Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Quantification of Personnel Film Badge Uncertainties A. DEVELOPMENT AND APPLICATION OF MEASURES OF TOTAL UNCERTAINTY In Chapter 4, several sources of uncertainty in film badge dosimetry were identified. This chapter quantifies those uncertainties, and assesses their effect, acting jointly, on estimates of exposure and of deep-dose equivalent obtained from personnel film badges (see Section 5.E). The assessment is made specific to each test series, to the magnitude of the estimated exposure, and to other rele- vant conditions surrounding the film badge reading. The uncertainty assessment is accomplished by developing an approach for calculating upper and lower limits for the exposure and deep-dose equivalent based on any film badge reading obtained during atmospheric nuclear tests. The method of calculation is intended to assure that there is a high probability that the limits include the actual exposure and deep-dose equivalent received by the individual. The intervals may be calculated for any specified probability level, with 95% being a common choice. Because the available data are inadequate to quantify all sources of uncer- tainty in a rigorous statistical manner, expert opinion must often be relied on for this assessment. For this reason, the limits are not "confidence intervals" in the classical statistical sense, and are sometimes referred to as "subjective confi- dence intervals." The appropriate interpretation of 95% intervals presented in this report is that, based on a careful assessment by experts of many individual 61

62 FILM BADGE DOSIMETRY IN ATMOSPHERIC NUCLEAR TESTS sources of uncertainty, there is a 95% probability that the interval includes the true value. It is also appropriate to interpret the intervals as indicating that there is only a 2.5% chance that the true value exceeds the upper limit, and only a 2.5% chance that the true value is less than the lower limit. Evaluating Individual Sources of Uncertainty To evaluate overall uncertainty, it is first necessary to specify probability dis- tributions for each uncertainty source; this is accomplished by specifying the probability that the estimated value falls within any specified range. The distri- butions for individual uncertainty sources are then used to evaluate the probabil- ity distribution for all uncertainty sources acting jointly. This process may re- quire complex calculations and possibly computer simulations. A discussion of approaches to uncertainty analyses is provided in a report of the National Coun- cil on Radiation Protection and Measurements (NCRP 1984~. The assessment in this report is based on the use of lognormal distributions for describing uncertainties from individual sources. The lognormal distribution is one in which the logarithms of the estimated values follow symmetric normal distributions, and is symmetric on a multiplicative scale; that is, the probability that an estimated value exceeds F times the median value, is the same as the probability that a value is less than 1/F times the median value. A major advan- tage of the use of lognormal distributions is that uncertainties from different sources can be easily combined without the need for extensive computations. The use of the lognormal distribution for uncertainty analyses is described by the NCRP (1984) and is illustrated by the National Institutes of Health Ad Hoc Working Group to Develop Radioepidemiological Tables (1985~. The general properties of the lognormal distributions are described in detail by Johnson and Kotz (1970) and by Aitchison and Brown (1969~. If the logarithm of an estimate follows a normal distribution with mean, m, and standard deviation, s, then the estimated value follows a lognormal distribu- tion characterized by its median M = em, and by its geometric standard deviation (GSD), S = eS. It is useful to express M as a factor B times the true value, and refer to B as the bias. If B > 1, the true value on average has been overesti- mated; while if B ~ 1, the true value on average has been underestimated. (It should be noted that the mean of the lognormal distribution is a factor e s2/2 higher than the median; however, this source of bias is negligible relative to the overall uncertainty, and can be safely ignored for most purposes). The GSD has the property that two-thirds of the estimated values fall between (1/S)M and SM. If adequate data are available, B and S can be estimated using standard statistical procedures. In the absence of such data, B and S must be estimated, .

5 FILM BADGE UNCERTAINTIES 63 based on judgments of scientists with relevant expertise. One approach is first to provide a subjective assessment of B and then to provide a factor K such that the interval obtained by multiplying the true value by (1/K)B and KB is thought to cover 95% of the estimated values. The GSD and s can then be determined by the relationships K = St96, and s = log(S). The values of K satisfying this relationship are referred to as 95% uncertainty factors. Once the parameters of the lognormal uncertainty distribution have been specified, the P% subjective confidence intervals can be determined as (E/B) e Zp s 5-1 where E is the exposure as determined from the film badge reading, and Zp is an appropriate factor determined from tables of the normal distribution. The values of Zp are 1.960, 1.645 and 1.00, respectively, for 95%, 90%, and 66.7% subjec- tive confidence intervals. Figure 5-1 shows a plot of two lognormal distributions with M = 1. The probability that an estimated value falls between two specified values is repre- sented by the fraction of the total area under these curves falling in the specified region. The plot with K = 1.5 (S = 1.23) represents a modest amount of uncer- tainty with 95% probability that the estimated value falls between 0.67 and 1.5. The plot for K = 4 (S = 2.03) represents a much greater amount of uncertainty; the range 0.25 to 4 is now required to cover 95% of the probability. Combining Several Sources of Uncertainty in One Badge Reading The approach used to combine uncertainties is based on the assumption that the uncertainties from specific sources follow independent lognormal distribu- tions. This assumption of independence requires that the direction and magni- tude of the error from one source have no influence on the direction and magni- tude of error from any other source. This assumption appears reasonable for combining uncertainties from most of the sources considered in this report. For example, it is unlikely that uncertainties resulting from the way a film badge is used in the field would be related to uncertainties resulting from laboratory processing. Where uncertainties from different sources were judged to be inter- dependent, they have been assessed in combination rather than individually. It is assumed that the film badge reading, E, can be written as the product of the true exposure (or dee~dose equivalent) and several factors Ei, i = 1, 2, ... N. It is further assumed that Be Ei follow independent lognormal distributions, that B. and S. represent the bias and the standard deviation on a logarithmic scale,

64 z 2 J 1 a, cr. o Ir o i_ In z LL J a' 1 CD O o FILM BADGE DOSIMETRY IN ATMOSPHERIC NUCLEAR TESTS I \ \ K = 1.5 \ \ \ I VALUE (UNITS ARBITRARY 3 ~_ ~_ K=4 - - - - MEDIAN= 1 FIGURE 5-1 Lognorsnal Distributions. 2 VALUE (UNITS ARBITRARY

5 FILM BADGE UNCERTAINTIES 65 respectively, for distribution i. It then follows that E will also follow a lognor- mal distribution with bias given by B = Hi Bi, with a logarithmic standard devia- tion, s, defined by 2_ 2+ 2+ + 2 S -Sat S2 ... SN and with a P% confidence interval given by (E/B) en Zp s 5-2 . 5-3 This is the same expression as Equation 5-1, but now s and B include uncertain- ties from several sources. It is useful to define the GSD and 95% uncertainty factor from source i as Si and Ki respectively, where Si = e i, Ki = Sit 96, and to define the overall GSD and 95% uncertainty factor by S = es and K = Sib, respectively. Example: A film badge reading, obtained in test A, provides an exposure estimate of O.X R. In Section 5.B, three major categories of uncer- tainty in exposure will be identified: laboratory, radiological, and environmen- tal. The following values for Bi, Ki, and si are typical and illustrate the combin- ing of uncertainty sources. Uncertainty source Bi Ki si Laboratory 1.0 1.2 0.093 Radiological 1.0 1.3 0.133 Environmental 1.2 1.1 0.049 In this case, B = 1.0 x 1.0 x 1.2 = 1.2 S2 = 0.0932 + 0.1332 + o.o492 = 0.0287 s =0.170,K= 1.39 The 95% interval is given by (0.~/1.2)e +i 96 ~ 0.~70 or as (0.48, 0.93~. Without the bias factor of 1.2, the interval would have been (0.57, 1.12), based on an overall 95% uncertainty factor of 1.39. This factor is not as large as the product of the individual Ki, which is 1.72. Because they are uncorrelated, uncertainties from different sources tend to cancel each other out. However, the overall uncertainty factor can never be smaller than the maximum Ki.

66 FILM BADGE DOSIMETRY me ATMOSPHERIC NUCLEAR TESTS In the sections that follow, the estimates of the parameters Bi and si, as well as Si and Ki, are determined for each of several uncertainty sources. These esti- mates are specific to each test series, and in some cases to the magnitude of the estimated exposure. In addition, the calculations necessary to combine uncertain- ties have been performed for the reader. In Chapter 6 tables are provided giving 95% subjective confidence limits for each test series as a function of exposure. Special Problems in the Application of Uncertainty Analyses to Film Badge Dosimetry Because uncertainties in film badge readings are often expressed in a form that is symmetric on an additive scale (e.g., + 50%), the use of the lognormal distribution in this report merits comment. In general, the lognormal distribu- tion, with symmetry on a multiplicative scale, is more appropriate for measures that cannot be less than zero, but with no clear upper bounds. For small uncer- tainties (K < 1.5, or 50% error), the lognormal distribution is very close to a symmetric normal distribution (see Figure 5-1) and thus the two distributions yield similar confidence limits. For large uncertainties, the symmetric normal distribution may permit negative estimates with high probability; this would be inappropriate for many film badge uncertainty sources. Nevertheless, certain sources may be more appropriately described on a sym- metric scale. In these cases, emphasis has been put on determining the correct upper bound; the effect of using a lognormal instead of a normal distribution in such cases will be a lower limit that is too high. For example, if the correct 95% limits for an estimate are M + 1.96 c,, K is taken to be 1 + 1.96 c/M. The upper limit of KM would be correct, but the lower limit (1/K)M is larger than the correct lower limit of M- 1.96 c,. (This result can be shown algebraically, or a few seal values for M and ~ should assure the reader of its validity.) Since laboratory uncertainties at low exposures are likely to be better de- scribed by the symmetric normal distribution than by the lognormal distribution, a special procedure has been used to treat such uncertainties. Note that at low exposures, negative estimates are possible because adjustment for background fog of a film is needed (although such estimates are generally recorded as zero). This special procedure is described in Section 5.B, and provides lower confi- dence limits of zero for very small estimated exposures.

5 FILM BADGE UNCERTAINTIES Uncertainty in Estimates of Total Dose Based on the Sum of Several Film Badge Readings 67 Although the scope of this report has been defined to include only the assess- ment of uncertainty in single film badge readings, uncertainties in the estimates of the total exposure for individual test participants, which are based on the sum of several fUm badge readings, are naturally of interest. Because the sum of several lognormally distributed variables does not follow a lognormal distribu- tion, and because uncertainties from some sources may not be independent for different readings from the same subject, the assessment of uncertainty of the estimated total exposure is complex. The following recommendations are made for assessing uncertainty in the to- tal exposure derived from the sum of several film badge readings. First, it is noted that the interval obtained from the sums of the upper (lower) P% limits for the individual film badge readings may in many cases provide useful limits, es- pecially if the number of readings is small and/or the estimated exposures are small. The confidence level associated with such an interval will be 2 P%, and, because intervals obtained in this manner do not account for possible cancelling of uncertainties as exposures are summed, they will generally be wider than nec- essary. However, if the limits obtained from this approach provide sufficient in- formation for the application of interest, it may not be necessary to proceed fur- ther. When the problem of lognormal summation is encountered, it is often solved by using a Monte Carlo simulation method Wee and Salem 1977~. In the case of film badge readings, it is possible to take advantage of a reasonable approxi- mation that greatly simplifies the calculation. Note in Figure 5-1 that when the uncertainty is relatively small (i.e., the 95% uncertainty factor is 1.5), the log- normal distribution approaches a normal distribution. In this case the mean and median are nearly equal. Thus to a reasonable approximation the mean of the sum is just the sum of the medians. For example, even when the 95% uncer- mnty factor is 2, the largest value encountered in this study, the mean is only 6% larger than the median. If uncertainties in readings from different badges for the same individual are independent, this approach also suggests that to a reasonable approximation, the variance, V, of the sum of M readings is given by V = ~ 1/~1.96~2 ~ ~,j (Kj - 1~2 (Ej/Bj)2 54

68 FILM BADGE DOSIMETRY IN ATMOSPlIERIC NUCLEAR TESTS where Kj, E., and Bj are respectively the 95% uncertainty factor, the film badge reading, and the bias for the jth reading. Approximate 95% confidence intervals for the total are then obtained as Ej/Bj + 1.96 ~ = EjtBj + N/2,j (Kj-1)2(E~/Bj)2 Thus, for example, if an individual's record consisted of the following readings, the total exposure and its 95% confidence interval could be calculated as fol- lows. Reading Ej Bj Kj Ej/Bj (Kj-1)2(Ej/Bj)2 95~o confidence (i) l~rruts for single readings 1 0.1 1.0 2.0 0.10 0.0100 (0.05, 0.20) 2 0.4 1.2 1.2 0.33 0.0044 (0.28, 0.40) 3 0.6 1.0 1.2 0.60 0.0144 (0.50, 0.72) Total 1.03 0.0288 (0.83, 1.32) The resulting confidence limits for the total, based on the assumption of inde pendence of uncertainties in the three readings, are 1.03 + ~ or (0.86, 1.20~. These limits are narrower than the more conservative limits (0.83, 1.32) obtained by summing the lower and upper limits from the three readings. B. CATEGORIES OF UNCERTAINTY The sources of uncertainty in radiation exposure determined from film badge dosimetry have been grouped into three categories: laboratory, radiological, and environmental. Uncertainties associated with each of these will be combined as described in Section 5.A. The three categories are interpreted as follows: Laboratory Uncertainties This category includes all the uncertainties introduced in film calibration, chemical processing of films, reading their optical densities, comparing these densities with the densities of unexposed and calibration films, and in interpret- ing the measured densities in terms of exposure. Even under the best controlled laboratory conditions, laboratory uncertainties are a strong function of exposure level, particularly at low exposure levels. This behavior is evident from the general mathematical form of the variation of film optical density, D, with exposure:

5 FILM BADGE UNCERTAINTIES 69 D=D (l-e~7E), 5-6 where D is the saturation density of the film at high exposures, E, and yis the sensitivity of the film. For the Du Pont Type 502 film illustrated in Figure 4-3, Do = 2.8, and it= 0.25 with exposure expressed in R; other films of comparable sensitivity to that of the Type 502 film should yield similar values. If it is assumed that the standard deviation of the measured optical density does not depend on exposure, and is given by a constant c' *, the standard devia- tion of the measured exposure, Is (E), can be shown to be approximately equal to {~*/(D T)le7E 5-7 If it is further assumed that measured exposures are approximately normally dis- tubuted, the upper confidence limit (for two-sided 95% limits) are given by E + 1.96 c, (E). The 95% uncertainty factor (the factor needed to multiply the mea- sured exposure, E, to obtain this upper limit) is then given by K(E) = 1 + 1.96 ~ (E)/E. 5-8 Because replicate density readings at the same exposure generally yield val- ues within + 0.03 density units, it is reasonable to take ~ * = 0.015. With the values of Do and ~ given above, we have K(E) = 1 + 0.042 e0 Is E/E. 5-9 In Figure 5-2 this K(E) is plotted (solid line) as a function of exposure. The 95% uncertainty factors K(E) for exposures between 0.5 R and 14 R are less than 1.1, with a minimum value of 1.03 at 4 R. However, below 0.2 R and above 14 R the uncertainty rises rapidly. In general, the exposure levels that delineate the useful range of the film, (small K(E)) depend on the sensitivity of the film. If a badge contains more than one film component, the overall exposure uncertainty using both film components may have a peak in the region of over- lap (see section 4.D) of the different components. The low sensitivity Du Pont Type 606 film, part of whose exposure curve is shown in Figure 4-3, has a D = 3.0 and a y= .006. The uncertainty, K(E), vs exposure, E(R), for this film (from Equation 5-8 with ~ * = 0.015) is shown by the dashed line in Figure 5-2. For the two film components shown in Figure 5-2, the overlap between the two films is sufficiently good that there is only a small rise in K (to K = 1.2) at the high- exposure end of He 502 film and the low-exposure end of the 606 film. If the

70 FILM BADGE DOSIMETRY IN ATMOSPHERIC NUCLEAR TESTS 508-1290 film combination shown in Figure 4-2 had been used, there would have been a much larger peak in K in the region of overlap. The intrinsic uncertainties in determining exposure discussed above are in- creased by uncertainties in the radiation field and in the time used in calibration, by variations in film processing conditions if calibration and unexposed films are not processed with each batch of field exposed film badges, and by possible inaccuracies in reading a calibration curve. For these reasons, the minimum laboratory uncertainty is never estimated to be as low as 1.03. Under controlled laboratory conditions it is conservatively estimated to be at least 1.2. Under less favorable conditions in some test series the minimum K is even larger. In almost all cases, the intrinsic uncertainty dominates at low exposures. The value of K for laboratory uncertainty is deduced as the appropriate combination (see Section 5.A) of the intrinsic uncertainty and estimated uncertainties in process- ing, calibration and interpretation. Unless stated otherwise, uncertainties for exposures in the overlap region of two different films were based on a K of 1.5 for laboratory uncertainty. 2.0 ' 1.8 - ~ 1.6 Z c' 1.4 By ~ . 1.2 1.0 \ DuPont 502 \ \\ DuPont606 I I I I ~1 1 - I ~ 10 100 .01 .1 EXPOSURE, E(R) FIGURE S-2 Plot of Uncertainty, K(E) vs Exposure, E(R) for DuPont 502 and 606 Film Components.

5 FILM BADGE UNCERTAINTIES 71 In Chapter 6, values for laboratory uncertainty are presented for each test series and are intended to be applied for exposures over 0.2 R with special consideration of larger exposures as indicated. As noted above, these laboratory uncertainties are never less than K= 1.2. The 95% uncertainty factor for the additional uncertainty for exposures below 0.2 R is obtained as K*(E) = e4(ln2K(E) - 1n21.2' 5-10 where K(E) = 1 + 0.042 eQ25E/E for Du Pont 502 film. Values of K*(E) are given in Table 5-1. TABLE 5-1 Additional Uncertainty Factors for Film Badge Readings Below 0.2 R Em) Kim) Kim) 0.02 3.11 3.07 0.04 2.06 2.01 0.06 1.71 1.66 0.08 1.54 1.47 0.10 1.43 1.36 0.12 1.36 1.28 0.14 1.31 1.22 0.16 1.27 1.17 0.18 1.24 1.13 These factors are to be combined with uncertainties from other sources as usual, including the "standard" laboratory uncertainty factor, which is 1.2 or 1.3 for most test series. Special treatment is required below the minimum detectable level ~L). The MDL is the minimum exposure that can be statistically distinguished from zero in the laboratory. It is usually established at the point where the laboratory uncertainty is + 100% at the 95% confidence level (see Section 5.C). It should be noted that the expression "minimum detectable level" is often used in a less precise sense; thus the MDL values indicated in various documents describing test series may not satisfy the above definition exactly.

72 FILM BADGE DOSIMETRY IV ATMOSPHERIC NUCLEAR TESTS For Du Pont 502 film, the MDL must satisfy MDL = 0.042 en 25~DL 5-11 implying that the MDL is approximately 0.04 R. In obtaining 95% confidence limits for recorded exposures below the MDL, the lower limit should be taken to be zero. To avoid problems of applying multiplicative factors to exposure esti- mates of zero it is recommended that exposures recorded as less than the MEL be considered as half the MDL, for purposes of defining the additional labora- tory uncertainty factor K*(E) and for calculating the upper subjective confidence limit, including all uncertainty sources. Note that this treatment of laboratory uncertainties at low doses is a departure from the use of the lognormal distribu- tion in that it allows for the inclusion of zero in the confidence limits. Labora- tory uncertainties may be better described by the symmetric normal distribution than by the lognormal distribution. To illustrate the above procedure, suppose that the worker in the example given in Section 5.A had a film badge exposure of 0.1 R instead of 0.8 R. For this exposure K(E) = 1 + 0.042e`°~5x~/O.1 = 1.43, and K*(E) = e,/(1n21 43 - in21 2) = 1.36, with corresponding s*(E) = (in 1.36~/1.96 = 0.157. If this additional uncertainty is added to that in the example, the overall s2 is 0.1702 + 0.1572 = 0.0535, and s = 0.231, K = 1.57. The 95% subjective confi- dence limits for exposure are (0.05, 0.13~. Radiological Uncertainties Three sources of uncertainty have been identified in the radiological cate- gory: photon energy spectrum, body wearing position and radiation backscatter. The influence of the low energy part of a photon energy spectrum on film badge exposure has been discussed in Section 4.A, particularly with regard to the thickness and material of the filter used to attenuate the lowest-energy pho- tons. A 0.028 inch thick lead filter was found to minimize the uncertainty in the exposure caused by uncertainties in the energy spectrum (see Section 4.A), but even at this thickness there is a residual bias B that is estimated to be 1.1 and an uncertainty in the consequences of the spectrum on the measured exposure which is estimated to give a K of no less than 1 2. For thinner and lower-atomic-num- ber filters such as used in early test series, both the B and K values are estimated to be larger. A film badge is normally expected tube worn on the chest. At such a posi- tion it is not experiencing the same radiation field as if it were freely exposed in air because the body attenuates radiation from the back. The magnitude of the

5 FILM BADGE UNCERTAINTIES 73 bias in the measured exposure clearly depends on the energy spectrum of the photon radiation, the spatial distribution of the radiation field and on the size of the wearer. It is estimated to have a typical value of B = 0.8 and an uncertainty associated with this effect for which a typical value of K = 1.1 is estimated. This uncertainty includes allowance for improper wearing, e.g., attached to the belt or carried in a pants pocket. The presence of the body on which a badge is worn increases the radiation field (as well as decreasing it due to attenuation, as discussed above) because the body backscatters photons. This is estimated to contribute a typical B of 1.1 and to have an uncertainty of at least K = 1.1 as well. Notice that the net effect of the wearing and backscatter contributions to the radiological effect with the above values of B tend to compensate in bias (1.1 x 0.8) but their K uncertain- ties are cumulative (not compensating). The radiological situation for pilots and other crew members exposed to radiation in aircraft is different from that for personnel exposed on the ground or on ships. The structure of-an airplane provides substantial shielding to persons within and preferentially removes low-energy photons from the spectrum and thus reduces the bias due to the low-energy spectrum (toward 1.0~. The shield- ing is greater from behind and below as a result of the seat. This increases the value of B attributed to body shielding, "wearing", toward 1.0. The reduction in the low-energy part of the photon spectrum also reduces the B due to backscatter (toward 1.0~. The net effect of the three radiological contributions on the overall radiological B is not very different from those for ground personnel. Because aircraft personnel have relatively little mobility within an airplane, there is less uncertainty associated with the radiological effects than for typical ground personnel. Therefore, film badge readings for aircraft personnel are more reproducible measures of exposure than for ground personnel, and perhaps more accurate as well. Nevertheless, in order to provide a conservative estimate of uncertainty, the same values of K are adopted as for Wound personnel in most test series. Exceptions are IVY and ~MBLER-SNAPPER where special condi- tions warranted special treatment. Environmental The final category of uncertainty combines all those uncertainties related to the field environment in which film badges are exposed. Section 4.G discusses the consequences of exposure to moisture, light, high temperatures, and radioac- tive contamination. As noted in that section, with expert examination of pro- cessed films, these effects can often be recognized and even taken into account. However, in some of the test series in this report where environmental effects were known to be present, it is not reasonable to conclude that such expert ex

74 FILM BADGE DOSIMETRY TV ATMOSPHERIC NUCLEAR TESTS aminations were made and reinterpretation is not feasible nor even possible be- cause many of the original films are no longer available. The environmental bias and uncertainty are estimated from a knowledge of the environmental condi- tions of each test series. In general, these were quite different for test series con- ducted in the Pacific, where conditions of high humidity prevailed, than for test series conducted at the NTS. These differences are reflected in the estimates of uncertainty in individual test series. For GREENHOUSE and TUMBLER-SNAPPER, fallout contamination in- creased the uncertainty in estimates of low-dose exposures. These effects are in- cluded under the environmental contributions to uncertainty and are discussed for those test series. Environmental conditions for personnel exposed to radiation while in aircraft was different from that for ground personnel in several respects. Badges usually were issued and collected on a daily basis, so no long-term environmental ef- fects took place. There was no effect attributable to high humidity or tempera- ture and there was no fallout on the badge of a wearer. As a result, environ- mental uncertainty in determining the exposure of pilots and other crew mem- bers is lower than for ground personnel. Consequently, uncertainties estimated for the latter provide a conservative estimate for the former. Exceptions occur in the cases of the IVY and TUMBLER-SNAPPER tests. Aircraft ground crews who often encountered radiation as they cleaned aircraft flown in proximity to nuclear tests, or as they removed air filters used to collect radioactive debris from detonation clouds, are estimated to have bias and uncertainty values associ- ated with their dosimetry readings that are similar to those of other ground personnel. C. MINIMU1VI DETECTABLE LEVEL OF RADIATION EXPOSURE MEASURABLE WITH A FILM BADGE As described in Section 5.B, the laboratory uncertainty factor increases as film badge exposure readings approach zero, and this results in a level below which readings are not statistically distinguishable from zero. This minimum detectable level (MEL) is usually established at the point where the laboratory uncertainty of the reading at the 95% confidence level is + 100% in normal dis- tribution terms. A series of exposures at the MDL should yield film badge read- ings, 95% of which would fall between 0 and twice the MDL, and which follow a symmetric normal distribution. Because the uncertainty of readings below the MDL is greater than the reading such readings are indistinguishable from zero or the MDL itself. Exposures midway between zero and the MDL are as likely to be interpreted as zero as they are to be read at the MDL. Similarly, as exposures increase to the MDL, they are more likely to fall into the readable range just as those am

5 FILM BADGE UNCERTAINTIES 75 preaching zero are more likely to be interpreted as zero. The general practice in film badge dosimetry is therefore to make the best possible interpretation of the exposures in this region, reporting zero for those that favor that end of the range and a positive reading for those approaching the MEL, bearing in mind that there is no statistical difference between the two. In Section 5.B, the Committee suggested using one-half the MDL for deter- mining the upper limits in a consistent manner for exposures reported below the MDL. It should be noted that this does not imply a recommendation to modify the existing records of exposures recorded below the MOL. D. COMMONALITY AMONG THE TEST SERIES The particular personnel firm badge selected for one multiple-detonation test series or single-detonation testing operation was not always the same as the next. Film badge use, however, included identical firm badges or film packets for some series, the same containers for firm packets during several series, and the same metallic filter during most series. After the third test operation, SAND- STONE, only single film packets containing two or three film types, or compo- nents, were used in personnel film badges. Environmental conditions during the use of film badges in atmospheric test- ing were similar within each of two categories, continental and oceanic testing locations. Except for the first nuclear detonation, TRINITY, in an arid New Mexico location, the remaining continental atmospheric test series were con- ducted in Nevada at either Frenchman Flat or nearby Yucca Flat in a semi-desert environment. Oceanic test operations were all at Pacific locations, except for ARGUS detonations which occurred outside the atmosphere above the Atlantic. Environmental effects on personnel film badges, therefore, were comparable during operations on the continent and similar during oceanic operations, where different protective measures against environment film damage were employed. Film badge calibration and processing techniques during the test operations were similar and became more uniform as testing continued. Radium 226 cali- bration sources were common in early test operations. These generally were re- placed by cobalt 60 sources later, but techniques for film badge calibration in air were similar. Radium exposure rates were calculated during early series, and both radium and cobalt exposure rates in later operations were related to NBS calibrations either by direct NBS-calibrations or by use of NBS-calibrated "R- meters". Processing became fairly uniform after the first few series. An important change was maintaining developing solutions within + 0.5°F rather than within + 1°F, as during CROSSROADS. Another important evolution was developing standard calibration films with known exposures for each developed batch of

76 FILM BADGE DOSIMETRY IN ATMOSPHERIC NUCLEAR TESTS personnel films, in addition to unexposed control films, to monitor and adjust for variations in the developing solutions and processing. Perhaps the most important common factor in personnel film badge use is the characteristic shape of the H & D curve for any personnel dosimeter film type (see Figure 2-3~. This leads to a uniform variation of laboratory uncertainty ver- sus exposure for each film type (see Section 5.B). As previously mentioned, the ARGUS I, II, and III events were detonations outside the atmosphere, high above the earth. Detonation yields were between one and two kilotons for each test, and no fallout was detected at the earth's sur- face. Film badges were worn during Operation ARGUS, but no personnel doses were recorded from ARGUS fallout. Thus, any discussion of exposure assign- ment accuracy during ARGUS is moot. Personnel-dosimetry accuracy during the Plowshare program tests (GNOME, SEDAN) is not discussed separately because these detonations were not part of the atmospheric test senes, but were underground tests between atmospheric test senes. The film badge used and the associated processing program during both Plowshare tests were the same as were used in DOMINIC II, and the same uncertainty considerations apply. E. CONVERSION FROM EXPOSURE TO DEEP-DOSE EQUIVALENT Dunng the period of atmospheric nuclear testing? film badge results were customarily expressed in roentgen, R. the unit of the radiological quantity, expo- sure. This approach proved useful as a quantitative means to control and limit the radiation exposure received by test participants. However, exposure is a measure of the electrical charge created by ionization of air by x or gamma radiation, and as such only indirectly reflects the amount of radiation energy absorbed within the body, or the risk of an adverse biological effect. The Committee therefore related film badge readings to deep-dose equivalents that are more relevant to health effects. By converting film badge exposure to deep- dose equivalent, the film badge readings from atmospheric nuclear tests are easily compared to current results for other activities, including underground weapons testing, nuclear power plant operation, diagnostic radiology, and nu- clear medicine. Procedures have been developed for conversion among the various radiologi- cal quantities defined for external radiation. Use is made of extensive computer calculations because some of the quantities cannot be directly measured. Where measurements have been made, there is good agreement with calculations. Pub- lication 51 of the International Commission on Radiological Protection (ICRP 1987) is the most recent compilation of relevant data.

5 FILM BADGE UNCERTAINTIES 77 Several factors must be considered when making conversions among the various quantities. Among these are: type and energy of radiation exposure geometry · dose equivalent of interest The type and energy of radiation were established by radiation conditions at nuclear tests and have been described in Chapter 3. Undoubtedly many exposure geometries were encountered, but an area (ex- tended plane) source of photons from the radioactive products of a nuclear detonation seems to be most representative. (Beta-particle exposures were not adequately monitored and are excluded in the dose assessments). ICRP Publica- tion 51 presents conversion factors for different geometries. Geometries that are most frequently evaluated are: · Anterior-posterior (front to back irradiation) · Posterior-anterior (back to front irradiation) · Lateral (irradiation from the side) · Rotational (uniform irradiation from front, back and sides as would occur if one stood in a cylinder made of radioactive material or if a vertical line source was rotated about oneself) Isotropic (uniform irradiation from the front, back, sides, top and bottom as would occur if one was suspended in a uniformly radioactive cloud) . None of the above is truly representative of the area source most commonly en- countered in atmospheric weapons testing. The rotational geometry was selected as the best approximation to the area- source geometry, although the first three geometries are inappropriate because the radiation is too directional, i.e. personnel entering a contaminated area would not be irradiated from one side only. Compared to the isotropic geometry, the other reasonable alternative, the dose to various body organs per unit exposure is greater for the rotational geometry. Furthermore, the isotropic condition was re- jected because uniform exposure from the top and bottom at the same time was not likely, even for pilots submerged in a radioactive cloud. The rotational condition appears to offer the best compromise between conservatism and appli- cability. The Individual Dose Equivalent, Penetrating, Hp. (as defined in ICRU 1985) was selected as the endpoint dose equivalent quantity for this study. This is the operational quantity for personnel monitoring. Hp(10) is the dose equivalent from penetrating radiation to soft tissue located at a depth of 10 mm in the body.

78 FILM BADGE DOSIMEI RY IN ATMOSPHERIC NUCLEAR TESTS Also called the deep-dose equivalent, Hp(10) can be evaluated with film badges or other types of personnel dosimeters. Such devices are normally calibrated using body phantoms to simulate backscatter conditions. A 30-cm-diameter sphere or a 30 cm x 30 cm x 15 cm slab of tissue-equivalent material is com- monly used. The deep-dose equivalent is also the quantity specified for per- formance testing of personnel dosimetry systems by the American National Standards Institute (ANSI 1983) and the Department of Energy (DOE 1986~. The Effective Dose Equivalent, He, and the dose equivalent to specific organs (e.g., the red bone marrow) were considered by the Committee but not selected for conversion. The effective dose equivalent is a conceptual quantity estab- lished by the ICRP. It cannot be measured, only calculated (ICRP 1977~. It is defined as the sum of the weighted dose equivalents for the major radiosensitive organs that exhibit stochastic (carcinogenic or genetic) effects. Weighting is based on the relative risk of a stochastic death per unit dose equivalent to the various tissues or organs. The effective dose equivalent is thus the quantity that most closely associates exposure to radiation with the risk of an adverse biologi- cal effect. Its advantage is that it provides a mechanism for combining dose equivalents from uniform and non-uniform body irradiation from either external or internal sources, to arrive at a single risk estimate. The deep-dose equivalent, however, is more practical, has been in use for several years, and is implicit in current regulations. The relation between effective dose equivalent and deep-dose equivalent for anterior-posterior and rotational exposure geometries is presented in Table 5-2. For the rotational geometry, the deep-dose equivalent and the effective dose equivalent are nearly identical for photon energies above 0.08 MeV. The deep- dose equivalent overestimates the effective dose equivalent by 10% to 15% for anterior-posterior irradiation. Table 5-3 relates the deep-dose equivalent to the quantity exposure for rota- tional irradiation. The quantitative value of the deep-dose equivalent (in rem) is 70 - 80% of the value of the exposure (in R) for the photon energies associated with nuclear weapons tests. Consequently, the Committee selected a bias of B. = 1.3 for converting film badge exposure data to deep-dose equivalent. A value of 1.2 was selected as the uncertainty factor (K) at the 95% confidence level to account for possible dissimilarities of irradiation geometries actually encountered and those assumed for the conversion, as well as variations in the shapes and sizes of people. Deep-dose equivalent does not indicate dose equivalent to specific organs. To assess the risk of a clinically detectable effect (e.g., cancer) to a specific organ, it necessary to estimate the dose equivalent to that organ. Calculations may be performed to estimate an organ-dose equivalent from deep-dose equiva- lent. Tables 5-4 and 5-5 are examples for red bone marrow and lung, respec- tively, for rotational irradiation.

5 FILM BADGE UNCERTAINTIES Table 5-2 The Ratios of Effective Dose Equivalent H to the Deep-Dose Equivalent, Hp~l6)a 79 Table 5-3 The Ratio of the Deep-Dose Equivalent Hp(10) to Exposurea Anterior- Ratio for Photon Posterior Rotational Photon Rotational Energy (Mev) Lrradiation Irradiation Energy (MeV) Irradiation, (rem/R) 0.05 0.08 0.10 0.20 0.40 0.60 0.80 1.00 2.00 0.73 0.85 0.87 0.87 0.87 0.88 0.89 0.90 0.90 0.87 1.06 1.09 1.06 1.03 1.03 1.02 1.02 0.99 aCalculated from data presented in ICRP Publication 51 (ICRP 87). Table 54 The Ratio of the Red Bone Marrow-Dose Equivalent to the Deep-Dose Equivalent, HptlO)a o.os 0.08 0.10 0.20 0.40 0.60 0.80 1.00 2.00 0.69 0.78 0.77 0.71 0.70 0.70 0.71 0.72 0.77 aCalculated from data presented in ICRP Publication 51 (ICRP 1987). Table 5-5 The Ratio of the Lung-Dose Equivalent to the Deep-Dose Equivalent' Hp~lO)a Ratio for Ratio for Photon Rotational Photon Rotational Energy (MeV) Irradiation Energy (MeV) Irradiation 0.05 0.08 0.10 0.20 0.50 1.00 0.69 0.92 0.99 1.04 1.00 1.00 aCalculated from data presented in ICRP Publication 51 (ICRP 19873. 0.05 0.08 0.10 0.20 0.50 1.00 0.93 1.12 1.14 1.12 1.08 1.07 aCalculated from data presented in ICRP Publication 51 (ICRP 1987).