3
Thermal-neutron and Fast-neutron Measurements

Neutrons are central to the operation of the atomic bomb. Fast neutrons are emitted in fission; for every fission, more than a single neutron leaks out of the active material. Measuring those neutrons can indicate the bomb yield directly. However, many fast neutrons can be greatly degraded in energy by materials in the bomb, such as the metal in the casing of the Hiroshima bomb, and by the environment, such as high-explosive gases in the Nagasaki bomb.

DS86 relies on calculations of neutron and gamma-ray fluence and kerma. Although the uncertainty in these calculations can be estimated from the combined uncertainty in the various components of the system, the ability to validate various aspects of the radiation-transport method by comparisons with measurements is essential for demonstrating the overall validity of the method used in DS86. An important comparison with respect to the neutron-fluence calculations at various distances in free air is that between calculated and measured thermal-neutron (low energy) and fast-neutron activation of rocks, building materials, and so on. The neutron dose is smaller than the gamma dose and the neutron kerma (or neutron dose) to survivors results primarily from exposure to higher-energy neutrons. Even thermal-neutron measurements are valuable in testing the transport in air because the thermal-neutron fluence at any site results primarily from downscattering of higher-energy neutrons. Thermal and epithermal neutrons have ranges of only a few meters in air and thus are produced locally. However, the thermal and epithermal fluence incident on a given sample can vary substantially because of variations in local scattering and downscattering. Furthermore, there could be errors in the transport system that affect only the lower-energy neutron transport and not the neutron-dose estimates. Thus, it is desirable to measure the higher-energy fluence directly with activation reactions that have a relatively high-energy threshold.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 3 Thermal-neutron and Fast-neutron Measurements Neutrons are central to the operation of the atomic bomb. Fast neutrons are emitted in fission; for every fission, more than a single neutron leaks out of the active material. Measuring those neutrons can indicate the bomb yield directly. However, many fast neutrons can be greatly degraded in energy by materials in the bomb, such as the metal in the casing of the Hiroshima bomb, and by the environment, such as high-explosive gases in the Nagasaki bomb. DS86 relies on calculations of neutron and gamma-ray fluence and kerma. Although the uncertainty in these calculations can be estimated from the combined uncertainty in the various components of the system, the ability to validate various aspects of the radiation-transport method by comparisons with measurements is essential for demonstrating the overall validity of the method used in DS86. An important comparison with respect to the neutron-fluence calculations at various distances in free air is that between calculated and measured thermal-neutron (low energy) and fast-neutron activation of rocks, building materials, and so on. The neutron dose is smaller than the gamma dose and the neutron kerma (or neutron dose) to survivors results primarily from exposure to higher-energy neutrons. Even thermal-neutron measurements are valuable in testing the transport in air because the thermal-neutron fluence at any site results primarily from downscattering of higher-energy neutrons. Thermal and epithermal neutrons have ranges of only a few meters in air and thus are produced locally. However, the thermal and epithermal fluence incident on a given sample can vary substantially because of variations in local scattering and downscattering. Furthermore, there could be errors in the transport system that affect only the lower-energy neutron transport and not the neutron-dose estimates. Thus, it is desirable to measure the higher-energy fluence directly with activation reactions that have a relatively high-energy threshold.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) FAST-NEUTRON ACTIVATION MEASUREMENTS Following the Hiroshima explosion, Japanese physicists made a number of measurements of the fast-neutron activation of 32S near the epicenter (Roesch 1987). The sulfur was contained in insulation material of electric-power poles. Japanese investigators made these measurements during the first few weeks after the event. The threshold for 32S activation is about 2.5 MeV, and the half-life of the 32P produced is only 14 d. The calculated and measured 32S activation tended to agree well close to the epicenter, particularly when corrections were made for the expected anisotropy due to bomb tilt (Roesch 1987). The exact degree of agreement, however, depends on the assumed yield. In fact, the comparison between 32S activation calculation and measurement was a factor in determining the yield that was used in DS86. The 32S data tended to diverge from DS86 calculations, and the calculated values appeared to be lower than measured by an amount that increased as the distance from the epicenter increased. However, the measured activities at these distances were very low, and the uncertainties very high. Because of the short half-life of 32P, the 32S activation measurements could not be repeated to confirm the original data. Documentation of similar measurements made at Nagasaki has not been found. THERMAL-NEUTRON ACTIVATION MEASUREMENTS When DS86 was released, a number of thermal-neutron activation measurements had been made at various slant ranges at Hiroshima and Nagasaki. Additional measurements have since been made of thermal-neutron activation of cobalt (Co) and europium (Eu) and, with a different technique, the generation of 36Cl by thermal neutrons. Those measurements have indicated that the thermal neutrons were more abundant at great distances than was predicted from the neutron spectrum calculated for the bomb explosion by DS86. It appeared that more high-energy neutrons penetrated the iron casing than was calculated. In general, the measured thermal-neutron activation of 60Co at Hiroshima appeared to be higher than the calculated values by an amount that increased as the slant range increased. Although some 152Eu thermal-neutron activation data were also reported, the data for larger distances were too few to confirm the 60Co comparisons. Measurements of 60Co and 152Eu activation at Nagasaki also suggested the possibility of a similar trend, but the data for distances greater than 1000 m was sparse. Near the epicenter at both Hiroshima and Nagasaki, 60Co and 152Eu activation data tended to be about 50% lower than calculated from the DS86 neutron fluence. RERF has surveyed the literature and communicated with investigators directly and has created a database of all known activation measurements (see Appendix A). Many of the newer measurements were made at increasing distances from the epicenter to resolve the apparent discrepancy observed in the DS86 calculation-

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) measurement comparison.1 A report including recent 36Cl measurements (Straume and others 1992) confirmed the divergence of DS86 calculations and measurements at great distances in Hiroshima and suggested that the differences could exceed a factor of 10 at slant ranges greater than 1500 m. Newer data for Nagasaki, in conjunction with improved radiation-transport calculations (Kaul and others 1994; Straume and others 1994) appeared to show good agreement between calculation and measurement at all distances, although the data for greater distances were still relatively sparse. However, very recently reported 152Eu activation data for Nagasaki (Shizuma 2000a) again suggest the possibility of similar, although perhaps smaller, discrepancies at great distances. The thermal-neutron activation measurements near the epicenter at Hiroshima continued to be about 50% lower than the calculations. Since Straume and others reported the apparent discrepancy in thermal-neutron activation calculations at Hiroshima, numerous studies have attempted to explain the lack of agreement. The present committee has examined all the proposed solutions and some new data that have become available since the original publication by Straume and others (1992). The committee and its consultants have examined the various measurement data in great detail to determine whether part or all of the disagreement could be due to measurement errors (including failure to account properly for background contribution) and to determine better the exact extent of the potential discrepancy in calculated DS86 neutron fluence as a function of distance. Previous assessments have treated all the data as essentially equivalent with respect to accuracy. Because of the varied quality of the reported measurements, the degree of the possible discrepancy might have been overestimated at the greater distances, where many of the data have large uncertainties. FAST-NEUTRON ACTIVATION MEASUREMENTS OF 63NI An important new development is the ability to measure the production of 63Ni in copper samples (n,p). Because the activation takes place only at energies above about 1.5 MeV, such measurement provides a method for confirming the 32S measurements at Hiroshima and for directly measuring the high-energy neutron fluence at large distances from the epicenter. One technique (Straume and others 1996; Ruehm and others 2000b) uses accelerator mass spectrometry (AMS). It is very sensitive and is able to give results on fast-neutron intensity from a few hundred meters to 1500 m. 1   The term discrepancy used in this discussion describes the trend in the measurements in contrast with DS86-calculated values. This is related loosely to the notion that the measurements as a function of slant range, in a semilogarithmic plot, have a shallower slope, corresponding to a greater “relaxation length,” than the calculated values. The other aspect of this trend, which is often left implicit, is that the measured and calculated values tend to be equal not near the source, but at some middling slant range of about 800 to 1000 m.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 63Ni can also be assayed by direct beta-counting (Shibata and others 1994); but because of the long half-life (100 y) and low specific activity, the measurement sensitivity is low except fairly close to the epicenter, where the neutron fluence was high. Preliminary measurements have however been made by this method (Shibata 2000). 63Ni measurements should permit an empirical determination of the numbers of fast neutrons emitted by the bomb. Together with the additional thermal-neutron measurements on 36Cl also being carried out now and a careful reevaluation of the reported thermal-activation data, a complete verification or determination of the neutron spectrum at Hiroshima might become possible, permitting the determination of neutron kerma at the important distance of 1000–1500 m, where the average total doses to survivors lie between 0.2 and 2 Gy. With the new measurement of fast neutrons using 63Ni, it is hoped that it will be possible to reconstruct a neutron source directly from the fast-neutron measurements combined with the augmented thermal-neutron data. Thus, the problem in Hiroshima is primarily to explain the neutron discrepancy, assuming that it is real. Previous attempts to provide a source term that might fit the neutron data have led to a neutron source that seemed physically unacceptable (see Chapter 4). The most penetrating fast neutrons in air are above 2.3 MeV. Thus, the basic problem in the neutron measurement is that the long relaxation length in air implied by the thermal-neutron activation data suggests a neutron source with a significantly greater number of neutrons at energies above 2 MeV than expected. But, only neutrons at the same or higher energies cause the capture of neutrons by sulfur and copper. It is exceedingly difficult to construct a neutron source that can provide enough thermal neutrons at 1500 m and not have too large a sulfur capture. It is hoped that the new neutron measurements will allow us to revisit and resolve this discrepancy. The remainder of this section discusses the committee’s evaluation of the various reported activation data and their estimated uncertainty. We estimate the minimum detectable concentrations (MDCs) of the various measurement protocols for different investigators. A number of measurement problems might have resulted in reported activation data that were biased high at low activities. On the basis of this evaluation, we provide our current best estimate of the degree of the calculation-measurement discrepancy relative to distance at Hiroshima and Nagasaki on the basis of best fits of activation measurements to calculated activation as a function of slant range, considering measurement and calculation uncertainties and appropriate background corrections. Finally, we recommend additional measurements that are required to refine the estimate of the discrepancy at Hiroshima. SUMMARY OF AVAILABLE NEUTRON ACTIVATION MEASUREMENTS Thermal-neutron and fast-neutron measurements have been reported on the basis of various reactions. These are summarized in Table 3–1.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) TABLE 3–1 Thermal-neutron and Fast-neutron Reactions Used for Measuring Neutron Fluence at Hiroshima and Nagasaki Reaction Threshold T 1/2 σth (barns)a σth (barns)b %c 59Co(n,γ)60Co Thermal 5.3 y 37 30 27 151Eu(n,γ)152Eu Thermal 13.4 y 5900 4400 13 153Eu(n,γ)154Eu Thermal 8 y 346 320 41 35Cl(n,γ)36Cl Thermal 3×105 y 42 28 8 63Cu(n,p)63Ni 1.5MeV 100 y — — — 32S(n,p)32P 2.5 MeV 14 d — — — aThermal capture cross section at 300°K. bThermal-neutron cross section averaged over energies less than 0.4 eV (Kaul and Egbert 1989). cFraction of activation from neutrons above thermal, estimated (Kaul and Egbert 1989). A description of the various samples analyzed, their locations, sample type, degree of shielding, and so on, is included in the database assembled and maintained at RERF. Appendix A describes this database and contains a list of the literature and other sources of the measurements in the database. Additional measurements are still in progress and will be added to the database when available. To evaluate the various data, a questionnaire was sent to each investigator reporting activation data (see Appendix A). The responses to the questionnaire are also included in the database. Letters requesting additional information regarding measurement protocols, uncertainty analyses, and background subtractions were sent to select investigators to clarify various issues. Committee consultants also visited the laboratories of several of the investigators and interviewed the principal investigators. Not all the investigators responded fully to the questionnaire or the followup requests for clarification. Because the published material did not usually contain sufficient information to evaluate the total uncertainty in reported measurements, the committee and its consultants were required to make their own estimates, as described in Appendix B. Appendix B also discusses the definition and estimates of MDCs for each investigator. For the committee to provide its best assessment of the most critical data sets—those at sites greater than 1000 m—it is essential that investigators be encouraged to provide the necessary information and, as discussed later, agree to cooperate in sharing samples and participating in comparisons. A number of the measurements described in the RERF database were heavily shielded or were measurements at increasing depth in cores from bridges or buildings. The comparisons between measurement and calculation for these samples involve a shielding correction that was made with the DS86 fluence calculations, so these data have not been used in our assessment of the extent of the disagreement between DS86 calculations and measurements. Any error in the DS86 energy spectrum of incident neutrons would confound the comparison of free-field calculated and measured fluence. Although even the calculation of activation in near-surface samples involves some additional calculation uncertainty,

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) these corrections are (as discussed later) relatively small compared with the apparent discrepancy. They do, however, need to be considered in evaluating the extent of the potential discrepancy in the DS86 neutron-fluence calculations. Appendix A discusses all the available reported activation measurements that the committee believes were at locations in the direct line of sight of the epicenter and were minimally shielded (that is, near the surface). Preliminary 63Ni data have been provided to the committee and are discussed in this report (Ruehm 2000; Straume 2000a), but they have not been reported and must therefore be considered tentative. Additional preliminary 36Cl data have also been provided to the committee and are discussed in this report (Straume and others 2000), but these data also must be considered tentative and subject to revision based on additional measurements and calculations. COMPARISON OF ACTIVATION MEASUREMENTS WITH DS86-BASED CALCULATIONS Figures 3–1 through 3–4 compare the measured activation of 60Co, 152Eu, 36Cl, and 63Ni at Hiroshima with the corresponding DS86-calculated free-field values. (Where appropriate, a small correction was made to the measured 60Co and 152Eu activation data to correct for shielding; see Appendix B.) To better illustrate the spatial dependence of the data at large distances, both measured and calculated values have been multiplied by the square of the slant range. Assuming that most of the neutrons originate from a point isotropic source at the epicenter, one would expect the thermal fluence, and thus thermal activation, to decrease approximately exponentially with distance, provided that the spectral distribution of neutrons in the epithermal and thermal region remained about the same. The DS86-calculated fluences show this near-exponential decrease, although the relaxation length increases slightly as the distance increases and the neutron spectrum becomes harder (Roesch 1987). The error bars reflect our best estimate of the total uncertainty (1 SD) and, as discussed in Appendix B, include possible errors not included by the investigators in their uncertainty estimates. Thus, our estimates of uncertainty are sometimes larger than those reported by the investigators. As discussed in Appendix B, the total uncertainty includes consideration of possible competing reactions and other sources of error.2 Also shown in Figures 3–1 through 3–4 is the best fit to the measurement data based on a weighted least- 2   Because we were unable to obtain sufficient information to estimate total uncertainty, the values listed in Table 3–2a and 3–2b are our best estimates and subject to revision. The actual uncertainty is still probably underestimated in as much as several issues discussed later in this chapter and in Appendix B might have contributed to additional errors that cannot be quantified now. As discussed in Appendix B, other reasonable weighting schemes other than that chosen could be used. However, any reasonable weighting scheme would still give little weight to the uncertain measurements at large slant ranges, so the difference between the fit to the measurements and the DS86 calculations would be similar to that shown.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) square fit, assuming an exponential decrease with distance (with varying relaxation length). The fitting procedure is discussed in detail in Appendix B. Weighting the data by the inverse of the variance provides a fit that is less influenced by poor-precision data and thus presumably more reflective of the actual spatial trend in the data. Because the 63Ni data are preliminary and subject to revision as additional measurements are completed, no uncertainty estimates are available, and the data have not been fitted. Similarly, the 36Cl data are being reevaluated because of various measurement and calculation issues discussed later in this chapter. Thus, the uncertainties shown reflect only the precision associated with measurements of several aliquots of the same sample. A cosmic-ray background activation value at the time of measurement has been subtracted from the 60Co and 36Cl measurement data before decaying to “at time of bombing” (ATB) as discussed below and in Appendix C. An appropriate cosmic-ray background for 63Ni is still under investigation, and no correction has been made to the data shown in Figure 3–4. As can be seen from the figures, notwithstanding the large uncertainties for samples from the more distant sites, the available data clearly indicate that a discrepancy that increases with increasing slant range exists in the DS86 calculations that cannot be explained by measurement error alone. There are only a few samples at very large slant ranges for 60Co, and they all deviate significantly from the best fit to the remaining data if the measurements at the Yokogawa Bridge (see Table 3–2a) are included in the fit. Note that the same investigator analyzed all these samples. They all have large uncertainty, and the uncertainty might be even larger than indicated because of the possible problems regarding cross-contamination and sample-selection bias discussed below. Because of their large uncertainties, these three points have very little influence on the shape of the fitted curve when the Yokogawa Bridge samples are included (see Appendix B). However, the Yokogawa Bridge measurements were shielded, and the effect of the shielding (about 50%) might have been overestimated. It is possible that if the distal data have much greater uncertainty than estimated for this report, the actual discrepancy lies somewhere between the two fitted curves shown in Figure 3–1. The fit to the 152Eu data (Figure 3–2) suggests a discrepancy similar to that indicated by the fits to the 60Co data. The ratio of measured to calculated, M/C, is about 5–10 at 1500 m. However, all the 152Eu data at the larger slant ranges are from two investigators, and the Shizuma data (indicated by squares in Figure 3–2) (Shizuma 2000a) appear to be consistently higher than data from Nakanishi and others (1983). As shown in Table 3–2a, all the Shizuma data at the larger slant ranges are less than our estimated MDC. However, all the data at great distances might be biased high (and the uncertainty underestimated) because of measurement issues described later in this chapter such as cross contamination of samples, quality control, background activity, cosmic-ray corrections, and other measurement issues. The more sensitive 36C1 and 63Ni measurements provide the strongest evidence for concluding that the discrepancy is less by a factor of only 5 or less at distances

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) TABLE 3–2a Hiroshima Neutron Line-of-Sight Measurements at Slant Range over 1000 m (Surface or Near-Surface Samples Except Indicated) Range, m Investigator (ref.#)a Site Name Specific Activity Estimated % S.D. invest. net activ ATB % S.D. Rev. DS86 Free field M/C S.D. stable mg Estimated MDC, Bq/mg msmt/ MDC Ground Slant ATB ATM ATM ATB 60Co (activity in Bq mg−1; estimated cosmic-ray activity=3.3×10−6) 1014 1168 Shizuma (106) Hiroshima City Hall 1.05×10–1 1.47×10–4 10 1.03×10–1 12 4.1×10–2 2.5 0.8 36.4 2.23×10–5 1.80×10–2 6.6 1197 1330 Hashizume (23) Powder Magazine 2.30×10–2 1.72×10–3 19 2.30×10–2 20 6.75×10–3b 3.4 1.2 9.0 1.49×10–4 2.07×10–3 11.5 1295 1419 Kerr (8) Yokogawa Bridge 5.60×10–3 1.96×10–5 9 4.66×10–3 12 2.2×10−3c 2.1 0.7 675.0 4.34×10–6 1.24×10–3 4.5 1295 1419 Hamada (29) Yokogawa Bridge 5.15×10–3 1.80×10–5 16 4.21×10–3 20 2.2×10–3c 1.9 0.7 239.0 1.14×10–5 3.26×10–3 1.6 1481 1591 Shizuma (106) Red Cross Hospital, pipe-1997 2.90×10–2 3.11×10–5 72 2.59×10–2 80 1.20×10–3 22 18 9.6 8.45×10–5 7.88×10–2 0.4 1481 1591 Shizuma (106) Red Cross Hospital, pipe-1995 1.50×10–2 2.09×10–5 27 1.26×10–2 34 1.20×10–3 11 5 56.4 1.44×10–5 1.03×10–2 1.5 1484 1593 Shizuma (106) Red Cross Hospital, ladder-1997 3.10×10–2 3.33×10–5 32 2.79×10–2 36 1.20×10–3 23 11 15.0 5.41×10–5 5.04×10–2 0.6 1484 1593 Shizuma (106) Red Cross Hospital, ladder-1995 3.40×10–2 4.74×10–5 29 3.16×10–2 33 1.20×10–3 26 12 26.3 3.09×10–5 2.21×10–2 1.5 1703 1799 Shizuma (106) Hiroshima Bank of Credit-1997 1.10×10–2 1.18×10–5 82 7.92×10–3 109 2.20×10–4 36 41 15.3 5.32×10–5 4.96×10–2 0.2

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 1703 1799 Shizuma (106) Hiroshima Bank of Credit-1995 2.10×10–2 2.93×10–5 14 1.86×10–2 18 2.20×10−4 85 30 73.0 1.11×10–5 7.96×10–3 2.6 4571 4608 Shizuma (106) Army Food Storehouse 0.00 0.00             2.2 3.70×10–4 2.65×10–1 0.0 152Eu (activity in Bq mg−1; estimated cosmic-ray activity=8×10−5) 849 1028 Shizuma (18) Choukaku-ji 2.4 2.28×10–1 17 2.40 20 1.45 1.7 0.6 0.036 1.16×10–1 1.220 2.0 880 1054 Nakanishi (11) Tamino’s House (Kawara-machi) 3.5 5.54×10–1 31 3.50 32 1.120 3.1 1.4 0.033 4.27×10–1 2.700 1.3 881 1055 Shizuma (18) Hiroshima Prefectural Office 2.2 2.09×10–1 9 2.20 12 1.110 2.0 0.6 0.069 6.08×10–2 6.40×10–1 3.4 893 1065 Shizuma (18) Honkei-ji 1.5 1.42×10–1 27 1.50 29 1.010 1.5 0.6 0.031 1.36×10–1 1.430 1.0 912 1081 Shizuma (18) Enryu-ji 2.4 2.28×10–1 63 2.40 64 8.70×10–1 2.8 2.0 0.027 1.58×10–1 1.660 1.4 924 1091 Shizuma (18) Yorozuyo Bridge stone wall 1.4 1.33×10–1 21 1.40 22 7.90×10–1 1.8 0.7 0.098 4.28×10–2 4.50×10–1 3.1 927 1093 Shizuma (18) Shingyo-ji 2 1.90×10–1 45 2.0 47 7.70×10–1 2.6 1.4 0.038 1.10×10–1 1.160 1.7 949 1112 Shizuma (18) Teramachi stone wall 1.7 1.61×10–1 53 1.70 54 6.46×10–1 2.6 1.6 0.046 9.07×10–2 9.56×10–1 1.8 988 1146 Shizuma (18) Hiroshima Radio Station 1.8 1.71×10–1 22 1.80 24 4.71×10–1 3.8 1.5 0.049 8.61×10–2 9.07×10–1 2.0 1017 1171 Shizuma (18) Hiroshima City Hall 1.1 1.04×10–1 27 1.10 29 3.73×10–1 2.9 1.2 0.043 9.65×10–2 1.020 1.1 1060 1208 Nakanishi (10) Hiroshima City Hall 1.15 1.21×10–1 13 1.150 15 2.65×10–1 4.3 1.5         1060 1208 Nakanishi (10) Hiroshima City Hall 1.02 1.07×10–1 50 1.020 51 2.65×10–1 3.8 2.3         1060 1208 Nakanishi (10) Hiroshima City Hall 1.15 1.21×10–1 13 1.150 16 2.65×10–1 4.3 1.5         1163 1300 Shizuma (18) Kozen-ji 1.1 1.04×10–1 45 1.10 47 1.15×10–1 9.6 5.3 0.022 1.90×10–1 2.000 0.5 1197 1330 Shizuma (18) Iwamiya-cho 0.9 8.54×10–2 50 8.99×10–1 52 8.80×10−2 10 6.1 0.034 1.25×10–1 1.310 0.7 1255 1383 Nakanishi (10) Hiroshima University 0.53 6.84×10–2 45 5.29×10–1 46 5.50×10–2 9.6 5.3         1255 1383 Nakanishi (10) Hiroshima University 0.18 2.32×10–2 72 1.79×10–1 73 5.50×10–2 3.3 2.6        

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) Range, m Investigator (ref.#)a Site Name Specific Activity Estimated % S.D. invest. net activ ATB % S.D. Rev. DS86 Free field M/C S.D. stable mg Estimated MDC, Bq/mg msmt/ MDC Ground Slant ATB ATM ATM ATB 1274 1400Nakanishi (11)Hiroshima University 0.41 4.31×10–2 44 4.09×10–1 45 4.80×10–2 8.5 4.6         1274 1400 Nakanishi (11) Hiroshima University 0.36 3.79×10–2 39 3.59×10–1 40 4.80×10–2 7.5 3.7         1274 1400 Nakanishi (11) Hiroshima University 0.48 5.05×10–2 50 4.79×10–1 51 4.80×10–2 10 5.9         1298 1422 Nakanishi (11) Hiroshima University 0.34 3.57×10–2 35 3.39×10−1 36 4.00×10–2 8.5 4.0         1328 1449 Nakanishi (11) Hiroshima University 0.114 1.20×10–2 32 1.13×10–1 34 3.23×10–2 3.5 1.6         1335 1456 Shizuma (18) Hiroshima University, Primary School 0.5 4.75×10–2 64 4.99×10–1 65 2.97×10–2 17 12 0.028 1.52×10–1 1.60 0.3 1357 1476 Shizuma (18) Kyo Bridge, Railing 0.55 5.22×10–2 71 5.49×10–1 72 2.48×10–2 22 17 0.047 9.00×10–2 9.48×10−1 0.6 1370 1488 Shizuma (18) Teishin Hospital (Communications Hospital) 0.51 4.84×10–2 43 5.09×10–1 44 2.25×10–2 23 12 0.053 7.98×10−2 8.41×10−1 0.6 36Cl (activity in 36Cl atoms/1015 Cl atom; estimated cosmic-ray background=108) 889 1045 Ruehm et al. (182) Sinkoji gravestone 2500 2.50×103 15 2392   2600 0.9 0.3         1029 1181 Ruehm et al. (182) Ganjioji-gravestone 400 4.00×102 10 292   680 0.4 0.1         1061 1209 Straume Hiroshima 700 7.00×102 17 592   474 1.2 0.4        

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86)   et al. (181) City Hall   1140 1279 Ruehm et al. (182) Tokneiji-gravestone 300 3.00×102 27 192 92 2.1 0.8 1217 1348 Ruehm et al. (182) Jyunkyoji-gravestone 240 2.40×102 16 132 150 0.9 0.3 1225 1355 Ruehm et al. (182) Hosenji-gravestone 260 2.60×102 29 152 140 1.1 0.5 1386 1502 Straume et al. (181) Teishin Hospital 300 3.00×102 53 192 39 4.9 3.0 1470 1580 Straume et al. (181) Red Cross Hosp. 212 2.12×102 36 104 20 5.2 2.4 1470 1580 Straume et al. (181) Red Cross Hosp. 325 3.25×102 59 217 20 11.0 7.0 1606 1708 Straume et al. (181) Postal Savings 198 1.98×102 39 90 8 11.0 5.4 1606 1708 Straume et al. (181) Postal Savings 185 1.85×102 47 77 8 10.0 5.6 63Ni (activity in atoms/micro gram Ni; no cosmic-ray background subtracted) 948 1111 Straume, Marchetti, Ruehm, et al. (181) Soy Sauce 0.51     0.60 0.50 1   1014 1168 '' City Hall 0.41     0.41 0.36 1   1304 1427 '' Univ. elem. School 0.17     0.17 0.055 3   1461 1572 " Univ. radioisotope bldg 0.15     0.15 0.028 5   a36Cl and 63Ni data are all preliminary and subject to change; see text. DS86 values are free-field except where indicated. bThe numbers in parentheses refer to Table 2 in Appendix 1. cShielding factor of 0.75 applied. dShielding factor of 0.51 applied.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) more than that for Hiroshima at large slant ranges in Figures 3–6 and 3–7. That probably reflects the different sources for the two cities. A much larger fraction of the thermal-neutron fluence (and thus activation) at Nagasaki was due to delayed neutrons from the fireball than at Hiroshima (about 33% vs. 8%) (Roesch 1987). Few 36Cl data have been reported at low activities in Nagasaki (see Table 3–2b). The plot of M/C at distances greater than 1000 m for Nagasaki (Figure 3–8) suggests that when uncertainty in both measurement and calculation is considered, the available reliable data do not support the existence of a discrepancy with distance. However, good data at large distances are sparse, and there are no good data at distances beyond a 1300-m slant range. Because the neutron fluence at Nagasaki was much lower than that at Hiroshima, activation at 1000–1200 m at Nagasaki corresponds to that at about 1050–1350 m at Hiroshima. If the lack of a discrepancy up to 1300 m at Nagasaki is confirmed by additional data, it would strongly suggest that the discrepancy at Hiroshima at low activities cannot be due only to measurement error and background subtraction errors in that these errors would also have been expected to occur at comparable activities at Nagasaki. Thus, it is very important to analyze additional samples from distances beyond 1000 m at Nagasaki. Because the source spectrum of neutrons at Nagasaki is very different from that FIGURE 3–8 Ratio of measured to calculated activation. Data from Table 3–2b. Error bars are 2SD. Data below MDC are not included.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) at Hiroshima and over one-third of the activation at large distances is calculated to be due to the delayed-neutron component, as opposed to only about 8% at Hiroshima, a similar discrepancy at Nagasaki might also indicate that at least some of the discrepancy is due to terrain or other effects, rather than source-term errors. The terrain at Nagasaki is irregular compared with Hiroshima, so one might expect increased scattering and possibly more degraded spectra at large distances (and correspondingly greater thermal activation) than calculated with DS86. MEASUREMENT ISSUES A more detailed discussion of measurement and uncertainty issues is in Appendix B, which is a rigorous evaluation by special scientists to the committee of the uncertainty in reported measurement data. The following summarizes some of the discussion in Appendix B. Cross Contamination of Samples Because the half-lives of 152Eu and 60Co are relatively short and it has been over 50 years since the events, the activation measurements reported in recent years, which include most of the data at larger slant ranges, are very low. The amounts of natural europium and cobalt in the samples are small, so enough must be extracted from the sample (that is the sample must be enriched sufficiently) to produce adequate measurement sensitivity. Many of the reported 152Eu and 60Co measurements at the greater distances are only slightly higher than or even less than the estimated MDCs (see Table 3–2). Thus, those data are highly suspect and should be used with caution. Because the lower-activity samples are so close to the MDCs, even very slight cross contamination of low-activity samples by previously prepared high-activity samples might have resulted in overestimation of the lowest-activity samples and underestimation of total uncertainty. As discussed below and in Appendix B, it appears that strict quality-control procedures to prevent cross contamination (use of blanks, blind sample analysis, and so on) were not generally followed; thus, one cannot discount the possibility that cross-contamination of samples occurred, particularly when the range of sample activity often varied over several orders of magnitude. For example, the range of 60Co activity in the samples prepared by Shizuma (1998) varied over a range of more than 1000. Even slight cross-contamination could have severely contaminated the lowest-activity samples. All the analyses of low-activity samples rely on chemical procedures to enrich the cobalt, europium, chlorine, and nickel and are thus subject to cross-contamination. Cobalt samples were prepared by using a milling machine to scrape chips from the surface of steel samples for chemical separation (Shizuma and others 1998). Tools were used to grind concrete and rock samples to prepare samples for 36Cl analysis. Although the committee has no direct knowledge that any such contamination occurred, the apparent flattening out of the measured activities for europium

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) and cobalt at great distances and the fact that some samples were apparently obtained at similar distances that did not show measurable activity (see Appendix B) suggest that the possibility exists and should be explored further. The 152Eu and 60Co data at great distances in particular are also suspect because of possible selection bias in deciding which samples were actually exposed to bomb neutrons. There is some evidence that some samples were assumed unexposed if spectrum photo peaks were not clearly evident and thus data that might have reduced the average activity at distal sites was not reported (see Appendix B). Quality Control It appears that few investigators performed rigorous quality control or participated in a quality-assessment program. Provenance of samples is not always well documented, and the mislabeling of samples is always a potential problem. Data-quality assessment has been defined by the US Environmental Protection Agency (EPA) as a procedure by which existing data can be used to make a value judgment or decision (US EPA 1994a; US EPA 1994b). The US Department of Energy has established a similar approach for environmental data based on the guidelines published by EPA (Tindal 2000). The procedure is used to establish whether a data set is adequate for making some decision or estimate. It can be used to answer two fundamental questions: Can a decision or estimate be made with a desired level of confidence, given the quality of the data set? How well can the data set be expected to perform over a wide range of possible outcomes? Assessment of environmental data must unambiguously establish the reliability of the data. Furthermore, monitoring data must be confirmed before application of statistical evaluation or other interpretations. Uncertainties arise from sampling procedures (sampling variance) and analytical procedures (analytical variance). The process of systematic and independent verification of data and the associated uncertainties or variances is data-quality assessment (Miller and Fitzgerald 1991). In the case of environmental measurements to ascertain the neutron yield of the Hiroshima and Nagasaki bombs, sampling variance is minimized by considering only samples whose exact locations at the time of the bombs are well established. Sample shielding must also be known. Sample variance also depends on sample size (Kratochvil and others 1984) and concentration of the analyte in the sample (Boyer and others 1985). For some measurements, analyte concentrations are quite small, and large quantities of environmental materials are needed. In other cases, heterogeneity in the distribution of the analyte in the sample dictates that a specific sample size be used. Those kinds of variance can be assessed with methods described by Clark and others (1996).

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) Analytical variance reflects a combination of systematic and random errors associated with measurement. Systematic errors, which can be detected and corrected, can be reduced by various means, including analysis of standard reference materials, analysis of blank samples containing no analyte, use of a different method to obtain the same result, and interlaboratory comparisons. In interlaboratory comparisons, several laboratories analyze identical samples with the same or different methods. Ideally, analytical variance is accurately reflected in the estimate of uncertainty reported with an analytical result. This variance reflects the total uncertainty arising from chemical and physical manipulations of the sample during preparation for analysis and from the established uncertainty of the analytical method used. It is always desirable to test any analytical result obtained with one method by repeating the measurement with a different method. Good agreement between measurements made with multiple methods affords confidence in the result and is useful in establishing the uncertainty in the measured value. Apparently, only a few intercomparisons were carried out to test counting accuracy; to our knowledge, no intercomparisons were carried out to test sample preparation, and so on. Preparation and analysis of duplicate aliquots were rare, as was splitting of samples. Exchange of prepared samples between laboratories doing the same types of analyses (such as germanium gamma-spectrometric counting) to test counting accuracy was apparently not routine. That is unfortunate, because the MDCs for a given type of analysis varied considerably from investigator to investigator, and an intercomparison of low-activity samples might have identified a systematic bias. For the 152Eu samples, at least, the half-life is long enough so that available samples could still be shared and counted in facilities with lower background. It would also be desirable to analyze some of the same concrete samples for both 36Cl and 152Eu at the more distant sites. Background Activity An important potential source of error in the reported activity at the time of bombing for sites at very large slant ranges is the correction for environmental background activity. All investigators corrected for background counting errors from their detector system due to radiation, but some of the corrections can be highly uncertain, particularly at low activities. Most investigators did not correct for activation of the samples in situ by neutrons produced by cosmic-ray secondaries in the atmosphere. Appendix C discusses cosmic-ray activation in some detail. Shizuma (1999) attempted to calculate the contribution of cosmic rays or measure the activation in laboratory reagents. He concluded that the contribution was negligible. For the case of 60Co, he apparently based that conclusion on a comparison with his measured 60Co activation in reagents, which is somewhat uncertain and only about one-fifth of the activation measured by another investigator (see Appendix C). It is important to note that the estimated cosmic-ray activation must

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) be compared with the measured activation, not with the activation at the time of the bombing after correction for decay. Because 60Co activation measured in the 1990s has decayed by as many as 10 half-lives, the possible cosmic-ray activation contribution at the time of measurement is not a negligible fraction of the measured activity in samples obtained at great distances from the epicenter and could account for about 20–30%, or more, of the reported activity at the time of the bombing at the most distant locations, as shown in Table 3–2. Cosmic-ray Corrections The cosmic-ray corrections shown in Table 3–2 are our current best estimates based on measurements in laboratory reagents and calculations (see Appendix B and Appendix C). The cosmic-ray activation of samples in situ could be higher or lower than that in laboratory reagents that were presumably stored in a building and thus could have been substantially shielded from the full cosmic-ray fluence, so the actual 60Co background for some samples might have been even larger than our crude estimate. The 152Eu cosmic-ray activity based on the reagent measurements and calculations appears to be too small to account for any significant error in the measurements, particularly because the decay correction due to the elapsed time between the bombing and sample measurement is only up to about a factor of 10 compared with about a factor of 1000 for 60Co. Owing to the short half-lives of 60Co and 152Eu, all the cosmic-ray activation took place in situ after the bombing and saturation (production rate=decay rate) should have been essentially achieved.3 Measurements by Straume (2000b) of the attenuation in deep concrete and at sites far from the epicenter indicate that the cosmic-ray contribution is significant for measurements beyond about 1200 m—36Cl/Cl of around 100×10–15. The cosmicray activation in concrete can be expected to vary somewhat depending on the history of the material (sand, and so on) used to make the concrete, so the estimated background subtraction for 36Cl might not be the same for all samples.4 Cosmicray 36Cl activities from less than 100×10−15 to 600×10−15 have been reported for sands and rocks (see Appendix C). As discussed in Appendix C, because of the lower geomagnetic latitude of Japan, a cosmic-ray fluence of about one-half to three-fourths that at higher latitudes is likely, therefore the value estimated from 3     An additional consideration with respect to 152Eu in roof and wall tiles is whether the cosmic-ray activation of tiles that were blown off buildings and thus exposed to cosmic rays while lying on the ground or elsewhere was greater than would have occurred if the tiles remained in place. 4   Note that because of the long half-life of 36Cl (300,000 y), all the activation took place over many thousands of years before the production of the concrete in the sample. For much of that time, the sand and other material used to make the concrete was probably shielded to some degree from cosmic rays. The fact that the estimated activities are only a few percent of that expected if complete saturation (production rate=decay rate) of a surface source had occurred substantiates this.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) the core measurements is reasonably consistent with the reported data. However, values 2–3 times higher than estimated from the core measurements, depending on the history of the sample materials, would also not be inconsistent with the reported data. In Table 3–2a and 3–2b, we have used the mean of 108×10−15 for Hiroshima and 122×10−15 for Nagasaki with an estimated coefficient of variation (CV) of 25%. Because it is clear that cosmic-ray activation could have been an important contributor to the measured 60Co and 36Cl activity for samples at great distances and because the actual cosmic-ray contribution depends on the location and exposure conditions, it is important to obtain good measurements of cosmic-ray activation for samples similar to those analyzed (that is steel plates or lightning rods for 60Co and rocks, tiles, and concrete for 36Cl) far enough from the epicenter for bomb-fluence effects to be neglected so that the cosmic-ray exposure conditions can be considered comparable. As discussed in Appendix C, cosmic-ray activation in thick slabs of material first increases because of spallation of the much higher incident-energy spectrum, which produces a shower of evaporation neutrons, and then decreases exponentially with depth, so estimating the cosmic component by measuring the variation in activity in cores, as was done for 36Cl, would not be an appropriate substitute for measurements of surface samples at great distances. Other Measurement Issues Several other measurement issues considered in Appendix B might increase the uncertainty reported by various investigators. The location of some samples was not always documented. Such samples as roof tiles were blown off buildings and later recovered, so their location at the time of the bombing is certain only to within several tens of meters. On the basis of a calculated DS86 free-field relaxation length of about 126 m from thermal activation at slant ranges of 500–1000 m, an error of 10 m in the slant range would result in an error of about 8% in the calculated activation. In many cases, the surface of the buildings from which concrete cores were obtained for 36Cl analysis apparently were subjected to repairs, so the depth at the time of bombing corresponding to a particular measured value is quite uncertain. Some surface samples, in particular those for 36Cl analysis, might also have been contaminated because of dilution of the surface 36Cl activity by chlorine in rainwater or enhancement of the signal due to repairs made with contaminated cement. Furthermore, 36Cl can be produced by activation of 39K, and this could have resulted in an additional error in the 36Cl/total chlorine estimates for samples with high potassium and relatively low chlorine content (Straume 2000b). The AMS technique used to measure 36Cl results in a higher sensitivity for 36Cl activation than for europium and cobalt. However, a detailed uncertainty analysis has not yet been reported for these data, and replicate measurements of the same sample indicate fairly poor measurement precision even at high activity (Straume 2000b). The problems of surface contamination and potassium activation dis-

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) cussed above are also of concern. Furthermore, thermal activation in concrete is very sensitive to water content and elemental composition of the concrete and surrounding medium. As discussed below, that might result in a substantial difference between the DS86 free-field calculation and the actual activation in a sample core from a building or bridge. Similarly, the activation in samples taken from granite gravestones might not be adequately reflected by the DS86 free-field calculations at ground level. Thus, the comparisons between the 36Cl surface activity and DS86-calculated values are more uncertain than previously reported and are being reevaluated by the various authors. The 63Ni data, also from AMS measurement, have relatively poor sensitivity at great distances, particularly when the amount of stable nickel in the copper sample is high. If the concentration of nickel in the sample is relatively high, the 63Ni created by thermal activation can also be high. Before the 63Ni data can be considered reliable, there must be a complete uncertainty analysis that accounts for potential errors due to the chemical separation of nickel from copper, AMS measurement, thermal-neutron activation of nickel, and uncertainty in total nickel content of the sample. UNCERTAINTY IN CALCULATED ACTIVATION It is important to note that the calculation of activation even in a near-surface line-of-sight sample using free-field DS86 fluences is somewhat uncertain. That would be true even if the calculated DS86 fluence in air were highly accurate. Lillie and others (1988) estimated the uncertainty in the free-field fluence at about 20% (1 SD). The calculated activations shown in Figures 3–1 through 3–8 are based on free-field values 1 m above the ground. Sample-specific calculations were carried out for only a few samples—generally samples that required a substantial shielding correction. The energy spectrum traversing the sample was assumed to be the same as that in the air. The actual spectrum—particularly the energy spectrum for the neutron energies below about 0.1 keV—traversing the sample, even near the surface and due to backscattering, depends heavily on the sample material, water content, sample geometry, and depth. The slowing down and the resulting spectral distribution of these neutrons and the thermal neutrons they produce in the sample account for most of the observed activation even in near-surface samples (Roesch 1987). The incident thermal neutrons account for only a small fraction of the activation. For most samples, the effect of sample depth, orientation, composition etc. (in particular water and boron content), was estimated on the basis of benchmark calculations. Kaul and Egbert (2000) presented the results of some benchmark calculations illustrating the sensitivity of the calculated fluence to building height, sample orientation, depth of sample, and so on. Straume and others (1994) also discussed the results of benchmark calculations and the resulting uncertainty in the activation calculations. Although all the estimated corrections from free-field to surface samples are relatively small (around tens of percent at most) and none of

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) these effects alone can explain the large apparent discrepancy between measured and calculated values at great distances, the total uncertainty in the calculated activation due to all perturbations might be large enough (50–100%) to obscure the magnitude of the possible discrepancy in the DS86 free-field air fluence, especially at greater distances. Furthermore, the true calculation of the activation in a sample depends on a folding together of the activation cross section with the energy spectrum of neutrons traversing the sample. In DS86, only a few energy groups were used to describe this energy distribution. In fact, DS86 contains only one thermal group consisting of all neutrons below 0.4 eV, although more extensive calculations have since been made (see Chapter 4). The activation by thermal neutrons is thus estimated by assuming the spectral shape in this bin to be a Maxwellian distribution with average energy corresponding to a temperature of 300°K and calculating a weighted thermal cross section (Kaul and Egbert 1989). Because the epithermal cross sections have substantial structure, particularly for 60Co, the calculations must assume a spectral shape in the epithermal energy bands (generally 1/E if an appropriately weighted cross section is to be used). For a sample imbedded even slightly, the shape of the spectrum in the epithermal and thermal region will depend on the material temperature and depth. The thermal activation will come primarily from a slowing of incident epithermal neutrons in the sample.5 For the case of 60Co, about 25% of the activation is due to neutrons above thermal energy from resonances in the cross section (see Table 3–1). Thus, the uncertainty in the calculated activation is sensitive to the actual energy distribution of the fluence in the first few centimeters of the sample and to the change in this distribution from that calculated for the free field because of slowing in the sample itself. The DS86 neutron fluence varies substantially with height above the ground (over 40% higher thermal fluence at 1 m than 25 m at 1500 m). That reflects the greater thermalization and backscattering of neutrons by soil than by air as shown in Table 3–3. The thermal fluence near the ground is larger than at 25 m, but the epithermal fluence is lower (Cullings 2000). The calculated activation assumes that the thermal and epithermal fluences near the surface of a structure are similar to those 1 m above the ground, on the basis of limited benchmark calculations. However, variations of around 20–30% can easily occur, depending on the location of the sample and its composition (water and boron content in particular). At the more distant slant ranges, where the line-of-sight angle is fairly small, it is also possible that the low-energy fluence is greatly underestimated by DS86 because of increased scattering 5   Many of the investigators reporting on thermal-neutron activation measurements in cobalt or europium estimated the DS86 (calculated) activation by multiplying the published (DS86) fluence by the thermal-activation cross section at 300°K shown in Table 3–1 rather than by a Maxwellian-weighted average as SAIC apparently does (Kaul and Egbert 1989). The difference is about 20–30% for thermal activation.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) TABLE 3–3 DS86-calculated Neutron Fluence (n/cm2) at 1 m Versus 25 m (Cullings 2000)   Thermal Neutrons   Epithermal Neutronsa   Ground Range Height 1 m Height 25 m Ratio Height 1 m Height 25 m Ratio 1000 m 2.6×1010 2.0×1010 1.3 3.8×109 5.4×109 0.70 1500 m 5.7×108 4.1×108 1.4 8.6×107 1.2×108 0.72 a29–100 eV energy bin. of neutrons by terrain features and structures. Finally, the activation cross sections themselves have uncertainty, and some, such as the 63Ni cross sections, are very uncertain (Egbert 1999). Thus, calculation uncertainty might account for some of the scatter in the calculated M/C ratios of samples collected at about the same distance as shown in Figures 3–5 and 3–8 and some of the observed discrepancy. Furthermore, previous comparisons of M/C ratios as a function of slant range, using a nonweighted regression that failed to reduce the influence of the higher uncertainty in more-distant measurement data and to include the possible calculation uncertainty, might have unduly emphasized the distant data and led to an overestimate of the discrepancy in the DS86 neutron fluence at large slant ranges. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS A number of measurement issues might explain in part why the thermalactivation measurements reported for large slant ranges are too high. They include possible cross-contamination, sample-selection bias, giving too much weight to data below the MDC, and inadequate background subtraction. However, all the Hiroshima activation measurements still show a consistent pattern of discrepancy, with measured values exceeding calculated values at greater distances, but lower than calculated values near the epicenter. When corrections are made to account for cosmic-ray activation and samples with very high uncertainty (near or less than the MDC) are disregarded or given less weight, the discrepancy appears to be somewhat smaller than previously reported. The preliminary 63Ni data suggest that the M/C ratio at 1500 m might be only around 3–5 for higher-energy neutrons. The 152Eu data, although of questionable accuracy, still tend to support a larger discrepancy than the 60Co activation data, so it would be useful if some of the available samples could be reanalyzed by multiple laboratories in an intercomparison exercise. That might indicate possible measurement bias at low activities and explain the larger apparent bias. Participation in this exercise with extremely low-background counting rooms would be very helpful. Such intercomparisons would require the cooperation of the owners of the processed and previously counted samples.

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) Some of the discrepancy (about 30% higher calculated neutron kerma at 1300 m) is already known to be due to the DS86 method on the basis of changes in cross sections and improvements in transport methods discussed in Chapter 4. Some additional discrepancy may be explained by errors in the calculation of activation. However, on the basis of the discussion earlier in this chapter, calculation uncertainty can probably account for no more than about a factor of 2 of the M/C discrepancy for any particular sample. Neutron fluences at Nagasaki at 1000–1300 m correspond to fluences at Hiroshima at about 1200–1500 m. Thus, the apparent good agreement at Nagasaki at distances up to about 1300 m implies that the discrepancy at Hiroshima cannot be due primarily to measurement errors or an error in the cosmic-ray background subtraction. However, there are few data at distances beyond 1000 m at Nagasaki. It is therefore important to obtain additional data at Nagasaki at slant ranges of 1000–1400 m to confirm the agreement with DS86 calculations. The nearly exponential decrease in both calculated and measured scaled activation (see Figures 3–1 through 3–4) at Hiroshima suggests that the remaining discrepancy at large distances can be explained by a slightly harder source spectrum that would allow more higher-energy neutrons to penetrate to greater distances and thus produce a larger local thermal and epithermal fluence at that distance. However, a higher proportion of higher-energy neutrons from either the bomb or the fireball would make the current agreement with the 32S measurements worse to a greater extent than it would improve the agreement at large distances (Kaul and others 1994). The committee offers the following recommendations: The highest priority should be given to making additional measurements of 63Ni at sites near the Hiroshima epicenter to compare with the 32S measurements and to making as many measurements as possible at distances greater than 1200 m. Additional 36Cl measurements should be obtained. Because of concerns about the reliability of 36Cl measurements from concrete, priority should be given to measurements in granite unless investigators can provide a protocol for measurements in concrete that address the issue of reliability. Because the 36Cl data are the only thermal-activation data with sufficient sensitivity to provide reliable results at large distances, it is important to resolve the surface-contamination and potassium-activation issues for 36Cl and to obtain more reliable estimates of the uncertainty in these data, particularly at large distances. Measurements of 63Ni at Nagasaki are needed to determine whether a similar but smaller discrepancy exists at activities comparable with those corresponding to Hiroshima slant ranges of 1200–1500 m. Emphasis should be on measuring line-of-sight minimally shielded samples. Additional 36Cl measurements are also desirable at these distances in Nagasaki. Because the actual cosmic-ray activities cannot be accurately calculated, owing to the dependence on location and local scattering and attenuation, it is im-

OCR for page 31
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) portant that background measurements be made on samples similar to those already analyzed but collected further away at locations and in positions similar to those of the actual bomb samples. That is particularly important for 63Ni, for which no cosmicray background data are available,6 and would be highly desirable for 152Eu. Shibata (2000) has demonstrated the capability of measuring at least the close-in copper samples for 63Ni with direct liquid-scintillation beta counting. Duplicate aliquots of samples collected close to the epicenter should be prepared and analyzed with both AMS and beta counting techniques. The possibility that some of the reported activity in the samples collected at large slant ranges is a result of cross-contamination during sample preparation and chemical-enrichment or measurement-selection bias should be further investigated. Investigators should be asked to document that procedures used to ensure that such cross-contamination did not occur and to see whether blanks and quality-control samples were a part of every batch sample preparation and analysis. All new measurement programs should include a quality-assurance component with established data-quality objectives and procedures to provide assurance that cross-contamination problems will be identified and eliminated. For the committee to provide its best assessment of the most critical data sets, those at sites greater than 1000 m, it is essential that the investigators that have reported activation measurements be encouraged to provide the necessary information and agree to cooperate in sharing samples and participating in intercomparisons. Substantial environmental data on isotopes produced by neutron activation in the Hiroshima and Nagasaki bombs exist and are documented in the RERF database (Appendix B, Table B-1). Whether those data will be useful in resolving the apparent discrepancy between measured and calculated neutron fluences depends on the completion of a data-quality assessment for this database. The working group should establish data-quality objectives and a data-quality assessment procedure to evaluate the existing data, with procedures described by EPA (US EPA 1994a) and DOE (Tindal 2000). This activity must be a joint and integrated effort involving both US and Japanese researchers and involving researchers who make the measurements and theoreticians who estimate neutron fluences. The uncertainty in the DS86 activation calculations has not been thoroughly investigated. A thorough uncertainty assessment of DS86 or its successor should include an assessment of the uncertainty in these calculations. The committee further recommends that any new dosimetry system utilize the best current technology and cross sections in calculating the neutron activation of samples. Furthermore, when measurements and calculations are compared, the method used for the calculations should be clearly specified. 6   Although calculations indicate that the cosmic-ray activation in copper is likely to be small relative to the measured signal (Ruehm and others 2000a) even at large distances, the cosmic-ray fluence might be increased considerably in high-Z materials because of spallation of incident very-high-energy neutrons, and so the calculated fluence based on cosmic-ray spectra in air might be too low.