Barometric Pressure Correction to Gamma-ray Observations and its Energy Dependence
DDraft version April 24, 2020
Barometric Pressure Correction to Gamma-ray Observations and its Energy Dependence
Gauri Datar, Geeta Vichare, and Selvaraj Chelliah Indian Institute of Geomagnetism, Navi Mumbai, 410218, India Equatorial Geophysical Research Laboratory (IIG), Tirunelveli, 627011, India
ABSTRACTCosmic rays (CRs) have been studied extensively in the last century to understand the processesin the universe as well as in the solar system. The CR studies around the world continue even todayusing the latest technologies. In today’s satellite era, although many observations are made fromspace, CR observations from the ground are still viewed as an important tool. These observations,however, mostly detect the secondary cosmic rays (SCRs) produced via nuclear spallation processesduring the interactions of the primary CR with the atmospheric nuclei. Neutron, muon, and gammaparticles are the major components of SCRs as observed from the ground. It is well known that theatmospheric pressure plays an important role in the SCR flux observed on the ground. Barometricpressure correction is standard practice for neutron monitor (NM) data. However, no such correctionis applied to gamma-ray, being massless. But the pressure affects the particles such as e ± , µ ± , whichproduce gamma rays in the cascade. Subsequently, the indirect pressure dependence of the gamma-rayflux can be anticipated.We examine this aspect in detail by studying the gamma-ray counts detected by the NaI (Tl)detector. The present study confirms that there is no correlation between the atmospheric pressureand the total gamma-ray counts collected from all energies. However, the scenario differs when thegamma-ray fluxes of different energies are investigated separately. The gamma rays of energy below ∼ Keywords:
NaI (Tl) detector; gamma-ray experiments; barometric pressure correction INTRODUCTIONCosmic rays (CRs) interacting with the atmosphere produce various particles known as secondary cosmic rays (SCRs).Among these, neutron, proton, π + , π , π − are generated first. Neutral pions decay into photons and charged pionsdecay into muons, which can later produce photons. These reactions result in the abundance of photon, neutron, andmuon on the ground. Mostly, the neutron is used as a probe to study primary CRs (solar, galactic, or extragalacticorigin), sun-earth connection, etc. Neutron monitors (NMs) have been used to collect the neutron data for morethan seven decades. The meteorological effects on neutron flux due to pressure, temperature, humidity, wind, snow,atmospheric electric field have been studied extensively by Dorman (1972, 2004); Belov et al. (1993); Clem & Dorman(2000); Gerontidou et al. (2015); Platanos et al. (2015); Thomas et al. (2017). The most prominent effect is due to theatmospheric pressure (Rochester 1962; Lindgren 1962; Carmichael et al. 1968; Chiba 1976). Increase in the barometricpressure causes a decrease in the CR intensity, which has been reported since the 1920s and was interpreted as theeffect due to absorption of CR by the atmosphere (Myssowsky & Tuwim 1926). Later, Dorman (1972) discussed the Corresponding author: Gauri [email protected], [email protected] a r X i v : . [ phy s i c s . s p ace - ph ] A p r Datar et al. complex nature of barometric effect as a result of absorption, decay, and generation of SCRs. It is also known thatthe barometric effect varies with geomagnetic cutoff rigidity and altitude of the location of observation.To obtain useful data from NMs, correcting the NM data for the effect of barometric pressure is regarded as one of themost important corrections of raw data (http://cosray.phys.uoa.gr/index.php/data/nm-barometric-coefficient) . Rawdata from a neutron detector is corrected for pressure using experimental calculation of the barometric coefficient. Themethod to define the barometric coefficient of the different components of secondary cosmic rays has been studied bymany researchers in the past (Carmichael et al. 1968; Dorman 1972). The literature describes the effect of barometricpressure on the counting rate of NMs as dN = − b.dp, (1)where dN is the rate of change in counts, dp is the change in pressure and b is the barometric coefficient. IntegratingEq. 1 and assuming that there are N counts for pressure p , the corrected counts ( N ) are calculated as N = N exp[ − b ( p − p )] , (2)where p is the current atmospheric pressure. The parameters N and p are the reference values (usually, the averagevalues) of counting rate and atmospheric pressure, respectively, over the specified time period. Eq. 2 can be used forexperimental calculation of the barometric coefficient, ‘ b ’ by linear regression of N and p measurements for a specificperiod. Calculating ‘ b ’ for an NM station has become much easier because of an online tool developed by the AthensCosmic ray team as described by Paschalis et al. (2013), which uses the data available on NMDB.During the last 60 years, the factors responsible for the change in b have been investigated using the data fromnetworks of the neutron as well as muon detectors. b depends on many factors such as latitude (rigidity), altitude,and time of observation. It is found to have 11-year solar cycle dependence (Platanos et al. 2015; Gerontidou et al.2015). b is different for different types of particles detected at the same location and time. The absolute value of b forneutrons is more than that for muons (Chilingarian & Karapetyan 2011). Based on ASEC muon data, Chilingarian &Karapetyan (2011) found that the absolute value of b is inversely proportional to the muon energy, indicating that bvaries with the energy of the detected particles. Thus, we can say that the barometric pressure effect on neutron andmuon components of SCR has been studied comprehensively in the past.On the other hand, similar studies discussing the barometric pressure effect on the photon component are scarce.Being massless, photons are not expected to have a dependence on atmospheric pressure. Unlike neutron and muon,for which SCR is the only major source, photons originate from different sources. In addition to the SCRs, terrestrialradioactivity is also a major contributor to photon production. Terrestrial radioactivity produces photons up to 2.8MeV, while SCR can produce photons of energies ranging from 100 keV to tens of MeV. Therefore, understandingthe pressure dependence can be complex. To our knowledge, only a couple of papers have attempted to study therole of pressure in gamma-ray flux (Bukata et al. 1962; Chin & Standil 1968). They reported an observation showinga correlation of photon intensity with the atmospheric pressure. Though the aerial radiation surveys are known tocorrect their data for atmospheric pressure, they are actually height corrections. Their interest lies in the terrestrialradioactivity emanating from the ground. Therefore, the height correction is required for the radiation coming fromthe ground (high pressure) to the aircraft (low pressure). The pressure correction for gamma-ray data measured on theground is different than such height correction. Chin & Standil (1968) evaluated b for different energy ranges and theyfound the energy dependence of b in the range 3.8 – 183 MeV. They reported an increase in b with energy, althoughthey did not discuss the physics behind this observation. It should be noted that their observations were made at alocation with the geomagnetic cutoff rigidity, R c ≈ ◦ N, 97.2 ◦ W) and an elevation of 236 m a.s.l. Theirset-up consisted of a NaI detector of size 9 .
5” diameter and 8” length (volume 238.76 inch (606.45 cm )), which wascompletely surrounded by an efficient scintillating plastic anticoincidence shield. In their paper, they have reportedthe barometric effect for energies above 3.8 MeV, and have not discussed the effect on energies below 3.8 MeV. In fact,NaI (Tl) is considered to be very effective for gamma-ray spectroscopy up to 3 MeV, therefore, it is important to studythe pressure aspect on energies below 3 MeV as well. In the present paper, we investigate barometric pressure effectson the gamma-ray data of energies between 150 keV and ∼
10 MeV, collected using NaI of size 4 × ×
16” (volume256 inch (650.24 cm )) located near the equator (R c ≈ DATAThe gamma-ray data used in the present analysis is obtained from the NaI (Tl) detector located at EquatorialGeophysical Research Laboratory (EGRL), Tirunelveli (30 m a.s.l.; Geographic Coordinates: 8.71 ◦ N, 77.76 ◦ E). The arometric Pressure Correction to Gamma-ray Observations Figure 1.
Barometric pressure correction for NM data (Thailand station) and NaI (Tl) data detector is placed inside a temperature-controlled cabin constructed six feet above the ground. The detector is insidethe lead shielding that covers the bottom and all four sides but not the top. The top of the cabin is covered byplywood. Digital pulse processing for pulse height analysis (DPP-PHA) is implemented to record an energy histogrameach minute. The spectrum is calibrated using standard sources ( Co and
Cs) and distinct background radioactivitypeaks such as K, Tl,
Bi. The details of the experimental set-up are described in Vichare et al. (2018). Thedata obtained from this set-up have been used successfully to study the diurnal variation in gamma rays (Datar et al.2019) and the response during a cyclone event (Datar et al. 2020). The atmospheric pressure data used in the presentanalysis is obtained from an in-house automatic weather station (AWS) at EGRL. γ -RAYS AND PRESSURE VARIATIONS3.1. Total γ -ray counts The dependence of the total gamma-ray counts collected by the above set-up on the barometric pressure is ex-amined in Figure 1. Here, we have also presented the neutron data obtained from an NM station – PSNM (Thai-land), which has similar latitude and rigidity as Tirunelveli. We have randomly selected NM data from 1 to 31January 2018. For NaI (Tl), data from 1 to 12 May 2019 is presented. The x-axis denotes time in days. Thetop panels show the graphs of ln ( N/N ) vs p − p . For NM, p is the reference pressure value obtained from , which is a long-term average value. Similarly, for NaI data, p is ref-erence pressure taken as an average at Tirunelveli. N is a base reference value of counts, which is taken as theminimum value of counts in the data. The plots on the left represent the NM data from PSNM, Thailand, while theplots on the right show gamma-ray data from NaI (Tl) located at EGRL, Tirunelveli. The second panels from thetop show the barometric pressure recorded at the observation site. Third and bottom-most panels show uncorrectedand corrected counts, respectively. It can be observed that the pressure and uncorrected counts of NM data areanti-correlated with each other over a long period ( ∼
10 days) and also on a diurnal scale. The uncorrected counts ofNM data show anti-correlation with the semi-diurnal variations of the atmospheric pressure. On the other hand, nosuch dependence is observed in uncorrected counts from NaI (Tl) detector.For PSNM, the scatter plot of ln ( N/N ) vs p − p is linear and the slope of a linear fit is taken as b. Whereas forEGRL, the plot is totally scattered, indicating a lack of dependence and no reliable linear fitting is possible. Thus, itis evident from Figure 1 that the total counts collected by NaI (Tl) do not show any dependence on the atmospheric Datar et al. pressure, which is supported by the scatter plot depicted in the top panel. It should be noted that these are the totalcounts detected by the NaI (Tl) detector integrated over all the energies. However, considering various sources ofgamma rays (terrestrial radioactivity is the major source for gamma rays up to 2.8 MeV, while SCR is the dominantsource above 3 MeV), it would be interesting to conduct a similar analysis for different energies, which is presented inthe next subsection. 3.2. γ -ray counts of different energies The analysis for different energies of gamma flux during May 2019 is presented in Figure 2 ((a) 0.3-2.7 MeV; (b)3-5 MeV; (c) 5-7 MeV; (d) 7-10 MeV). The top panels in each subfigure show the graph of ln (N/N ) on the y-axisand (p-p ) in mbar on the x-axis. As mentioned earlier, N and p are the reference values of counts and atmosphericpressure, respectively. The middle panel shows the atmospheric pressure ( mbar ) and the uncorrected counts from NaI(Tl) detector within the specified energy range. The bottom panel shows the corrected counts wherever applicable,i.e., whenever good linear fit with negative slope is obtained, the correction is applied using the barometric correctionformula (Eq. 2). The goodness of the fit is determined using the statistical parameters displayed in Table 1. It canbe observed from the scatter plots that the scatter is more for the lower energy band (0.3-2.7 MeV) than the otherenergy bands. Among the rest of the energies, 3-5 MeV and 5-7 MeV have less scatter and it is possible to fit a lineartrend with a negative slope. For 7-10 MeV energy range as well, it is possible to fit a line with a negative slope. Theslopes of the linear fits in the scatter plots represent b values and are shown in Table 1. It can be noticed from Table1 that the b value is negative above 3 MeV and small but positive for the energy range of 0.3-2.7 MeV. The b valuecomputed for all energies as shown in Figure 1 is also a positive small value. The plots of pressure and uncorrectedcounts above 3 MeV seem to be anti-correlated on semi-diurnal as well as long time scales. But for energy range 0.3-2.7MeV, such relationship is not clearly seen. The parameters such as Pearson correlation coefficient (CC), R , SSE, andRMSE enlisted in Table 1 describe the statistical significance of the linear fits and hence that of b values. The CCvalues for the parameters displayed in the scatter plots are in general good above 3 MeV. CC values are ∼ − . − .
294 for 7-10 MeV range, though it is statistically significant (p < . > . .
5. Further, based on the statistical significance of CC, wedecided whether to apply the pressure correction or not. This information is shown in the last row of Table 1.We carried out similar analysis for two more data sets viz. six days from June 2019 and five days from July 2019, toconfirm the results obtained above. The important observation in the present work is about the pressure dependenceof gamma rays that varies with the energy of the gamma. The results from all these three data sets are plottedtogether in Figure 3. Figure 3(a) shows the variation of obtained barometric coefficient with energy. The mean bvalues in each energy range along with standard error (shown by vertical lines) are depicted in Figure 3 (b). Thesignificance of obtained b value is indicated by averaged CC values which are displayed by bar plots in Figure 3 (b).
Table 1.
Goodness-of-fit statistics for γ -ray counts of different energies (May 2019) Parameter All energies 0.3-2.7 MeV 3-5 MeV 5-7 MeV 7-10 MeVb (%/mbar) R CC b -0.6038 -0.6012 -0.2941SSE c RMSE d Correction
NA NA Yes Yes Yes a R indicates the success of the fit in explaining the variation of the data as it is the square of the correlation between the response valuesand the predicted response values. Value of R-square closer to 1 indicates good correlation. b CC is the Pearson correlation coefficient. c SSE is ‘sum of squares due to error’, which measures the total deviation of the response values from the fit to the response values. It isalso called the summed square of residuals. A value closer to 0 indicates that the fit will be more useful for prediction. d RMSE is ‘root mean squared error’ or the standard error of the regression. arometric Pressure Correction to Gamma-ray Observations Figure 2.
Barometric dependence of γ -ray counts of different energies during May 2019 (a) 0.3-2.7 MeV; (b) 3-5 MeV; (c) 5-7MeV; (d) 7-10 MeV. Datar et al.
Figure 3.
Summary and statistics of data from May, June, and July 2019
It can be observed that for 0.3-2.7 MeV energy range, the b values are very small and the average CC is also verysmall (0 . ∼ − .
26 and − .
27, respectively with mean CC greater than ∼ − .
34 with mean CC of 0 .
36. Larger error bars inthis higher energy band can be due to the lower number of counts in this energy range. Thus, it is clear from Figure 3which is based on the data collected during different periods that there is no pressure dependence for the energy range0.3-2.7 MeV, while clear dependence exists for the gamma of energies greater than 3 MeV and value of b increaseswith the energy above 3 MeV. DISCUSSIONThe main objective of the present study is to investigate the pressure dependence of gamma-ray data collected byNaI(Tl) scintillation detector. NaI (Tl) detector is considered best for gamma spectroscopy up to 3 MeV. However, itcan also detect gamma rays of energies higher than 3 MeV. Also, apart from gamma rays, it can detect other particlessuch as electrons, muons, neutrons as well, but the efficiency to detect these particles is very low. It may be importantto discuss the contributions of other particles in the spectrum obtained from the NaI detector. As pointed out byAvakyan et al. (2013), the ratio of efficiencies of a NaI (Tl) detector to detect gamma and neutrons varies from 5:1to 8:1 in the energy range 3 MeV – 10 MeV. This means that when a NaI (Tl) detects 50 to 80 % of the incidentgamma photons, it also detects 10 % of the incident neutrons. Similarly, it might be having some efficiency to detectother particles like muons and electrons. However, as per EXPACS simulations, which compute the flux of variouscomponents of SCR (Sato 2015, 2016), in the energy range of 113 keV –11.3 MeV, a total flux of gamma photonsequals to 2.67E-01 cm − s − . Similarly, for neutrons, it is 9.61E-03 cm − s − and 3.14E-05 cm − s − for muons( µ ± ). For electrons and positrons together, it is 1.04E-02 cm − s − . Thus, with respect to gamma, the neutrons and e ± have fluxes of 3.66 % and 3.95 %, respectively, while muons have very less abundance of 0.012 % in this energyrange. Therefore, we can rule out the role of muons affecting our observations significantly. We cannot yet ignore thecontribution of neutrons, electrons, and positrons ( ∼ ∼
10 %, and the average abundancew.r.t. gamma is 3.66 %. That means if 100 gamma particles are incident on the detector, then 3.66 neutrons will alsobe incident on it. Out of those 100 gammas, ∼ arometric Pressure Correction to Gamma-ray Observations total gamma-ray countscollected from all energies. However, the scenario differs when the gamma-ray flux of different energies are investigatedseparately. Besides the SCR, gamma rays observed on ground originate from other sources as well, such as terrestrialradioactivity, thunderstorms. Therefore, understanding gamma-ray dependence on atmospheric pressure is complex.The gamma particles with energies up to 2.7 MeV have a major contribution from the terrestrial radioactivity comingfrom the ground. While gamma as the SCR component produced in the CR cascade has a broad energy range fromfew keV to tens of MeV. Datar et al. (2019) have reported the presence of a diurnal variation in the gamma-raycounts with energy less than 2.7 MeV, which they attributed to the transport of radon in the air emanating from theground to the atmospheric boundary layer. Therefore, the gamma radiation of energy < > π + , π − , π are produced in the interactions ofprimary CRs with the atmospheric nuclei. π directly produce photons, but charged pions decay into muons, whichcan produce photons after travelling some distance in the air. Thus, photons being tertiary or quaternary particles inthis chain, their flux on the ground is a mixture of all those produced by different generation processes. This makes itsdependence on the barometric pressure not so straightforward and maybe the reason for a not very good value of CC( ∼ − . > . − .
25 to − .
30 %/mbarfor 3-10 MeV, which is similar to that of low-energy muons.CONCLUSIONGamma rays, being massless, are not expected to be affected by pressure variations. However, atmospheric pressureaffects the particles such as e ± , µ ± , which produce gamma rays in the cascade of SCRs. Thus, the indirect pressuredependence of the gamma-ray flux is observed in gamma rays of energies above 3 MeV. However, gamma rays of energiesbelow 2.7 MeV do not show such pressure dependence because of the dominance of the component of terrestrial origin,which does not get affected by atmospheric pressure variations.Present work suggests that the gamma-ray data collected on the ground for energies above 3 MeV need to becorrected for atmospheric pressure. Furthermore, the present study has been carried out at a location where themaximum pressure variation is only ∼ http://phits.jaea.go.jp/expacs/ .AUTHOR CONTRIBUTIONS STATEMENTAll authors participated in designing the experiment. GD carried out the data analysis. GD and GV analysed theresults and finalised the manuscript. ADDITIONAL INFORMATION Datar et al.
Competing interests : The authors declare no competing interests.REFERENCES: The authors declare no competing interests.REFERENCES