
Hicks HG. Results of calculations of external gamma radiation
exposure rates from local fallout and the related radionuclide compositions of selected U.S. Pacific events. Livermore, CA: Lawrence Livermore National Laboratory; Report
UCRL-53505; 1984.
International Commission on Radiological Protection. Conversion coefficients for use in radiological protection against
external radiation. New York: Elsevier; ICRP Publication
74; Annals of the ICRP 26(3/4); 1996.
Jacob P, Rosenbaum H, Petoussi N, Zankl M. Calculation of
organ doses from environmental gamma rays using human
phantoms and Monte-Carlo methods. Part II: Radionuclides
distributed in the air or deposited on the ground. Neuherberg:
GSF-Institut für Strahlenschutz; Report GSF 12/90; 1990.
Land CE, Bouville A, Apostoaei I, Simon SL. Projected lifetime
cancer risks from exposure to regional radioactive fallout in the
Marshall Islands. Health Phys 99:201–215; 2010.
Lessard E, Miltenberger R, Conard R, Musolino S, Naidu J,
Moorthy A, Schopfer C. Thyroid absorbed dose for people
at Rongelap, Utrik and Sifo on March 1, 1954. Upton, NY:
Brookhaven National Laboratory; Safety and Environmental Protection Division report No. 0061648; 1985.
National Council on Radiation Protection and Measurements.
Recommended screening limits for contaminated surface
soil and review of factors relevant to site-specific studies.
Bethesda, MD: National Council on Radiation Protection
and Measurements; NCRP Report No. 129; 1999.
Robison WL, Noshkin VE, Conrado CL, Eagle RJ, Brunk JL,
Jokela TA, Mount ME, Phillips WA, Stoker AC, Smart ML,
Wong KM. The northern Marshall Islands Radiological Survey: data and dose assessments. Health Phys 73:37– 48; 1997.
Sharp R, Chapman WH. Exposure of Marshall Islanders and
American military personnel to fallout. Bethesda, MD:
Naval Medical Research Institute; Report WT-938; 1957.
Simon SL. The role of Merril Eisenbud and HASL in monitoring the Marshall Islands during the Castle nuclear testing
series. Technology 7:145–156; 2000.
Simon SL, Graham JC. Findings of the nationwide radiological
study: summary report. Report submitted to the Cabinet of
the Government of the Republic of the Marshall Islands.
Majuro, Marshall Islands: Ministry of Foreign Affairs,
Government of the Republic of the Marshall Islands; 1994.
Simon SL, Graham JC. Findings of the first comprehensive
radiological monitoring program of the Republic of the
Marshall Islands. Health Phys 73:66 – 85; 1997.
Simon SL, Vetter RJ. Consequences of nuclear testing in the
Marshall Islands. Health Phys 73:66 – 85; 1997.
Simon SL, Bouville A, Land CE, Beck HL. Radiation doses and
cancer risks in the Marshall Islands associated with exposure to
radioactive fallout from Bikini and Enewetak nuclear weapons
tests: summary. Health Phys 99:105–123; 2010a.
Simon SL, Bouville A, Melo D, Beck HL, Weinstock RM.
Acute and chronic intakes of fallout radionuclides by
Marshallese from nuclear weapons testing at Bikini and
Enewetak and related internal radiation doses. Health Phys
99:157–200; 2010b.
Tipton WJ, Miebaum RA. An aerial radiological and photographic survey of eleven atolls and two islands within the
Northern Marshall Islands. Las Vegas, NV: EG&G; Report
EGG-1183-1758; 1981.
United Nations Scientific Committee on the Effects of Atomic
Radiation. Sources and effects of ionizing radiation. New York:
United Nations; United Nations Scientific Committee on the
Effects of Atomic Radiation 1993 report to the General Assembly
with scientific annexes; Publication E 94:IX.2; 1993.
f f