A calibration neutron monitor: Energy response and instrumental temperature sensitivity

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113,, doi: /2008ja013229, 2008 A calibration neutron monitor: Energy response and instrumental temperature sensitivity H. Krüger, 1 H. Moraal, 1 J. W. Bieber, 2 J. M. Clem, 2 P. A. Evenson, 2 K. R. Pyle, 2 M. L. Duldig, 3 and J. E. Humble 4 Received 31 March 2008; accepted 1 May 2008; published 2 August [1] Neutron monitors are integral detectors of secondary cosmic rays. Since each of them has its own detection efficiency, energy spectra cannot readily be derived from their observations. To circumvent this problem, latitudinal surveys have been conducted for many years with mobile neutron monitors to derive such spectra. Another way, however, is to use the worldwide stationary neutron monitor network, but then the counting rates of these monitors must be normalized sufficiently accurately against one another. For this reason, two portable calibration neutron monitors were built at the Potchefstroom campus of the North-West University, South Africa. When calibrations of an adequate number of the worldwide neutron monitors have been done, this paper demonstrates that the differential response functions derived from them will provide experimental data for modulation studies in the rigidity range of 1 to 15 GV. Several properties of these calibrators were investigated, in order to achieve sufficient calibration accuracy. The energy response over the cutoff rigidity range from the poles to the equator, as well as the instrumental temperature sensitivity, are described in this paper. The main conclusion is that the calibrator has a difference of almost 4% in its energy response over the cutoff rigidity range 0 16 GV in comparison with a standard 3NM64 neutron monitor. Furthermore, it is shown that not only the calibrator, but also the NM64 and IGY monitors, have fairly large instrumental temperature sensitivities. Correction coefficients for these effects are given. Citation: Krüger, H., H. Moraal, J. W. Bieber, J. M. Clem, P. A. Evenson, K. R. Pyle, M. L. Duldig, and J. E. Humble (2008), A calibration neutron monitor: Energy response and instrumental temperature sensitivity, J. Geophys. Res., 113,, doi: /2008ja Introduction [2] Cosmic-ray modulation studies in the heliosphere are primarily done by observations in space, but they continue to be supported by a network of 50 neutron monitors on the surface of the Earth. [3] In the atmosphere, the primary cosmic rays interact with atmospheric nuclei to form a cascade of secondary particles. The worldwide neutron monitor network records these secondary cosmic rays, mainly the neutrons, with energies about a decade higher than detected by most spacecraft. 1 School of Physics, North-West University, Potchefstroom, South Africa. 2 Bartol Research Institute and Department of Physics and Astronomy, University of Delaware, Newark, Delaware, USA. 3 Australian Government Antarctic Division, Kingston, Tasmania, Australia. 4 School of Mathematics and Physics, University of Tasmania, Hobart, Tasmania, Australia. Copyright 2008 by the American Geophysical Union /08/2008JA [4] The counting rate, N, of a neutron monitor at cutoff rigidity, P c, is given by Z 1 dn NP> ð P c Þ ¼ P c dp dp; where dn/dp is called the differential response function, which is related to the primary cosmic ray intensity, j(p), above the atmosphere by ð1þ dn dp ¼ SP; ð x ÞjðP; tþ; ð2þ where S is the yield function for secondary cosmic rays at atmospheric depth x. Details pertaining to the functioning of neutron monitors can be found in Hatton and Carmichael [1964], Hatton [1971], Clem and Dorman [2000], Moraal et al. [2000], Simpson [2000], Stoker et al. [2000], and Krüger [2006]. [5] Since neutron monitors are integral detectors, each with its own design, detection efficiency and local environment, energy spectra cannot readily be derived from their observations. One way to circumvent this is by conducting 1of6

2 Figure 1. A typical response function for 11 intercalibrated neutron monitors as function of cutoff rigidity. It is based on the Dorman et al. [1970] function N = N 0 [1 e ap c k ], with N 0 = 1.0, a = 10.0, and k = latitudinal surveys with mobile neutron monitors. The drawback is that these surveys do not take place on a continuous and regular basis. Another way to derive such spectra would be to use the worldwide stationary neutron monitor network. In principle, differential response functions can be calculated from dn dp NP ð c 2 Þ NP ð c1 Þ ; ð3þ P c2 P c1 where P c1 and P c2 are reasonably near cutoff rigidities. The only condition is that the counting rates of these monitors have to be normalized sufficiently accurate against one another. [6] Moraal et al. [2000] showed an example of such an intercalibration, by calculating a typical differential response function (using the so-called Dorman function) as function of cutoff rigidity, for eleven intercalibrated neutron monitors. This is represented in Figure 1. The cutoff rigidity intervals increase in steps of 30%. The vertical error bars represent a 0.2% error in the counting rate of each individual neutron monitor. The size of these error bars show that in order to derive meaningful spectra from neutron monitor counts, the calibration accuracy will have to be 0.2%. [7] In order to execute such intercalibrations, two identical portable calibration neutron monitors were built at the Potchefstroom campus of North-West University, South Africa, and completed in Figure 2 shows one of the calibrators on its cradle. It has a length of 753 mm, i.e. about 1/3 of the length of a standard NM64 neutron monitor. The counter is filled with 3 He at a pressure of 4 atmospheres. Its mass with the cradle and wheels is 223 kg. Six people can carry it as a unit without the need to dismantle it (which would inevitably affect its efficiency). Thus, it can be brought as a unit to neutron monitors in the worldwide network for intercalibration. It is even possible to put it inside most neutron monitor huts. This calibration neutron monitor is described in greater detail by Krüger [2006], and design specifications are available on fakulteite/natuur/fisika/navorsing/kalibrasie_n_m.html. Preliminary work on this experiment was published in Moraal et al. [2001, 2003] and Krüger et al. [2003, 2005]. [8] Several properties of the calibrator were investigated in order to achieve the 0.2% accuracy. Firstly, its energy response over the cutoff rigidity interval from the poles to the equator was investigated, with the result that it was found to be almost 4% larger than that of a standard 3NM64 neutron monitor. Secondly, it was also determined that not only the calibrator, but also the stationary NM64 and IGY neutron monitors, have fairly large instrumental temperature sensitivity, which must be accounted for in calibration procedures. Thirdly, the calibrator shows a large sensitivity to the surface beneath it. These three effects were investigated by Krüger [2006]. In this paper we summarize the energy (latitudinal) dependence and the temperature sensitivity, while we will report on the surface effects in a subsequent paper. 2. Energy Response 2.1. Experimental Measurement of the Energy Response of the Calibrator [9] Neutron monitors of different design have different responses to the energy dependence of primary intensity variations, as well as different atmospheric (pressure and temperature) responses. Thus, to achieve a calibration accuracy of 0.2%, any difference in energy response between the calibrator and the standard NM64 type stationary neutron monitors must be known to within this accuracy. Figure 2. The calibration neutron monitor (see puk.ac.za/fakulteite/natuur/fisika/navorsing/kalibrasie_ n_m.html). 2of6

3 Table 1. Latitudinal Surveys With the Calibrator Year Vessel Departure Seattle Crossing Equator Arrival McMurdo Departure McMurdo Crossing Equator Arrival Seattle 2002/03 Polar Sea 4 Nov 02 day 308 day Jan 03 day 369 day 428 day Apr 03 day /04 Polar Sea 17 Nov 03 day 321 day Dec 03 day 365 day 401 day Mar 04 day /05 Polar Star 5 Nov 04 day 310 day Dec 04 day 365 day 406 day Mar 05 day Polar Star day 47 day Mar 06 day /07 Polar Sea 18 Nov 06 day 322 day Jan 07 day 374 day 410 day Apr 07 day 465 Cutoff rigidity 1.7 GV 15.5 GV GV GV 15.5 GV 1.7 GV Such differences can be expected due to differences in design, such as the relative masses and dimensions of the polyethylene moderators and reflectors, and the lead producers, of the two neutron monitors. An obvious way to measure this difference is to compare the differential response function, dn/dp, of the calibrator with that of a standard neutron monitor during a latitudinal survey. [10] Neutron monitor latitudinal surveys have been conducted annually since 1994 by the Bartol Research Institute, in collaboration with the Australian Government Antarctic Division and the University of Tasmania, from Seattle, USA, to McMurdo, Antarctica, and back, over a 5 6 month period. They use a standard 3NM64 neutron monitor aboard one of two US Coast Guard ice breakers, the vessels Polar Sea and Polar Star. These surveys cover cutoff rigidities from <0.1 GV at McMurdo to 15 GV in the mid-pacific. Details of these annual surveys were described by Bieber et al. [1997, 2001, 2003]. One of the two Potchefstroom calibrators was sent with this 3NM64 on five of these voyages. [11] The results of the first latitudinal survey of the calibrator in 2002/03 were described by Krüger et al. [2003], which revealed that the calibrator indeed has a different energy response than that of a standard 3NM64 neutron monitor, such that its cutoff rigidity response is almost 4% larger over the whole latitudinal range than that of the 3NM64. A second survey in 2003/04 confirmed this result. Due to the low counting rate of the calibrator and the resulting relatively large statistical errors of the counting ratio as a function of cutoff rigidity, it was decided to send the calibrator on a third voyage. However, due to a problem with the pre-amplifier during this voyage in 2004/05, the results are unreliable. This attempt was repeated in the 2005/06 season, but for logistical reasons data could only be recorded on the return leg. The final survey was conducted in the 2006/07 season. These voyages are summarized in Table 1. [12] The hourly counting rates of the two monitors for the five voyages as a function of time are shown in Figure 3. The upper curves show the counts of the 3NM64 (divided by 10), while the bottom curves show those of the calibrator. We note that the 3NM64 counting rate is approximately 28 times higher than that of the calibrator. The lowest sections of each trace represent equatorial crossings, which occurred on different dates on different voyages (see Table 1). No corrections were made for pressure variations or anything else, but data points with obvious electronic pick-up were eliminated. Note that no data are available for the calibrator for the extended period from day 366 to day 413 in 2003/04, when the electronics head had to be removed to replace the hard disk. [13] The cutoff rigidities at the position of the ship were determined with the standard code obtained from K. G. McCracken (private communication, 2005), using the 1995 geomagnetic field representation. [14] The ratio of the calibrator to 3NM64 counting rate was calculated for each hour. These ratios were then binned into cutoff rigidity intervals of 1 GV each, ranging from 0 to 16 GV, as shown by the data points in Figure 4. The error bars are based on the statistical fluctuations expected from the total number of counts in that interval. For this reason the point at 0 GV is very accurate, because of the long time spent in the vicinity of McMurdo (see Figure 3). For the same reason the scatter in the points and the large error bars in the 6 GV P c 10 GV interval, are due to the shorter periods spent by the ship in each latitude zone as it transited between the polar and the equatorial regions. [15] The middle solid line in Figure 4 represents a linear regression of (8.4 ± 0.7) 10 5 /GV, where the error is one standard deviation in the value of the normalized counting rate. This gives a fractional change in the slope of (0.235 ± 0.018) %/GV. The error on this value implies that over the ridigity range of 0 15 GV the upper limit on the uncertainties is approximately ± = 0.27%, which surpasses the desired accuracy of 0.2% for corresponding smaller separation of the neutron monitors. However, it will Figure 3. Average hourly counting rates of the calibrator and 3NM64 for five latitudinal surveys. The horizontal axis starts on day 290 of each year, and continues past day 365 into the first part of the next year. Notice that the 2004/05 calibrator counts are unreliable and were not used in the subsequent analysis. 3of6

4 be a constant number throughout all calculations, and will therefore have little effect on spectral variations Simulations [16] One of us (J. M. Clem) calculated the calibrator/ NM64 counting ratio as a function of cutoff rigidity by using a Monte Carlo simulation, the FLUKA particle transport code of Clem [1999], together with programs to simulate the proportional counter and electronics response. The latest version of FLUKA, which now includes its own heavy ion package, was used. The response for all the incident particles was taken into account. The calibrator was simulated to be above a 10 cm thick slab of steel without any roof. Standard dimensions and compositions for the geometry of an NM64 were used. [17] This calculated ratio of the counting rates as a function of cutoff rigidity is also shown in Figure 4 by the dashed lines. The simulation was done for a differential response function at both solar minimum and maximum conditions. As expected, the solar minimum response is noticeably steeper, due to the fact that more low-energy particles are included. [18] These simulated ratios vary with 1.45% at solar minimum, and 1.82% at solar maximum, over the rigidity range of 0 to 15 GV. This is equivalent to 0.10 to 0.12 %/GV, respectively. These values are less than half of the measured ones Summary and Conclusions on Energy Response [19] The latitudinal surveys have shown that there is a significant difference in the energy/latitudinal response of the calibrator relative to a standard 3NM64. In fact, the overall change of 3.8% from pole to equator is 20 times larger than the ±0.2% calibration accuracy needed for realistic spectral studies. The statistical accuracy at present is such that the ratio of (0.235 ± 0.018) %/GV will lead to a calibration error of 0.27% between a polar and equatorial neutron monitor. This is, however, a constant error, and therefore will have little effect on spectral variations. It is also clear that numerical simulations of neutron monitor response are not yet sufficiently accurate to be useful in these studies. Figure 4. Counting ratios of the calibrator and 3NM64 for the years 2002 to 2007 as a function of cutoff rigidity. Observations, binned into 1 GV cutoff rigidity intervals, are shown by the dots. The continuous lines represent a linear regression with a 68% confidence band, while the dashed lines are simulated values as described in the text. 3. Temperature Sensitivity [20] The second calibration monitor was taken to Sanae, Antarctica, between 19 December 2002 and 2 February 2003, in a first attempt to calibrate the Sanae monitor. In the course of this experiment, however, an unexpected large instrumental temperature sensitivity was discovered, as reported by Moraal et al. [2003]. This temperature effect, with a magnitude of 0.126%/ C, is very large in relation to the overall calibration error of 0.2% that must be attained. This necessitated a detailed investigation over and above these preliminary results. [21] Quite coincidentally, Evenson et al. [2005] discovered the same temperature effect on their Nain NM64 neutron monitor, due to thermostat runaways. [22] This instrumental temperature sensitivity is different from the atmospheric temperature effect, described by Hatton [1971] and Iucci et al. [2000]. We note that this atmospheric effect will have no influence on the calibration accuracy. [23] In this section, the measured temperature sensitivities of the calibrator, NM64, and IGY neutron monitors are compared with one another, and also with the sensitivities calculated from the FLUKA simulation Experimental Observations Calibrator [24] Several experiments to determine the temperature coefficient of the calibrator were performed in the vicinity of the Potchefstroom 15-counter IGY neutron monitor. [25] First, the calibrator was placed in an air-conditioned room 18 m from the IGY neutron monitor hut on the same level. The temperature of this room was varied gradually, letting the calibrator s temperature change over a range of 25 C, while the IGY was kept at a constant temperature inside the monitor hut. [26] The ratios of the counting rates of the IGY to the calibrator are shown in Figure 5, as a function of the Figure 5. The temperature sensitivity of the calibrator, obtained during days in The continuous lines represent a linear regression with a 68% confidence band. 4of6

5 Figure 6. The temperature effect determined for the IGY, obtained during days in The continuous lines represent a linear regression with a 68% confidence band. temperature difference. Data are binned at 1 C intervals. The vertical error bars are the expected statistical errors. The regression line has a positive slope of a =(0.116± 0.007) %/ C. This agrees well with the value a = (0.126 ± 0.015) %/ C, determined by Moraal et al. [2003] at Sanae. [27] This experiment was repeated with the calibrator in another room closer to the monitor hut, to test whether the environment has any influence on this temperature sensitivity. The same coefficient of a = (0.119 ± 0.010) %/ C was found. Thus, these two experiments, together with the one at Sanae, demonstrate the reliability of the method. By using the standard weighting by the inverse of the invariance, the composite of the three experiments determines the temperature sensitivity of the calibrator as a =(0.118 ± 0.005) %/ C IGY [28] Knowing the temperature coefficient of the calibrator, a simpler method could be used to determine the temperature coefficient of the IGY neutron monitor. This was to put the calibrator together with the IGY inside the monitor hut and vary the temperature of the hut. [29] The ratios of the counting rates of the IGY to the calibrator are shown in Figure 6, as a function of the temperature. Since N cal = N c0 e a cal(t T 0 ) for the calibrator, and N IGY = N I0 e a IGY(T T 0 ) for the IGY, this implies that N IGY =N cal ¼ N I0 =N c0 e ð aigy acal ÞðT T0Þ ð4þ The regression line gives a slope a IGY a cal = (0.065 ± 0.011) %/ C. Since a cal = (0.118 ± 0.005) %/ C, it follows that a IGY = (0.053 ± 0.012) %/ C NM64 [30] The Bartol group operates six neutron monitor stations, with combinations of 10 BF 3 and 3 He counters, but all of them with the standard NM64 moderator, producer, and reflector geometries. As mentioned in the introduction, problems with temperature stabilization led to an independent investigation of the temperature sensitivity of their neutron monitors. Evenson et al. [2005] described these experiments with the Thule and Nain monitors, and how they found it necessary to apply temperature corrections to their data, even before applying pressure corrections. The average temperature coefficients obtained for their 3NM64 neutron monitors, with 10 BF 3 and 3 He counters, are shown in lines 2 and 5 of Table 2. It is clear that the NM64 has a much smaller temperature sensitivity with 10 BF 3 counters than with 3 He counters. In addition, it is remarkable that the 3 He counters in the NM64 configuration show a lower temperature sensitivity than in the calibrator configuration Simulation of the Temperature Effect [31] These temperature sensitivities were further investigated with numerical FLUKA simulations of the 3NM64, for both the 10 BF 3 and 3 He counters. The calculated values of the temperature coefficients in lines 3 and 6 of Table 2 show that a 3 He neutron monitor is about four times more sensitive to temperature changes than a 10 BF 3 monitor. [32] Evenson et al. [2005] pointed out that the differences can be understood by noting that the overall temperature coefficient of a neutron monitor is determined by the separate coefficients of the reflector, lead producer, moderator and counter. Heating each of these components separately in the simulations showed that the first three have positive coefficients. The 3 He counter also has a positive coefficient, but significantly smaller than the other three components, while the 10 BF 3 counter has a negative coefficient that cancels a large fraction of the positive coefficient of the other three components. The reason for this difference in the temperature coefficients of the counters is the different cross-sections of the gases Summary and Conclusion [33] A summary of the temperature coefficients obtained is shown in Table 2, ordered from large to small. The calibrator has the largest sensitivity, followed by the 3 He 3NM64, the IGY and the 10 BF 3 3NM64. The coefficients of both the 3 He and 10 BF 3 measurements are considerably larger than those of the simulations. Evenson et al. [2005] planned to conduct more simulations to try to establish the reason for this discrepancy. [34] This experiment has demonstrated that instrumental temperature effects on neutron monitors are large with respect to the desired calibraton accuracy of 0.2% and has to be taken into account. This means that all neutron monitors have to be kept at a fixed temperature or that the counting rates have to be normalized to a standard temperature. This effect will be included in the planned calibration procedure for neutron monitors. Table 2. Temperature Coefficients Neutron Monitor Value 3 He Calibrator ± %/ C 3 He 3NM64 (Thule/Nain) ± %/ C 3 He 3NM64 (Simulation) ± %/ C 10 BF 3 IGY (Potchefstroom) ± %/ C 10 BF 3 3NM64 (Thule) ± %/ C 10 BF 3 3NM64 (Simulation) ± %/ C 5of6

6 [35] Acknowledgments. This work is based upon research supported by the South African National Research Foundation, by NSF grant ATM , by the Australian Government Antarctic Division and the University of Tasmania. The authors acknowledge André Benadie and Danie de Villiers of the North-West University for the construction and programming of the calibrator. We are grateful to Leonard Shulman and James Roth of the Bartol Research Institute for their technical support and installation of the calibrator on the Polar Sea and Polar Star, and to Keith Bolton and Barry Wilson of the University of Tasmania for the construction and installation of the NM64 monitor on these vessels. [36] Amitava Bhattacharjee thanks D. Smart and another reviewer for their assistance in evaluating this paper. References Bieber, J. W., J. Clem, and P. Evenson (1997), Efficient computation of apparent cutoffs, Proc. 25th Int. Cosmic Ray Conf., Durban, Bieber, J. W., J. Clem, M. L. Duldig, P. Evenson, J. E. Humble, and R. Pyle (2001), A continuing yearly neutron monitor latitude survey: Preliminary results from , Proc. 27th Int. Cosmic Ray Conf., Hamburg, Bieber, J. W., J. Clem, M. L. Duldig, P. Evenson, J. E. Humble, and R. Pyle (2003), Cosmic ray spectra and the solar magnetic polarity: Preliminary results from , Proc. 10th Int. Solar Wind Conf., AIP Conf. Proc., 679, Clem, J. M. (1999), Atmospheric yield functions and the response to secondary particles of neutron monitors, Proc. 26th Int. Cosmic Ray Conf., Salt Lake City, 7, Clem, J. M., and L. I. Dorman (2000), Neutron monitor response functions, Space Sci. Rev., 93, Dorman, L. I., S. G. Fedchenko, L. V. Granitsky, and G. A. Rische (1970), Coupling and barometer coefficients for measurements of cosmic ray variations at altitudes of mb, Acta Phys. Acad. Sci. Hung., 29, suppl. 2, 233. Evenson, P., J. W. Bieber, J. Clem, and R. Pyle (2005), Neutron monitor temperature coefficients: measurements for BF 3 and 3 He counter tubes, Proc. 29th Int. Cosmic Ray Conf., Pune, 2, Hatton, C. J. (1971), The neutron monitor, in Progress in Elementary Particle and Cosmic-Ray Physics, vol. X, edited by J. G. Wilson and S. A. Wouthuysen, North Holland, Amsterdam. Hatton, C. J., and H. Carmichael (1964), Experimental investigation of the NM-64 neutron monitor, Can. J. Phys., 42, Iucci, N., G. Villoresi, L. I. Dorman, and M. Parisi (2000), Cosmic ray survey to Antarctica and coupling functions for neutron component near solar minimum ( ): 2. Determination of meteorological effects, J. Geophys. Res., 105(A9), 21,035 21,045. Krüger, H. (2006), A calibration neutron monitor for long-term cosmic ray modulation studies, Ph.D. thesis, North-West Univ., Potchefstroom, South Africa. Krüger, H., H. Moraal, J. W. Bieber, J. M. Clem, P. Evenson, K. R. Pyle, M. L. Duldig, and J. E. Humble (2003), First results of a mobile neutron monitor to intercalibrate the worldwide network, Proc. 28th Int. Cosmic Ray Conf., Tsukuba, Japan, 7, Krüger, H., H. Moraal, J. W. Bieber, J. M. Clem, K. R. Pyle, M. L. Duldig, and J. E. Humble (2005), Latitude surveys with a calibration neutron monitor, Proc. 29th Int. Cosmic Ray Conf., Tsukuba, Japan, 2, Moraal, H., A. Belov, and J. M. Clem (2000), Design and co-ordination of multi-station international neutron monitor networks, Space Sci. Rev., 93, Moraal, H., A. Benadie, D. De Villiers, J. W. Bieber, J. M. Clem, P. Evenson, K. R. Pyle, L. Shulman, M. L. Duldig, and J. E. Humble (2001), A mobile neutron monitor to intercalibrate the worldwide network, Proc. 27th Int. Cosmic Ray Conf., Hamburg, 8, Moraal, H., H. Krüger, A. Benadie, and D. De Villiers (2003), Calibration of the Sanae and Hermanus neutron monitors, Proc. 28th Int. Cosmic Ray Conf., Tsukuba, Japan, 7, Simpson, J. A. (2000), The cosmic ray nucleonic component: The invention and scientific uses of the neutron monitor, Space Sci. Rev., 93, Stoker, P. H., L. I. Dorman, and J. M. Clem (2000), Neutron monitor design improvements, Space Sci. Rev., 93, J. W. Bieber, J. M. Clem, P. A. Evenson, and K. R. Pyle, Bartol Research Institute and Department of Physics and Astronomy, University of Delaware, Newark, DE 19716, USA. (john@bartol.udel.edu; clem@ bartol.udel.edu; penguin@bartol.udel.edu; pyle@bartol.udel.edu) M. L. Duldig, Australian Government Antarctic Division, Kingston, Tas 7050, Australia. (marc.duldig@aad.gov.au) J. E. Humble, School of Mathematics and Physics, University of Tasmania, Hobart, Tas 7001, Australia. (john.humble@utas.edu.au) H. Krüger and H. Moraal, School of Physics, North-West University, Potchefstroom, 2520 South Africa. (helena.kruger@nwu.ac.za; harm. moraal@nwu.ac.za) 6of6