Available online at www.sciencedirect.com www.elsevier.com/locate/asr ScienceDirect Advances in Space Research 72 (2023) 5607–5625 Space weather monitoring with Health Canada’s terrestrial radiation monitoring network Chuanlei Liu a,⇑, Tamara Koletic a, Kurt Ungar a, Larisa Trichtchenko b,1, Laurel Sinclair b aRadiation Protection Bureau of Health Canada, 775 Brookfield Rd, Ottawa, ON K1A 1C1, Canada bCanadian Hazards Information Service of Natural Resources Canada, 2617 Anderson Rd, Ottawa, ON K1A 0E7, Canada Received 20 April 2023; received in revised form 5 June 2023; accepted 8 June 2023 Available online 15 June 2023 Abstract This work presents a feasibility study of utilizing Health Canada’s terrestrial radiation monitoring network, the Fixed Point Surveil- lance (FPS) network, for space weather monitoring through demonstrating detections of Forbush decrease and ground level enhance- ment events. The network is currently comprised of more than eighty sodium iodide spectrometers distributed across Canada. It was designed for terrestrial radiation monitoring but is also capable of registering cosmic radiation in a high-energy channel. Data from four- teen FPS stations for the period from 2003 to 2018 were analyzed and compared with data obtained by other ground-level cosmic radi- ation monitoring systems. The level of atmospheric impacts on measurements can be well explained, and signatures of both long-term solar cycle variations and sporadic solar events have been detected in the FPS network. The Forbush decrease amplitudes in FPS were found to be comparable to those obtained in the global muon detector network but about 2–3 times lower than those recorded by the global neutron monitoring network. This study suggests that the 20 years of cosmic ray data from the FPS network can be used for cli- matological space weather studies. In addition, the network can be readily available for real-time space weather monitoring. Crown Copyright � 2023 Published by Elsevier B.V. on behalf of COSPAR. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). Keywords: Sodium iodide detector; Terrestrial radiation monitoring network; Forbush decrease; Ground level enhancement; Solar cycle effect; Atmos- pheric effects; Space weather monitoring and forecasting 1. Introduction Space weather is initialized by a variety of solar activi- ties and manifests as significant changes in the conditions of the electromagnetic fields and the population of charged particles in the interplanetary space and near-Earth system. These changes occur in different time scales, as variations in the long-term (22-year solar magnetic cycle), interim (�27 days recurrent activity produced by Corotating Inter- action Regions (CIRs)) and short-term (days to hours), due to impacts of Coronal Mass Ejections (CMEs), solar ener- getic particles and solar flares. These eruptive and recurrent https://doi.org/10.1016/j.asr.2023.06.018 0273-1177/Crown Copyright � 2023 Published by Elsevier B.V. on behalf of C This is an open access article under the CC BY-NC-ND license (http://creativeco ⇑ Corresponding author. E-mail address: Chuanlei.Liu@hc-sc.gc.ca (C. Liu). 1 Retired. solar activities can disturb the geomagnetic and iono- spheric conditions and change the background radiation level, impacting the operations of ground-based critical infrastructure, communication and navigation networks, and potentially posing biological hazards to astronauts and aircrew (Bothmer and Daglis, 2007; Hanslmeier, 2007; Miroshnichenko, 2015). The need for operational space weather monitoring and forecasting has been recognised since 1940, significantly increasing since then due to the increased reliance of mod- ern society on infrastructure vulnerable to space weather impacts (Trichtchenko and Holmlund, 2021). Today, space weather monitoring and forecasting are achieved through continuous observations and measurements of solar activi- ties, ionospheric conditions, solar wind characteristics, magnetic field variations, and ground-level cosmic radia- OSPAR. mmons.org/licenses/by-nc-nd/4.0/). http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ https://doi.org/10.1016/j.asr.2023.06.018 http://creativecommons.org/licenses/by-nc-nd/4.0/ mailto:Chuanlei.Liu@hc-sc.gc.ca https://doi.org/10.1016/j.asr.2023.06.018 http://crossmark.crossref.org/dialog/?doi=10.1016/j.asr.2023.06.018&domain=pdf C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 tion intensities by a diversity of space-borne and ground- based monitoring systems (Bothmer and Daglis, 2007; Hanslmeier, 2007; Hochedez et al., 2005). At the ground level, monitoring of cosmic radiation variations including transient solar activity effects such as Ground Level Enhancement (GLE) and Forbush Decrease (FD) events have been united into two networks: the Global Neutron Monitoring Network (GNMN) (Mishev & Usoskin, 2020) and Global Muon Detector Network (GMDN) (GMDN, 2022). Both networks have provided continuous monitoring data on cosmic radiation intensity variations for decades. GLE events are sharp increases in the levels of cosmic radiation all the way from the upper atmosphere to the Earth’s surface as a consequence of the eruptive increase in the flux of solar energetic particles. The observation and measurement of these events provide data to derive their spectral and angular distributions, allowing for radia- tion exposure assessment for aircrew and astronauts (Mishev et al., 2017 and references therein). In addition to the existing GNMN and GMDN networks, other cos- mic radiation monitoring networks capable of GLE mea- surements with sufficient temporal resolution and spatial coverage may also be beneficial to the recently imple- mented operational space weather services developed under the auspices of the International Civil Aviation Organiza- tion (ICAO, 2018, 2019). The FD effect is a transient depression of the galactic cosmic radiation intensity resulting from the CIR- or CME-originated solar activities. Large FDs are often asso- ciated with geomagnetic storms (e.g., Forbush, 1938; Kudela et al., 2000; Belov, 2009; Kane, 2010), and both are often accompanied with precursory signs of cosmic radiation intensity anisotropy (Munakata et al., 2018; Rockenbach et al., 2011; Lingri et al., 2019; Papailiou et al., 2021). Such precursors have been detected in both GNMN and GMDN networks, where the lead time was reported to range from several hours up to a day (Papailiou et al., 2021 and references therein). Reliable identification and forecast of the geomagnetic disturbances are essential for operational space weather services, as their impacts on power grids can be devastating. Several guide- lines have been established by regulatory organisations for mitigation of the impacts of geomagnetic disturbances (e.g., NERC, 2019). This work exploits Health Canada’s Fixed Point Surveil- lance (FPS) network (FPS, 2022), a terrestrial gamma radi- ation monitoring network deployed across Canada, for space weather applications. It was motivated by the dissim- ilar type of detectors used in FPSwith respect to the GNMN and GMDN detectors, the ready availability of the network covering areas most susceptible to space weather, and the existence of similar types of instruments worldwide, suggest- ing a feasible extension to a global network. The provision and analysis of the FPS data also supports the strategies of the International Science Council’s Committee on Space Research in pursuit of the full development of space pro- grams at a national level and supports information exchange 5608 at an international level (e.g., Badruddin et al., 2021 and ref- erences therein; COSPAR, 2022). Specifically, this work analyzed the historical FD events that were observed in FPS from 2003 to 2018 and the GLE event recorded in January 2005 (GLE69). The objectives of this work are (a) to estimate the atmospheric impacts in FPS and the difference among different cosmic radiation monitoring networks; (b) to demonstrate the cosmic radia- tion response and space weather event detections in the FPS, and to quantify the response sensitivity; and (c) to understand the feasibility and value of utilizing the FPS for space weather applications. 2. Fixed point surveillance network Since 2002, Health Canada has been developing the FPS network to monitor environmental radiation levels and assess health impacts arising from both naturally occurring radioactive materials and anthropogenic radiation sources. To date, it comprises more than eighty stations deployed in Canada’s major municipalities and areas in proximity to ports or nuclear power plants. Its geographic coverage extends northward to the Arctic (Resolute, Nunavut), south to the Canada-United States border, and coast to coast from west to east. The elevation ranges from a few to a thousand meters above sea level. 2.1. RS250 detector Each FPS station is equipped with a RS250 spectrome- ter (Fig. 1, left) to detect and measure terrestrial and cos- mic radiation. RS250 is a 7.6 cm (Ø) � 7.6 cm thallium doped sodium iodide (NaI) detector, manufactured by Radiation Solution Incorporated, and has an energy reso- lution of 7.5% at 662 keV. In the normal operational mode, data collection, transmission and processing are automated on a 15-minute basis. However, this can switch to a minute- by-minute basis in the case of a nuclear emergency. A typ- ical 15-min spectrum at the Ottawa station is given in Fig. 1 (right), showing the characteristic gamma peaks of terres- trial radionuclides up to about 2.6 MeV and a cosmic radi- ation channel to register signals above �3 MeV. To facilitate radiation risk assessment, the RS250 response to terrestrial gamma radiation has been calibrated to air kerma and ambient dose equivalent (Liu et al., 2022a, 2022b) over the energy range from 15 keV to 3 MeV. The FPS dose results are integrated into two international data exchange platforms in nearly real time: EURDEP (Euro- pean Radiological Data Exchange Platform) and IRMIS (IAEA International RadiationMonitoring System). More- over, recent development has enabled cosmic ray dose esti- mation with the FPS data (Liu et al., 2018). 2.2. Cosmic ray channel The cosmic channel shown in Fig. 1 (right) forms the basis of cosmic ray monitoring in this work. An assump- Fig. 1. Left shows the RS250 detector deployed at Saanich station. Right shows a typical 15-min energy spectrum with the characteristic peaks of terrestrial radionuclides and a cosmic ray channel. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 tion here is that the contribution of non-cosmic origin (e.g., high-energy gammas from either natural radionuclide such as 214Bi decay or non-cosmic-related neutron activation processes) to this channel is marginal and therefore it is neglected. For the contributions of cosmic origin, the con- stituent particles and the associated proportions have been estimated with the Monte Carlo method. The intent of these estimations is to better understand the cosmic radia- tion response in the FPS network, with respect to other cosmic radiation monitoring networks, and the feasibility and implications of using it for cosmic ray monitoring and space weather applications. An analytical radiation model called PARMA v4.12 (Sato et al., 2008, 2017) was used to calculate the flux and generate events of each type of cosmic particles (i.e., m+/-, e+/-, c, p or n), while the detector modeling and the particle transport in medium were conducted with the FLUKA (FLUktuierende KAskade) simulation software (Battistoni et al., 2015). The muon flux above 1 GeV obtained from simulation for Ottawa was 145 m-2s�1, in near agreement with the sea-level value of 146.6 m-2s�1 derived from Patrignani et al. (2016). The simulations can well reproduce the measured counts of cosmic channel at seven selected representative stations within 7–8% accu- racy. For example, the simulation predicted a count of 1989 in a 15-min interval at the Ottawa station in July 2016, which agrees with the measured count of 2068 (one-day average) within 4% accuracy. In this case, the sim- ulated 1989 counts are composed of 63.2% muons, 22.6% photons, 11.3% electrons/positrons, 1.75% neutrons and 1.21% protons. 2.3. Cosmic radiation response difference between the FPS and GNMN/GMDN detectors The RS250 detector responds differently to cosmic rays than the GNMN and GMDN detectors. In the FPS net- work, the contributing cosmic muons (neutrons) are at a different energy level from the typical energy range to which the muon (neutron) detectors in GMDN (GNMN) 5609 sensitively respond. Moreover, the FPS signals also include cosmic electromagnetic contributions, whose generation process can be categorized into three groups: neutral pion decays, muon decays, and muon interactions with the atmosphere (radiative emissions and ionizations). The fol- lowing outlines the distinctive feature of RS250 in terms to its cosmic radiation response. � Both soft and hard muons can contribute signals in FPS, whereas the GMDN muon detectors respond to hard muons of at least hundreds of MeV (Munakata et al., 2018). � The neutron detections in FPS are expected to largely be from thermal and epithermal neutrons, while fast neu- tron detection dominates in the GNMN neutron detectors. � The pion-decay related electromagnetic components are largely produced at the �10 kPa level and their intensity decreases exponentially as altitude decreases. They account for about 10% of the overall electromagnetic intensity at sea level, but about 80% at 30 kPa height (Dorman, 2014). In contrast, the muon-related electro- magnetic components can develop along with muons’ transportation in air. � The dominant electromagnetic detections in the FPS are from muon interactions with the atmosphere that take place within a range of a few radiation lengths (i.e., sev- eral hundred metres to 1 km) above an observation site (Olah and Varga, 2017). � The muon-interaction related electromagnetic compo- nents mainly populate in the low energy region, whereas the muon-decay related electromagnetic components are mainly particles of high energy (e.g., above 100 MeV) (Olah and Varga, 2017). These differences, originating from instrumental design and response, have a few implications. The inclusion of the detection of soft muons, thermal neutrons and electro- magnetic components in FPS blurs the connection between ground-level detections and the originating primary cosmic Fig 2. A map showing the fourteen selected FPS stations in this study. 1, St. John’s; 2, Point Lepreau; 3, Iqaluit; 4, Quebec City; 5, Ottawa; 6, Pickering; 7, Thunder Bay; 8, Resolute; 9, Winnipeg; 10, Regina; 11, Calgary; 12, Yellowknife; 13, Vancouver Island and 14, Whitehorse. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 rays. In contrast, the hard muons and fast neutrons are clo- sely related to the first interactions of the primary cosmic rays in the atmosphere, thus bearing more and direct infor- mation about primary cosmic rays. As a result, with respect to GNMN or GMDN data, the FPS data are expected to be less sensitive to the primary cosmic radiation variations, resulting in relatively large fluctuation or poor precision. However, the cosmic electromagnetic signals, originated mainly from the bottom layer of the atmosphere, make FPS data especially useful for studying and modelling atmospheric effects. Moreover, the large proportion of muon contributions, as well as the muon-related electro- magnetic contributions, in FPS makes it similar to muon detectors in response to cosmic radiation variation and space weather events. In the following sections, these aspects are further studied and evaluated from an experi- mental point of view. 2.4. Station selection for FD studies In this work, a total of 14 FPS stations have been selected and retrospective data analysis was performed. A map of the selected stations and detailed information about them are presented in Fig. 2 and Table 1, respectively. Here the term ‘‘station” refers to either a single station or a clus- ter of closely deployed stations. The selection includes all the available FPS stations above 60 degrees north latitude and, below that latitude, ten approximately evenly dis- tributed stations in longitude. The vertical geomagnetic cut-off rigidity of these stations ranges from 0.01 GV to about 2 GV as in 2018. The vertical cut-off rigidity, as a good approximation to the effective cut-off rigidity, is defined as the minimum rigidity that primary cosmic rays must have in order to reach a cosmic radiation observation site (Gerontidou et al., 2021; Smart and Shea 2009). The starting year of data collection at these stations is also given in Table 1, with the earliest back to 2002. After excluding data of poor quality in the early testing stages, this work has reviewed the FD events from 2003 to 2020, covering the declining phase of the 23rd solar cycle (Aug 1996–Dec 2008) and the entire 24th cycle (Dec 2008–Dec 2019). For all these stations, their mean 15-min counts range from 2049 (at Pickering) to 3186 (at Calgary). The Poisson statistics implies that a relative uncertainty from 1.8% to 2.2% is achievable based on a single measurement. For the stations that consist of a cluster of detectors (i.e., stations #2, #6 and #13 in Table 1), their statistical uncer- tainties can be largely reduced if using the combined data. By combining all available data, the relative uncertainties are 0.90%, 0.74% and 0.98% for Point Lepreau, Pickering and Vancouver Island stations, respectively. 3. Atmospheric and instrumental effects The cosmic rays observed at ground level are nearly all secondary cosmic rays, which are generated in the cosmic showers initiated in the upper atmosphere. Their intensities 5610 at a detector site are therefore subject to changes in the atmospheric condition, and so are the measurements. By studying these dependences, the atmospheric effects can be quantified and largely eliminated so as to discern those impacts initiated by primary cosmic rays, which is of pri- mary interest in the context of space weather monitoring and forecasting. 3.1. Barometric effect One of these dependences is the atmospheric pressure measured at an observation site. In this work, the empirical function from (Malandraki and Crosby, 2018) was used to approximate the relationship between the variations of instrumental reading and the atmospheric pressure (p). The relationship is expressed as n ¼ n0eb� p�p0ð Þ; ð1Þ here b is the barometric coefficient, and n0 and p0 are the reference count and reference pressure, respectively. For all the fourteen selected FPS stations, the estimated bs are largely within a range from about �2.5% to �4.0%/ kPa. By comparison, b is typically about �7.5%/kPa (equivalently about �1% per mm Hg) for the sea level GNMN neutron stations at low rigidity cut-off (Malandraki and Crosby, 2018; Simpson et al., 1953; Dorman, 2014). For muon detectors such as ionization chamber, counter telescope, or plastic scintillators in GMDN, the barometric coefficients normally range from �1.1% to �1.7%/kPa (Dorman, 2014; De Mendonça et al., 2016). The different bs found in these instruments are largely attributed to the type of cosmic ray particles to which the instrument primarily responds. This is because the particle Table 1 The geographical information of the selected FPS stations (sorted by longitude). The number beside the station name refers to station multiplicity included in this station. In this case, the elevation Above the Sea Level (ASL) is an average over all stations. The vertical cut-off rigidity was given for 2018 according to the international geomagnetic reference field model (IGRF, 2022). ID Station Name Province Latitude (deg) Longitude (deg) Elevation ASL (km) Cut-off rigidity (GV) Start year 1 St. John’s NL 47.586 �52.737 0.116 1.98 2004 2 Point Lepreau (6) NB 45.27 �66.062 0.028 1.81 2004 3 Iqaluit NU 63.747 �68.545 0.026 0.15 2011 4 Quebec City QC 46.769 �71.292 0.115 1.43 2005 5 Ottawa ON 45.374 �75.686 0.086 1.54 2002 6 Pickering (9) ON 43.843 �79.103 0.08 1.71 2002 7 Thunder Bay ON 48.372 �89.312 0.2 1.1 2004 8 Resolute NU 74.705 �94.969 0.041 0.01 2012 9 Winnipeg MB 49.902 �97.213 0.235 0.98 2004 10 Regina SK 50.451 �104.674 0.576 1.05 2003 11 Calgary AB 51.108 �114.027 1.083 1.21 2004 12 Yellowknife NT 62.476 �114.469 0.215 0.26 2003 13 Vancouver Island (5) BC 49.25 �123.23 0.07 1.78 2003 14 Whitehorse YT 60.7 �135.0 0.731 0.66 2011 C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 and its energy level largely determine its atmospheric absorption/attenuation effect, so the intensity variation magnitude differ for different cosmic rays as atmosphere depth or barometric pressure changes. As detailed in Appendix A, the barometric effect has been quantitatively assessed in this work by using the particle absorption lengths in the atmosphere and the composition proportions in signal. The estimations can well explain the level of baro- metric effects found in all three types of instruments. These results in turn can be considered as a verification of the cos- mic radiation composition and proportion, especially in the RS250 cosmic radiation signal that were obtained from simulations. 3.2. Temperature effect As shown in Section 2.2, a large portion of the RS250 signal in the cosmic channel originates from cosmic muons. So, the temperature effect that was observed in GMDN is also expected in the FPS network. In this work, the effec- tive muon generation level method (De Mendonça et al., 2016) was used describe the temperature effect. In formula, it is given by Dn ¼ n� n0 n0 ¼ u H � H 0ð Þ þ v T � T 0ð Þ ¼ uDH þ vDT ; ð2Þ where m is the temperature coefficient in %/�C and u is the decay coefficient in %/km. The generation level here refers to the conventional 10 kPa atmospheric pressure level, of which the temperature (T) and height (H) variations are used to relate to the muon count rate variation (Dn) observed at ground level (Duperier, 1944; Dorman, 2014; De Mendonça et al., 2016). The temperature effect was analyzed in the FPS data, for further details, please see Appendix B. 3.3. Instrumental effect The RS250 detector currently deployed in the field is the upgraded version of the original design, of which the 5611 replacement time differs at different stations. By looking at the long-term FPS data, changes in the seasonal pattern were noticeable at some stations after certain specific times. Studies showed that the starting time of seasonal pattern changes coincides with the detector replacement time. Therefore, it is more likely these pattern changes are the result of detector replacement. The exact reason causing the seasonal pattern changes in FPS is under investigation. To account for the seasonal variations and pattern changes, the ground level temperature (Tgrd) was added into the fitting formula and its relationship with count vari- ation (Dn) is expressed as Dn ¼ aDT grd . Here a is the instru- mental or seasonal coefficient in %/�C. 3.4. Fitting and correction procedures In this work, the pressure effect was studied first and separately from other impacts (i.e., the temperature and instrumental impacts). This is the general practice as adopted in treating GNMN and GMDN data (De Mendonça et al., 2016), and the b results assessed this way make comparisons straightforward and more mean- ingful between different types of instruments. With the pressure-corrected data, a simultaneous fit on temperature and seasonal variations was performed afterwards to esti- mate these remaining effects. Because more work is required to interpret the exact meaning of a, no compar- isons were made for this variable. Note that for the FD analysis in this work, which mainly focuses on short-term variations in days, the change in seasonal coefficient a is not expected to cause any significant differences on the final results. The u and m results were used in comparing the tem- perature effects between FPS and GNMN/GMDN data. To check any possible temporal variations of these coef- ficients, fit was first conducted on a yearly basis. However, for the final coefficients used for raw data corrections, they are obtained specifically from two quiet solar periods (ei- ther 2008–2009 or 2018–2019). This is because the data col- lected over these years feature less variations in cosmic C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 rays, allowing for a better discernment and characteriza- tion of all atmospheric effects. More details on yearly- based fitting results can be found in Appendices A and B. For those stations that have operated throughout these two solar quiet periods (see Table 1), two sets of coefficients were obtained to correct data, one for each quiet period. However, for Iqaluit, Resolute and Whitehorse (started after 2011) stations, the first set of coefficients was esti- mated with the data from the first year of complete and good data, while the second set was from the second quiet solar period. For all stations, the raw data were split into two parts, separated by the time of detector replacement (which usually happened after the first quiet solar period). Then these two parts were corrected separately by applying the applicable set of coefficients for that part. 4. Other data used in this work 4.1. Global neutron and muon monitoring data The global neutron (GNMN) and muon (GMDN) monitoring data were used in this work for comparison purposes in terms of cosmic-ray response, atmospheric effects, and sensitivities to space weather events. The neutron data are the pressure-corrected hourly data from the Fort Smith neutron station from 2002 to 2021, retrieved from the Neutron Monitor Database (NMDB) website (NMDB, 2022). This station has a geomagnetic cut-off rigidity less than 0.3 GV. For muon monitoring data, they were from the multi-directional cosmic muon telescope at Nagoya (35.15N, 136.97E, elevation 77 m) and downloaded from the GMDN website (CREST, 2022). This work used only the vertical direction muon data collected from 2006 to 2018 (pressure-corrected), which have an effective geomagnetic cut-off rigidity at 11.5 GV. 4.2. Forbush decrease event list, information and selection The FD event information was retrieved from the Insti- tute of Terrestrial Magnetism, Ionosphere and Radiowave Propagation (IZMIRAN) database to characterize the effects and interplanetary disturbances. This is the Forbush Effects and Interplanetary Disturbances (FEID) database, accessible through the IZMIRAN Center for Space Weather Forecasts website (IZMIRAN, 2022). FEID is a historical FD event database that contains a series of FD characteristic attributes such as relevant time/date, sup- pression amplitude, its solar source, solar wind and cosmic ray variations, and disturbances of the interplanetary mag- netic and geomagnetic fields. The interplanetary cosmic ray variations and anisotropy were derived from the global sur- vey method (GSM) (Belov et al., 2018; Grigoryev et al., 2019). Multiple FD events have been selected in this work to study the FD effects in FPS. The selection criteria are based on two parameters that are extracted from the FEID data- 5612 base: the FD strength (Magn) and the FD identification reliability (Qs). Here the parameter Magn is the FD mag- nitude (in percent) for particles with 10 GV rigidity (FEID, 2022), calculated as maximal range of cosmic radi- ation density variations during the event, obtained by the Global Survey Method from GNMN data (Belov et al., 2018). Qs is the quality of event identification associated with the solar source information. Only those FD events that are reliably identified as pro- duced by solar activity (i.e., Q > =4) and have a Magn value equal or >5% have been selected. Table 2 lists all selected events with onset date/time, FD magnitude (Magn), the identification quality with the solar source association (Qs), geomagnetic disturbance indices (DstMin and Kpmax), the maximal Interplanetary Magnetic Field (IMF) strength in the event (HMax), the maximum equato- rial anisotropy (Axym), and the duration of the main depression phase (TMin). In total, 30 FD events were selected between 2003 and 2020. The majority of the selected events last from a few hours up to 34 h. They are mostly of CME origin preceded by significant solar flares, and about 20% of them have a mixed sources of CME and CIR. Most of them are charac- terised by the Dst index <�50 nT and Kp > 5. Because the Sun was in the ascending phase from 2018 to 2020, no large FD events occurred during this period. The magnitude of FD ranges from 5% to 27%, with maximum equatorial ani- sotropy (Axy) of cosmic radiation ranging from �1% to 10%. From 2003 and to 2021, a total of nine GLE events (GLE65 to GLE73) have been detected in the GNMN neu- tron network (GLE, 2022), with a maximum enhancement ranging found from 3.9% to �4800% based on 5-min data. Of them, only the January 2005 GLE event (GLE69) was observed in the FPS network. 4.3. Meteorological and atmospheric profile data To study the meteorological impact, the weather data, including hourly pressure and temperature records, were retrieved from the Environment and Climate Change Canada (ECCC) historical database (ECCC, 2021). Fur- thermore, the upper atmospheric impact has been taken into account by considering the atmospheric temperature and height variations at the 100 mbar (10 kPa) isobaric level. This level is assumed to be the atmosphere depth where the primary cosmic ray protons have their first inter- actions and the muon’s parent, charged pions and kaons, are largely produced. The data are extracted from the Glo- bal Forecast System (GFS) of the NCEP (National Centres of Environmental Protection) (Kalnay et al., 1996; NCEP, 1994), which have 6-hour temporal resolution and 1.0- degree spatial resolution. To get hourly data at a specified FPS location, linear interpolation has been performed on the original grid data. As a global weather prediction sys- tem, GFS runs numerical modeling and analysis four times a day and can provide up to 16-day forecasting data. An example of these data is provided in Appendix C. Table 2 The selected FD event list. All data were extracted from the IZMIRAN FEID database (IZMIRAN, 2022). The parameter Magn is the FD magnitude (in percent) for particles with 10 GV rigidity. Qs refer the quality of FD identification associated with solar source information. Datetime Magn (%) Qs DstMin (nT) Kpmax HMax (nT) Axym (%) TMin (hour) 0 2003-05-29 18:25 6.6 4 �144 9� 29.4 1.57 18 1 2003-10-21 18:00 6.9 4 �61 6 11.8 1.99 25 2 2003-10-24 15:24 5.8 4 �49 7� 34 2.97 8 3 2003-10-29 6:11 27 5 �353 9 47.3 10.42 10 4 2003-10-30 16:19 14.3 5 �383 9 38 4 10* 5 2004-01-22 1:37 8.6 4 �149 7 25.4 4.32 8 6 2004-07-26 22:49 13.5 5 �197 9� 26.1 2.6 6 7 2004-09-13 20:03 5 4 �50 5+ 25.5 2.6 34 8 2004-11-07 18:27 5.2 4 �340 9� 47.7 2.01 8 9 2004-11-09 18:25 8.3 4 �289 9� 40.3 4.17 8 10 2005-01-17 7:48 5.4 4 �74 7 36.8 1.59 17 11 2005-01-18 4:00 7.7 4 �121 8� 21.2 3.78 12 12 2005-01-21 17:11 9 5 �105 8 30.4 5.26 10 13 2005-05-15 2:38 9.5 5 �263 8+ 54.8 2.53 4 14 2005-09-11 1:14 12.1 4 �147 8� 18.4 5.76 10 15 2005-09-12 6:05 5.1 4 �90 7 9.2 2.58 16 16 2006-12-14 14:14 8.6 5 �146 8+ 17.7 4.35 12 17 2011-02-18 1:30 5.2 4 �30 5 31 1.55 7 18 2012-03-08 11:03 11.7 5 �143 8 23.1 2.68 21 19 2012-03-12 9:14 5.7 5 �51 6+ 23.6 3.27 18 20 2012-07-14 18:09 6.4 4 �127 7 27.3 2.62 23 21 2013-04-13 22:54 5.3 5 �6 3+ 12.9 2.98 29 22 2013-06-23 4:26 5.9 4 �49 4+ 7.6 1.7 20 23 2014-02-27 16:50 5.1 5 �99 5+ 16.6 2.35 25 24 2014-09-11 23:45 8.1 5 �16 5+ 14 1.58 9 25 2014-09-12 15:53 8.5 5 �75 6+ 31.7 2.09 9 26 2014-12-21 19:11 7.8 4 �51 5+ 16.7 1.17 13 27 2015-06-22 18:33 8.4 4 �204 8+ 37.7 3.29 13 28 2017-07-16 5:59 5.5 5 �72 6 23.7 2.17 17 29 2017-09-07 23:00 6.9 5 �124 8+ 27.3 3.29 14 * The TMin of the event 4 was take from (D’Andrea et al., 2009) instead because it was missing in the IZMIRAN FEID database. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 5. Results and discussions 5.1. Observation of long-term solar-cycle variations in the FPS network Fig. 3 shows an example of the long-term observation data and the corrected data that were obtained from the FPS Ottawa station. The overlaid sunspot data, extracted from the solar influences data analysis center of the Royal Obser- vatory of Belgium (https://www.sidc.be/silso/), are used to represent solar activity variations. The corrected data show a relatively smoother trace than the raw data, suggesting that the atmospheric impacts can be largely reduced and con- trolled. Overall, the data exhibit a long-term variation that anti-correlates with solar activities represented by the sunspot trace. At some specific times as marked by the vertical cyan lines, the detector replacement happened and the data before and after replacement may shift by a certain extent even though they were reconciled to our best knowledge. 5.2. Observation of Forbush Decrease (FD) effects in FPS – Individual event illustration Based on the onset time given in Table 2, each FD event has been investigated within a time window of multiple 5613 days in which the event took place. The date range chosen was one day before and four days after the onset time such that there are enough pre-event data to establish a baseline and the duration is long enough to study the recovery phase. The baseline was defined as the averaged counts (nbl) of the day before the event, same for all the muon, neutron and FPS data whenever applicable. The amplitude of depression (A) was calculated from the difference of the data (n) at any moment within the FD range relative to nbl, as A = (n � nbl)/nbl = n/nbl � 1. An example of Forbush depression event is shown in Fig. 4, which includes the hourly data from the combined nine Pickering stations, Fort Smith neutron station and Nagoya muon station. The black horizontal line over the first 24 h is the baseline used to determine the amplitude of depression (A), as illustrated by the downward arrows. The vertical dashed lines mark the times (TFD) when a local minimum amplitude occurs for a specific observation site. The amplitude A at TFD is then defined as the FD ampli- tude (AFD), as illustrated by the horizontal dashed lines in the figure. Here the TFD was searched within a time win- dow of few hours from the reference TFD time in the data- base. In this way, a locality in proximity to the same reference TFD is ensured to make the AFD comparison among different monitoring data more meaningful. In real- https://www.sidc.be/silso/ Fig. 3. The weekly-averaged raw (grey dot trace) and atmospheric-corrected data (darkblue trace) at the Ottawa station, as well as the sunspot observations (pink dash trace). The vertical cyan lines mark the time of detector replacement. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 ity, TFD could be searched throughout a wider time win- dow, leading to a minima time which may deviate from the local minima TFD found here. For example, if perform- ing a global amplitude minimum searching in the FPS data, the TFD is actually several hours earlier than what was marked in Fig. 4. This difference is due to stochastic fluctu- ation in the data and is also dependent on the geographical location of observation site. The resulting AFD difference between these two searching scenarios is generally at the level of statistical uncertainty in the data. The data from all three types of detectors share a com- mon pattern during the FD depression and recovery Fig. 4. An example of the FD event observed in the FPS Pickering system (#18 in Table 2), Fort Smith neutron station and Nagoya muon station. The black horizontal line in the first 24 h is the baseline from which the amplitude of suppression is determined, as the downward arrows show. The vertical dashed lines mark the times when a local maximum depression was found (TFD), whereas the horizontal dashed lines indicate amplitudes at the minima (AFD). 5614 phases, as shown in Fig. 4. They all start with a weak pre-increase in the first 24 h before the onset time, although the increases in muon and FPS data are less statistically sig- nificant. The onset time of this event at ground level is 2012-03-08 11:03 UTC, followed by about a 21-hour grad- ual decrease period before attaining its minimum. An upturn trend can be found in all data after TFD. As for the magnitude of the FD effect, it clearly shows that the neutron data are highly suppressed with an AFD of about �12.5%. In contrast, the FD impact found in the muon and FPS data is less significant with an AFD amplitude only at a level of �3% to �3.5%. Note that the statistical fluctuation in Pickering data is at a level of 1% or less. If using AFD to characterise the detector response sensitivity, the results in this event suggest that the GNMN neutron detector is more sensitive to FD impacts than muon and FPS detectors. In addition to the apparent difference on AFD, the depression level of the muon and FPS data is found to vary relative to the neutron data as FD evolves in time, as shown in Fig. 5 (left). Here the relative depression is defined as the ratio between two data sets, each of which normalized to its baseline value (n/nbl). In Fig. 5 (left), both ratios on the first day (baseline) have a unit ratio, followed by an increase on ratio during the following �21 h. The increase here suggests the depression level found at muon and FPS data is less than that in neutron data. During this gradual depression phase, muon and FPS data match rea- sonably well. In the recovery phase, these ratios are decreasing until somewhere around 2012-03-11, after which a relatively stable ratio with a diurnal pattern can be seen. To quantify the relationship between the FPS and muon data, Fig. 5(right) projects the two individual one-hour data sets found in Fig. 5 (left) into a scatter plot. This plot demonstrates that a general similarity in the depression C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 level exists between the responses in the GMDN and FPS networks. A linear regression fit predicts an approximately 1 to 1 ratio for this relationship. Several more examples are presented in Figs. 6 and 7 to demonstrate more details in the FPS’ response to a variety of FD events. Fig. 6(left) shows a fast depression event (#6 in Table 2) in July 2004 and Fig. 6(right) two consecutive FD events taking place within two days in November 2004 (#8 and #9 in Table 2). In both cases, the FPS cap- tures the overall varying features, and its response is similar to that of neutron detector. The corresponding Nagoya muon data for this period of time were not provided at (CREST, 2022); no muon data are therefore presented for both events in Fig. 6. Fig. 7 shows a weak FD event (#17 with a Magn of 5.2%). The stochastic fluctuations in the FPS data are nearly comparable to its depression level, indicating the low detection limit at this level. Only a slight geomagnetic disturbance was recorded in this event with DstMin = �30 nT and Kp = 5. 5.3. Observation of the ground level enhancement 69 event in the FPS network A more interesting but complicated showcase is given in Fig. 8, showing five FD events (the three strongest are numbered from #10 to #12 in Table 2) and one Ground Level Enhancement event (GLE69) that all took place in a short period of a week from Jan 17 to Jan 21, 2005. A total of five FD events took place with Magn values of 5.4% on Jan 17th at 7:48 UT, 7.7% on Jan 18th at 4:00 UT, 4.7% on Jan 18th at 19:00 UT, 0.5% on Jan 20th at 4:00 UT and 9.0% on Jan 21st at 17:11 UT, respectively. All FD events except for the weakest event on Jan 20th are marked in Fig. 8 by the vertical purple lines. The GLE69 event, as seen as the sudden rising in Fig. 8 on Jan 20th at 7:00 UT, was observed in the FPS as shown by the rise in count rate in the data recorded at the Ottawa, Fig. 5. Left: the temporal ratio series of the FPS and muon data. Right: the sc hourly basis and was scaled in reference to its baseline value (n/nbl). The FPS da data from Fort Smith station. 5615 Pickering and Yellowknife stations. The previous work (Liu et al., 2019) has reported a level of count rate increase at +16.39% based on the 15-min raw data collected at the FPS Yellowknife station. This amounts to a 12- to 15-fold lower sensitivity of the FPS response to GLE69 relative to that of the GNMN neutron detectors. In this work, this event was re-analysed with the atmospheric effect- corrected data. A slightly higher enhancement level of +16.53% was found in the Yellowknife station data when using the 15-min corrected data. If using hourly data (i.e., the data shown in Fig. 8), this level becomes +7.89% for corrected data and +7.74% for raw data. The small difference is understandable as during the very short rise time the change in the atmospheric conditions is not considerable enough to lead to a large difference. With respect to the pre-onset hours, the surface atmospheric pressure varies less than 0.1%, whereas the change in the 10 kPa level height is at 0.13% level at the peak time. GLE69 is the only GLE event observed in FPS so far with statistical significance. From 2003 to 2021, a total of nine GLE events (GLE65 to GLE73) have been detected in GNMN (GLE, 2022). For all these GLE events but GLE69, the maximum enhancements found in the GNMN neutron network range only from 3.9% to 44.7% based on 5-min data. By adjusting the GNMN data to 15-min count rate as in FPS and considering the 12–15 fold lower sensi- tivity of the FPS detectors, the expected enhancement in FPS is comparable to or much lower than the level of its statistical uncertainty. This explains why only the GLE69 event has been observed in the FPS during this period. 5.4. Comparison of the FD effects between the FPS, GNMN and GMDN data – Data ensemble effects In order to quantify the instrumental sensitivity to FD effects, the FPS AFD values were compared with these obtained from the GNMN and GMDN networks for atter plot of the ratios between muon and FPS data. Here each data is on ta is from Pickering system, muon data from Nagoya station, and neutron Fig. 6. Two FD examples observed in the FPS Pickering stations (blue) and the Fort Smith neutron station (dotted purple). Left: The FD event #6 in Table 2. Right: The FD events #8 and #9. The black horizontal line in the first 24 h is the baseline from which the amplitude of suppression is determined, as the downward arrows show. The vertical dashed lines mark the times when a local maximum depression was found, whereas the horizontal dashed lines indicate amplitudes at the minima. Fig. 7. The FD event #17 observed in FPS. Left is the amplitude of depression in time, and right is the scattered plot of scaled data between FPS and GMDN muon stations. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 ensembles of data collected over the period between 2003 and 2018. Three FPS data sets (Ottawa, Pickering and Yel- lowknife) are used in order to address the stochastic and geographical influence on comparisons. The neutron data from Fort Smith station (the GNMN network) and the muon vertical direction data from Nagoya station (the GMDN network) are used in comparisons. Fig. 9(left) shows the Ottawa AFD results, as well as those from the neutron and muon networks. The FD events are clustered in the two active solar periods (2003– 2006, and 2011–2016). Based on the neutron data (purple squares and right axis in Fig. 9-left), most FD events in the first active period are relatively strong with an AFD from several up to more than twenty percent, whereas the second period is populated mostly with weak FD events below several percent. These effects were also captured at the FPS (blue circles) and GMDN muon (black triangles) networks, however both are at only a few percent level (left axis in Fig. 9-left). The pre-2006 FD events imply a rela- tively stable relationship between the FPS and GNMN 5616 results, which is however obscured after 2006 with appre- ciable discrepancies at varying levels. Overall, the FPS tends to have a lower AFD (weaker effect or lower sensitiv- ity) than that of the GNMN neutron network. Fig. 9(right) shows the FPS AFD data from Fig. 9(left) as a function of the AFD data from the GNMN neutron network. Based on these event-by-event correlations, a lin- ear regression was performed to derive a numerical rela- tionship to describe the relative response sensitivity of the FPS detector to the GNMN neutron detector to these FD events, as shown in Fig. 9(right). The regression line has a slope of 0.360 ± 0.063, indicating the FD magnitude in FPS is about 36.0% of that from neutron detector. Put- ting it another way, the neutron detector is about three times more sensitive to FD events than the FPS detector. The fit has an r2 value of 0.577. In both left and right plots in Fig. 9 (as well in Figs. 10 and 11), the uncertainty on the FPS AFD is calculated based on Bayesian theorem with a uniform prior probability on AFD, as described on the treatment of efficiency errors in (Ullrich and Xu, 2007). Fig. 8. The January 2005 FD events and the GLE69 event observed in the one-hour data collected in the FPS (grey line: Ottawa station; green dotted line: Pickering station; and red dots: Yellowknife station) and the Fort Smith neutron (blue dashed line) stations. The vertical purple lines mark the four relatively strong FD events happened during this period. Fig. 9. Left: The AFD results obtained from the Ottawa station (blue circles), as well as those from the Fort Smith neutron (purple squares) and the Nagoya muon (black triangles) stations. The lines are guides for the eye. Right: GNMN neutron AFD vs FPS AFD. The line shows the result of a linear regression to this data. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 To increase the statistical significance, the count rate data from the nine stations of the Pickering system were combined first and then the FPS AFD was calculated from the combined count rate. In this case, the stochastic uncer- tainty in the FPS data which are purely associated with measurement statistics are expected to drop by 3-fold. The resultant FPS AFD was then compared with the muon (GMDN) and neutron (GNMN) data, as shown in Fig. 10, where the FPS results tend to be more correlated with the neutron results, especially during the 2nd solar active per- iod. The slope derived from the linear regression between the FPS and neutron results is 0.447 ± 0.050, while the r2 value becomes 0.793. This indicates a higher confidence that the variation in the FPS AFD is correlated with the GNMN AFD. However, there is still about 20% of the fluc- 5617 tuation in data unaccounted for, implying the existence of other type of stochastic effects than data statistics. Picker- ing is the largest clustered system in FPS. Therefore, the results from this system can be considered as the most pre- cise ones in the FPS network. All comparisons so far are based on the data collected at different locations. To reduce possible effects due to geo- graphical differences, the Yellowknife data were used for comparisons. Yellowknife is about 300 km north from Fort Smith, and the elevation and meteorological conditions are similar at both cities. Therefore, using the FPS data in Yel- lowknife is more sensible to compare with the neutron data in Fort Smith. This comparison result is shown in Fig. 11. The FPS AFD amplitudes, with this reduction in geo- graphic effect, are found to correlate more strongly with Fig. 10. Left: The AFD results obtained from the Pickering system (blue circles), as well as those from the Fort Smith neutron (purple squares) and Nagoya muon (black triangles) stations. The FPS points represent the FD decrease amplitudes calculated from the combined nine stations of the Pickering system. The lines are guides for the eye. Right: FPS AFD vs GNMN neutron AFD. The line shows the result of a linear regression to this data. Fig. 11. Left: The AFD results obtained from Yellowknife station (blue circles), as well as those from the Fort Smith neutron (purple squares) and the Nagoya muon (black triangles) stations. The lines are guides for the eye. Right: FPS AFD vs GNMN neutron AFD. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 the neutron results, than what was observed when the sta- tions were widely separated in latitude as shown in Fig. 9. The r2 value in Fig. 11(right) is 0.745. Among the fourteen selected FPS stations, three of them started to collect data around years of the 2nd solar active period. As seen in Table 1, these are stations Iqaluit (#3), Resolute (#8) and Whitehorse (#14). Owing to the lack of strong FD events, the muon and FPS AFD results fluc- tuate largely with respect to the neutron observations at these stations. As a result, the fit on these scatter plots is poor, with an r2 normally less than 0.25. In contrast, the remaining eleven stations have an r2 value typically from 0.50 to 0.86, with an averaged slope of 0.4 ranging from 0.33 to 0.50. Based on these results, it is reasonable to con- clude that the FPS station is about 2- to 3-fold less sensitive than the GNMN neutron detectors in response to FD events (i.e., based on the maximum depression amplitude AFD). The muon data used in this work do not cover the large FD events before 2006. Given the low magnitude and large 5618 fluctuation found in both the muon and FPS AFD results, reliable determination on their relationship is challenging. So, a descriptive conclusion is alternatively drawn in this work: the FPS system tends to respond to FD events in a similar way and at a comparable level as muon detectors. This conclusion is based on the ratio results as seen in Figs. 5–7 and the AFD results as seen in Figs. 9–11. 5.5. Implication for space weather monitoring and forecasting As demonstrated in the previous sections, the observa- tions of those relatively strong FD events (Magn > 5%) in FPS provide an empirical support on the use of FPS for space weather monitoring applications. The FPS data can be regarded as complementary to the other two ground-based radiation detection systems, with a sensitiv- ity comparable to the GMDN muon detector but a few times less than the GNMN neutron detector. Moreover, being an existing and well-maintained surveillance net- C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 work, FPS has a good geographical coverage above the mid-latitude of North America. Thus, the network is read- ily available for space weather monitoring providing the meteorological effects are corrected in real-time. The quasi-real time (15 min data cadence) FPS data can also complement the space weather monitoring programs of the Canadian Hazards Information Service, which cur- rently relies on geomagnetic, ionospheric and satellite observations. Nowadays, it is especially valuable and crucial to fore- cast the solar events that can potentially cause strong geo- magnetic disturbances and consequently lead to social disruption and economic loss. In these ground-based radi- ation monitoring networks, the current forecasting tech- nique largely relies on detection of the cosmic radiation anisotropy that is caused by impending FD events or geo- magnetic storms. This precursor anisotropy has been observed in both the GNMN (Belov et al., 2018) and GMDN systems (Munakata et al., 2000) with a typical lead time of several hours to 24 h (Papailiou et al., 2012; Leerungnavarat et al., 2003). By using the similar technique, the FPS network is also potentially useful for forecasting space weather events. However, additional efforts are required to attain this objective. To achieve a full and reliable picture of the cos- mic radiation anisotropy, a data coverage would be required over the celestial longitude region. For example, a ring of stations in the GNMN network and multiple directional data from a set of GMDN muon stations are often used. The FPS network cannot provide a full spatial coverage on its own. However, these gaps could be easily filled by the same or similar detection system existing in other locations around the world. Alternatively, it is also possible to develop a global space weather forecasting sys- tem by using a hybrid type of instrument (Chilingarian and Reymers, 2008; Kato et al., 2021) by integrating the FPS system into the existing networks. To derive the cosmic radiation anisotropy beyond the geomagnetic field, the cos- mic ray shower processes and the dynamic modulation effect of the geomagnetic field on primary cosmic rays have to be modelled and corrected. This involves modelling and possibly intensive computations. Data statistics is also a key factor influencing the accu- racy on forecasting. The current practice in GNMN and GMDN is based on hourly data, for which the stochastic variability should be below 1%, given that the level of the depleted region of FD events is at an order of 1–2%. A sin- gle FPS station has a statistical uncertainty normally >1%. However, a cluster of closely-located stations (e.g., the Pickering stations) could be a simple way to meet the sta- tistical requirement. 6. Conclusions This work studied thirty strong FD events (Magn > 5%) and a GLE event that were observed from 2003 to 2018 at Health Canada’s Fixed-Point Surveillance (FPS) network. 5619 The atmospheric effects are clear and moderate in the FPS data. The long-term solar cycle effect can be observed in both raw and corrected data. The RS250 detector of the FPS network was found to respond to FD events in a similar way as the global GNMN neutron and GMDN muon detectors. Particu- larly, the FD depression level observed in the FPS network is comparable to that obtained at the GMDN muon detec- tor. The maximum FD amplitudes in FPS were about 2–3 times lower than those observed in GNMN, implying a lower sensitivity of FPS in response to FD events than neu- tron detectors. Overall, the FPS network has collected about two dec- ades of cosmic ray data from a perspective different from but complementary to that of the GNMN and GMDN detectors. As an existing well-maintained network, the FPS system can be readily used for cosmic radiation and space weather monitoring at a regional scale, provided that atmospheric effects are corrected for. Given that the same or similar gamma radiation instruments are available around the globe, a low-maintenance global network of such a kind is economically feasible to achieve for space weather monitoring and possibly for forecasting. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments We acknowledge both the NMDB database www.nmdb. eu for providing high-resolution neutron data and the WDC-SILSO, Royal Observatory of Belgium Brussels for providing the sunspot data used in this work. Thanks are due to Environment and Climate Change Canada (ECCC) for the access to meteorological data from weather stations across Canada and to Shinshu University for providing us access to muon data from the Global Muon Detector Net- work (GMDN). We also acknowledge the cosmic ray group of the IZMIRAN of the Russian Academy of Sciences for providing the comprehensive catalog of For- bush decreases. Thanks to the reviewers and editor Dr. Peggy Ann Shea for their comments and assistance in eval- uating this paper. Appendix A. Barometric effect The atmospheric pressure measured at an observation site is indicative of the atmospheric overburden resulting from the full column mass density above that site. So, changing pressure usually implies changing mass density of atmospheric profile from the observation level up to high altitude. An empirical function (Malandraki and Crosby, 2018) exists to approximate the relationship between the varia- http://www.nmdb.eu http://www.nmdb.eu Fig. A1. The pressure effect observed in the 2008–2009 Ottawa data. The red line shows the linear regression of two variables, and the fit coefficient is given in legend. The two dashed circles in the figure show 68% and 95% confidence level regions, respectively. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 tions of instrumental cosmic reading n and atmospheric pressure (p), which is given as n ¼ n0eb� p�p0ð Þ ðA1Þ Here b is the barometric coefficient, and n0 and p0 are the reference count and reference pressure, respectively. Further simplification can be made by expanding the expo- nential term into power series and taking the first-order term. n ¼ n0 � ½1þ b p � p0ð Þ� Dn ¼ n� n0 n0 ¼ b p � p0ð Þ; ðA2Þ In this form, b has a clear meaning: the percentage change of instrumental reading as pressure changes. This work takes the averages over the quiet solar years as the references, and uses a linear regression modelling the rela- tionship between Dn and the change of air pressure Dp to determine b. The atmospheric pressure effect in the FPS data has been clearly identified and demonstrated in Fig. A1. The example was based on the FPS Ottawa data collected dur- ing the quiet part of solar cycle 24 (2008–2009) and a strong anti-correlation relationship can be found between count and pressure variations. The red line over the actual scattered data is the fitted regression function, whereas the green dashed circles illustrate the 68% (one standard devi- ation r) and 95% (2r) sample coverage regions, respec- tively. The Pearson’s r in this fit is �0.88, suggesting the pressure variation approximately explains about 77% of the overall variation on count. In this example, the coefficient b is estimated to be �2. 58 ± 0.01 %/kPa, implying a unit kPa change can cause about 2.58% count change in the RS250 reading. The baro- metric fluctuation within three-sigma limit is 5.1 kPa, cor- responding to �13.2% change in the RS250 cosmic counts. The variability of the pre-corrected counts is 3.1% at 1r. After correction, it drops to 2.1%, the level of statistical fluctuation of count in Poisson statistics. The barometric effect analysis was also done on yearly- basis, as shown in Fig. A2(left) with the Ottawa data and in Fig. A2(right) for all fourteen selected stations. The large negative rs indicate a strong anti-correlation throughout all years. The relatively weak correlation found from 2011 to 2016 can be attributed to solar activity influence. This period represents the most active years in the 24th solar cycle, when the variation of primary cosmic radiation accounts for a relatively larger portion of the overall vari- ation in data. The coefficient results obtained from the Ottawa data suggest that b ranges from �2.5%/kPa to �2.9%/kPa over these years. By comparison, b is typically at about �7.5%/ kPa (equivalently about �1% per mm Hg) for the sea-level GNMN stations at low rigidity cut-off (Malandraki and Crosby, 2018; Simpson et al., 1953; Dorman, 2014). For muon detectors such as ionization chamber, counter tele- scope, or plastic scintillators in GMDN, the barometric 5620 coefficients normally range from �1.1% to �1.7%/kPa (Dorman, 2014; De Mendonça et al., 2016). The different bs found in different type of radiation detectors are attributed to multiple factors such as detector configuration (e.g., shielding scenario; the cosmic radiation component and energy range that an instrument responds), local geographic condition (geomagnetic rigidity cut-off and altitude) and the solar cycle phase in study. Among these factors, of the most importance is the type of cosmic radiation particle to which the instrument primarily responds. This is because the particle and its energy level largely determine its atmospheric absorption effect and hence the intensity variation magnitude as atmospheric depth or barometric pressure changes. By using Fi and bi to represent the composition and barometric coefficient of each constituent particle, the overall b found in one type of radiation instrument can be calculated from a composition-weighted sum of bs, expressed as b = P Fi_s bi. To a first approximation, the typical absorption lengths of cosmic ray particles in atmosphere are ln � 145 g/cm2, lp � 110 g/cm2 (Dorman, 2014) and ll � 520 g/cm2 (Ziegler, 1996) for neutrons, protons and muons, respec- tively. Note that here ll is for combined soft and hard muons at low altitude. The reciprocal of each length (i.e., b) is �6.9%/kPa for neutrons, �9.1%/kPa for protons and �1.9%/kPa for muons. The signal composition in the GNMN neutron system comprises predominantly of fast neutrons (�81%), protons (�11%) and captured muons (�7%) (Dorman, 2014). The composition-weighted sum of bs of these particles gives an overall pressure coefficient of �6.7%/kPa, about 10% less than the bs typically found in GNMN (i.e., �7.5%/ kPa aforementioned). The underestimation is partially due to the b used for muonic contribution in the estima- tion; a larger b should be used because these captured muons in GNMN are mainly soft muons. Differently in Fig. A2. The yearly-based pressure effect parameters obtained in the Ottawa station (left) and at the fourteen selected FPS stations (right). C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 the GMDN detectors, their signals mainly consist of hard muons, whose b is therefore normally less than �1.9%/ kPa. This sets an upper limit of the barometric effect in GMDN, and also agrees well with observations (i.e., �1.1% to �1.7%/kPa aforementioned). As explained earlier, the FPS signals are largely from cosmic ray muons, electromagnetic components, and hadronic components (protons and neutrons) to a lesser extent. Here muons include both hard and soft muons, whose absorption length ll can be reasonably approxi- mated by 520 g/cm2 (equivalently bm = �1.9%/kPa). For the electromagnetic components, the coefficient bs are dependent on the origins. The pion-decay related electro- magnetic components have an lem of about 120 g/cm2 (equivalently bpd em = �8.3%/kPa). For both muon-decay and muon-interaction related electromagnetic components, they are largely produced locally and thus co-vary with their parent muons in response to barometric variations (Dorman, 2014). Therefore, their barometric coefficient bl em is equivalent to bm = �1.9%/kPa. The b at the FPS Ottawa station then can be estimated by assuming F and b of each contributing component as follows: F l = 63% and bm = �1.9%/kPa; F l em = 30% and bl em = �1.9%/kPa; F pd em = 4% and bpd em = �8.3%/kPa; F p = 1.2% and bp = �9.1%/kPa and F n = 1.7% and bn = �6.9%/kPa. Here for all the cosmic radiation compo- sition proportions are from subsection 2.2. For the electro- magnetic composition, the pion-decay related electromagnetic components account for 10%. A composition-weighted sum of bs give an overall barometric coefficient of �2.3%/kPa, which well explains the observed bs level at the Ottawa station with about 10% underestima- tion, as seen in Fig. A2(left). The bs in the fourteen selected FPS stations, as shown in Fig. A2(right), are largely within a range from about �2.5% to �4.0%/kPa. The altitude impact on b can be seen from the relatively large bs found at the four highest ele- vated stations (Calgary, Whitehorse, Regina and Win- nipeg). These stations tend to be more susceptible to 5621 barometric variations than others. This effect is attributed to the change in cosmic radiation composition; the pion- related electromagnetic component and hadronic compo- nent become more copious in relative to muons as altitude increases. Furthermore, the rigidity cut-off effect can also be observed if looking at the post-detector-replacement results at St. Johns and Point Lepreau stations. These two stations have the highest cut-off rigidity and, most of the time, experience relatively weak barometric effects (i.e., bs close to �2.5%/kPa). However, the cut-off rigidity effect is not conclusive because of different results found at times prior to the detector change. Appendix B. Temperature effect As shown in Section 2.2, a large portion of the RS250 signal originates from cosmic ray muons. So, the tempera- ture effect that was observed in GMDN is also expected in the FPS system. This effect can be either positive or nega- tive. A warm temperature tends to expand air and to reduce its density. It means, for these short-lived particles such as muon’s parents, fewer interactions occur at a unit geometric path in the atmosphere but are more probable to decay to muons. In this sense, temperature has a positive effect (Grieder, 2001; Dorman, 2014). Meanwhile, the air expansion induced by warm temperature likely renders muon generation at a higher altitude, which also augments the chance of muon decay on the way to the ground. This effect is negatively correlated with muon observation on the ground (Grieder, 2001). For hard muons, the negative effect can be relatively small if their relativistic/dilated life- times are long enough to ignore the production height change due to temperature variation. So, the negative effect is important for soft muons, whereas the positive effect pre- vails for hard muons and underground muon measurements. Several methods are available (Dorman, 2014; De Mendonça et al., 2016) to describe the temperature effect in the GMDN data. For simplicity in data processing Fig. B1. The temperature and instrumental effects (left) at the Ottawa station and the year-by-year fitting parameters for all stations (right). C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 and clarity on result interpretation, the IT method (i.e., the effective muon generation level method) is used in this work. The level here refers to the conventional 10 kPa atmospheric pressure level, at which the temperature (DT) and height variations (DH) are used to relate to the muon count rate variation (Dn) observed at ground level (Duperier, 1944; Dorman, 2014; De Mendonça et al., 2016). In formula, it is given by Dn ¼ n� n0 n0 ¼ u H � H 0ð Þ þ v T � T 0ð Þ ¼ uDH þ vDT ; ðB1Þ where m is the temperature coefficient in %/�C and u is the decay coefficient in %/km. With the pressure-corrected data, the temperature and instrumental effects have been estimated by using a simul- taneous fit. Three variables are used in the fit; these are the two variables in the IT method (DH and DT) and the ground surface temperature (DTgrd as explained in section 3.3). Fig. B1(left) shows an example of the fit results obtained from the Ottawa station. In this case, the adjusted R-squared suggests only up to 30% of variation (up to 60% for some other stations) is related to these variables. For the instrumental effect, it is clear that a positive a starts in 2015 in the Ottawa station, the year after the last detec- tor replacement on June 27, 2014. Fig. B1(right) shows the yearly based a s for all stations, where the starting years with appreciable positive a s differ from one station to another. This is caused by the different detector replace- ment time in these stations. The m results in Fig. B1 show a large year-by-year fluc- tuation around zero, indicating a weak correlation of this variable with data. In contrast, the height change of the 10 kPa level is anti-correlated with data variation at a level of �2% to �4% before 2014 and �1% to �2% afterwards. Note that during the active solar years from 2011 to 2016, 5622 the influence of varying primary cosmic radiation on fit can be large, affecting the fitting results and quality over these years. This seems to be the case for all stations as implied from the bottom plot in Fig. B1(right). The detector replacements were also largely done during the same period of time; the extent of its impact on the fitting results has yet to be understood. Moreover, one may argue the u plot in Fig. B1(right) shows a long-term trend throughout all years, irrespective of the large fluctuations during the active solar years. The height impact on the observed cosmic vari- ation becomes decreasing over years. This trend could be explained, at least partially, due to climate change. Over the last two decades, the troposphere got warmer and expanded in the north hemisphere (Meng et al., 2021). The less dense troposphere mitigates the height change impact. Nevertheless, the exact reasons causing the varia- tion of these impacts are under study. In comparison, the temperature effect is small in the GNMN neutron system (Krüger et al., 2008) and is of an order of �6%/km in the GMDN muon system (based on 2007–2013 data). As shown in Fig. B1, the u in FPS during the same period of muon data is at �2%/km to �3%/km level, weaker than the u in GMDN. The weaker u in FPS can be interpreted by its mixed signal components. The muonic signal in FPS may experience a similar level of u influence as that found in GMDN. However, the ability of detecting muon-decay related electromagnetic compo- nents in FPS can compensate this impact to a certain extent. Because the major electromagnetic components in FPS are from muon interaction processes, which are inde- pendent of the changing height of muon generation (Olah and Varga, 2017), the u from overall electromagnetic con- tribution should be small but positive. Therefore, the over- all u in the Ottawa station should be weaker than �6%/ km * 0.63 = �3.78%/km. This limit generally matches the observed u levels in the Ottawa data before detector replacement, as shown in Fig. B1(left). Fig. C1. The weekly-averaged raw data at the Ottawa station, as well as the environmental data used in this work. (a): the raw data. The vertical cyan lines mark the time of detector replacement. (b) and (c) are the ground level pressure and temperature data. (d) Temperature (black) and height (blue) of the 10 kPa isobaric level above Ottawa. C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 Appendix C. Meteorological and atmospheric profile data Fig. C1(a) shows the weekly-averaged raw count rat data from the Ottawa station, along with a series of envi- ronmental data used for studying the atmospheric impacts. In Fig. C1(a), the vertical cyan lines mark the detector replacement times, where the last replacement happened on June 27, 2014. Separated by this specific time, the change in the seasonal variation pattern before and after this time is noticeable in the raw data series. The plots (b) and (c) show the environmental pressure and tempera- ture data near the Ottawa station. The temperature and height data at the 10 kPa isobaric level are given in plot (d), where seasonal cycles are clearly seen. In Ottawa, the height is in phase with ground temperature, but in anti- phase with the temperature at 10 kPa level. At stations like Calgary, Resolute, Whitehorse and Yellowknife, the 10 kPa level temperature and height are found in the same phase (with certain shifts). Note that these series are weekly-averaged and therefore are less scattered than the original hourly data. The stan- dard deviations (r) of hourly data are 0.85 kPa and 0.25 km for ground surface pressure and height at 10 kPa level, respectively. To get an idea on the overall extent of the atmospheric impact, for example, within 3r deviations or 99.7% data coverage, one can use these rs and the coef- ficients found in Appendices A (e.g., b = �2.58%) and B (e. g, u = �3%) to compute the fluctuation range of the cosmic radiation count rate. In Ottawa, this comes to ±6.6% and ±2.3%, respectively. 5623 References Badruddin, B., Aslam, O.P.M., Derouich, M., Qutub, S., 2021. Study of the development and mechanism of large amplitude decreases in cosmic ray intensity during geomagnetic disturbances in the magne- tosphere. Adv. Space Res. 68 (11), 4702–4712. https://doi.org/10.1016/ j.asr.2021.08.019 (Accessed: October 6, 2022). Battistoni, G., Boehlen, T., Cerutti, F., et al., 2015. Overview of the FLUKA code. Ann. Nucl. Energy 82, 10–18. Belov, A.V., 2009. Forbush effects and their connection with solar, interplanetary and geomagnetic phenomena. Universal Heliophysical Processes. Proceedings of the International Astronomical Union, IAU Symposium, vol. 257, pp. 439–450. Belov, A., Eroshenko, E., Yanke, V., Oleneva, V., Abunin, A., Abunina, M., Papaioannou, A., Mavromichalaki, H., 2018. The global survey method applied to ground-level cosmic ray measurements. Sol. Phys. 293 (68). https://doi.org/10.1007/s11207-018-1277-6 (Accessed: Octo- ber 6, 2022). Bothmer, V., Daglis, I.A., 2007. Space Weather – Physics and Effects. Praxis/Springer, New York, p. 438, ISBN 978-3-540-23907-9. Chilingarian, A., Reymers, A., 2008. Investigations of the response of hybrid particle detectors for the Space Environmental Viewing and Analysis Network (SEVAN). Ann. Geophys. 26 (2), 249–257. https:// doi.org/10.5194/angeo-26-249-2008 (Accessed: October 6, 2022). COSPAR, 2022. Website of the International Science Council’s Commit- tee on Space Research: https://cosparhq.cnes.fr/ (Accessed: October 6, 2022) CREST, 2022. Cosmic Ray Experimental Science Team (CREST) at Shinshu University website. http://cosray.shinshu-u.ac.jp/crest/DB/ Public/Archives/GMDN.php. (Accessed: October 6, 2022). D’Andrea, C., Poirier, J., Balsara, D.S., 2009. Experimental data and analysis of the October 2003 Forbush decrease. Adv. Space Res. 44 (10), 1247–1251. https://doi.org/10.1016/j.asr.2008.11.032. De Mendonça, R.R.S., Braga, C.R., Echer, E., Dal Lago, A., Munakata, K., Kuwabara, T., Kozai, M., Kato, C., Rockenbach, M., Schuch, N. J., 2016. The temperature effect in secondary cosmic rays (MUONS) https://doi.org/10.1016/j.asr.2021.08.019 https://doi.org/10.1016/j.asr.2021.08.019 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0010 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0010 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0015 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0015 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0015 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0015 https://doi.org/10.1007/s11207-018-1277-6 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0030 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0030 https://doi.org/10.5194/angeo-26-249-2008 https://doi.org/10.5194/angeo-26-249-2008 https://cosparhq.cnes.fr/ http://cosray.shinshu-u.ac.jp/crest/DB/Public/Archives/GMDN.php http://cosray.shinshu-u.ac.jp/crest/DB/Public/Archives/GMDN.php https://doi.org/10.1016/j.asr.2008.11.032 C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 observed at the ground: analysis of the global muon detector network data. Astrophys. J. 830 (2). https://doi.org/10.3847/0004-637X/830/2/ 88 (Accessed: October 6, 2022). Dorman, L.I., 2014. Cosmic Rays in the Earth’s Atmosphere and Underground. Springer, p. 894, ISBN: 978-9401569873. Duperier, A., 1944. A new cosmic-ray recorder and the air absorption and decay of particles. Terrest. Magn. Atmosph. Electr. 49, 1–7. ECCC, 2021. Environment and Climate Change Canada (ECCC). https://climate.weather.gc.ca/. (Accessed: October 6, 2022). FEID, 2022. The Forbush Effects and Interplanetary Disturbances (FEID) description of parameters. http://spaceweather.izmiran.ru/ dbs/fds/full-list-parameters-eng.pdf. (Accessed: October 6, 2022). Forbush, S.E., 1938. On cosmic ray effects associated with magnetic storms. J. Geophys. Res. 43, 203–218. FPS, 2022. The Health Canada’s Fixed Point Surveillance network website. https://www.canada.ca/en/health-canada/services/health- risks-safety/radiation/understanding/measurements.html. (Accessed: October 6, 2022). Gerontidou, M., Katzourakis, N., Mavromichalaki, H., Yanke, V., Eroshenko, E., 2021. World grid of cosmic ray vertical cut-off rigidity for the last decade. Adv. Space Res. 67 (7), 2231–2240. https://doi.org/ 10.1016/j.asr.2021.01.011 (Accessed: October 6, 2022). GLE, 2022. The GLE database. https://gle.oulu.fi/#/. (Accessed: October 6, 2022). GMDN, 2022. The Global Muon Detector Network. http://cr0.izmiran. ru/gmdnet/. (Accessed: October 6, 2022). Grieder, P.K.F., 2001. Cosmic rays at Earth. Elsevier Science B.V., Amsterdam. ISBN 0080530052, 9780080530055. https://doi.org/10. 1016/B978-0-444-50710-5.X5000-3 (Accessed: October 6, 2022) Grigoryev, V.G., Gololobov, P.Y., Krivoshapkin, P.A., Krymsky, G.F., Starodubstev, S.A., Zverev, A.S., Yanke, V.G., 2019. Method of Global Survey by Data of Muon Telescopes. Phys. Atom. Nuclei 82, 879–885. https://doi.org/10.1134/S1063778819660220 (Accessed: October 6, 2022). Hanslmeier, A., 2007. The Sun and space weather, second edition. Springer, Netherlands. https://doi.org/10.1007/978-1-4020-5604-8. Hochedez, J., Zhukov, A., Robbrecht, E., Van der Linden, R., Bergh- mans, D., Vanlommel, P., Theissen, A., Clette, F., 2005. Solar weather monitoring. Ann. Geophys. 23 (9), 3149–3161. https://doi.org/ 10.5194/angeo-23-3149-2005 (Accessed: October 6, 2022). ICAO, 2018. International Civil Aviation Organization, Annex 3 to the Convention on International Civil Aviation, Meteorological Service for International Air Navigation, ICAO International Standards and Recommended Practices, Twentieth Edition, July 2018. http://store. icao.int/products/annex-3-meteorological-service-for-international- air-navigation. ICAO, 2019. International Civil Aviation Organization, Manual on space weather information in support of International Air Navigation, ICAO Doc 10100, First Edition, https://store.icao.int/en/manual-on- space-weather-information-in-support-of-international-air-navigation- doc-10100. IGRF, 2022. The International Geomagnetic Reference Field website. https://www.ncei.noaa.gov/products/international-geomagnetic-refer- ence-field. (Accessed: October 6, 2022). IZMIRAN, 2022. The IZMIRAN website. http://spaceweather.izmiran. ru/eng/dbs.html. (Accessed: October 6, 2022). Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K.C., Ropelewski, C., Wang, J., Leetmaa, A., Reynolds, R., Jenne, R., Joseph, D., 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc. 77 (3), 437–472. https://doi.org/10.1175/1520-0477(1996)077<0437: TNYRP>2.0.CO;2 (Accessed: October 6, 2022). Kane, R.P., 2010. Severe geomagnetic storms and Forbush decreases: interplanetary relationships reexamined. Ann. Geophys. 28, 479–489. https://doi.org/10.5194/angeo-28-479-2010. 5624 Kato, C., Kihara, W., Ko, Y., et al., 2021. New cosmic ray observations at Syowa Station in the Antarctic for space weather study. J. Space Weather Space Clim. 11, 45. https://doi.org/10.1051/swsc/2021028. Krüger, H., Moraal, H., Bieber, J.W., Clem, J.M., Evenson, P.A., Pyle, K. R., Duldig, M.L., Humble, J.E., 2008. A calibration neutron monitor: Energy response and instrumental temperature sensitivity. J. Geophys. Res. 113 (A8). https://doi.org/10.1029/2008JA013229 (Accessed: Octo- ber 6, 2022). Kudela, K., Storini, M., Hofer, M.Y., Belov, A., 2000. Cosmic Rays in Relation to Space Weather. In: Bieber, J.W., Eroshenko, E., Evenson, P., Flückiger, E.O., Kallenbach, R. (Eds.), Cosmic Rays and Earth. Space Sciences Series of ISSI. Springer, Dordrecht. https://doi.org/ 10.1007/978-94-017-1187-6_8. Leerungnavarat, K., Ruffolo, D., Bieber, J.W., 2003. Loss cone precursors to Forbush decreases and advance warning of space weather effects. Astrophys. J. 593 (1), 587–596. https://doi.org/10.1086/376408 (Accessed: October 6, 2022). Lingri, D., Mavromichalaki, H., Belov, A., Abunina, M., Eroshenko, E., Abunin, A., 2019. An extended study of the precursory signs of Forbush decreases: new findings over the years 2008–2016. Solar Phys. 294 (70). https://doi.org/10.1007/s11207-019-1461-3 (Accessed: Octo- ber 6, 2022. Liu, C., Zhang, W., Ungar, K., Korpach, E., White, B., Benotto, M., Pellerin, E., 2018. Development of a national cosmic-ray dose monitoring system with Health Canada’s Fixed Point Surveillance network. J. Environ. Radioact. 190–191, 31–38. https://doi.org/ 10.1016/j.jenvrad.2018.04.023 (Accessed: October 6, 2022). Liu, C., Ungar, K., Zhang, W., Korpach, E., Benotto, M., Pellerin, E., 2019. Observation of ground-level enhancement across Canada’s fixed point surveillance network during the 20 January 2005 solar event. Health Phys. 117 (3), 291–299. https://doi.org/10.1097/ HP.0000000000001098 (Accessed: October 6, 2022). Liu, C., Saull, P.R.B., Martin-Burtart, N., Hovgaard, J., Korpach, E., Tulk, C., Ungar, K., Zhang, W., 2022a. Calibration of Health Canada’s Fixed Point Surveillance system for environmental radiation monitoring in terms of air kerma and H*(10). . Environ. Radioact. 253–254 (4). https://doi.org/10.1016/j.jenvrad.2022.107009 (Accessed: October 6, 2022) 107009. Liu, C., Benotto, M., Ungar, K., Chen, J., 2022b. Environmental monitoring and external exposure to natural radiation in Canada. J. Environ. Radioact. 243, 106811. https://doi.org/10.1016/j.jen- vrad.2022.106811 (Accessed: October 6, 2022). Malandraki, O.E., Crosby, N.B. (Eds.), 2018. Solar Particle Radiation Storms Forecasting and Analysis. Series: Astrophysics and Space Science Library. Springer International Publishing (Cham), Vol. 444. ISBN: 978-3-319-60050-5. Meng, L., Liu, J., Tarasick, D.W., Randel, W.J., Steiner, A.K., Wilhelm- sen, H., Wang, L., Haimberger, L., 2021. Continuous rise of the tropopause in the Northern Hemisphere over 1980–2020. Sci. Adv. 7 (45). https://doi.org/10.1126/sciadv.abi8065 (Accessed: October 6, 2022). Miroshnichenko, L., 2015. Solar cosmic rays: Fundamentals and Appli- cations, second edition. Springer Cham. https://doi.org/10.1007/978-3- 319-09429-8. (Accessed: October 6, 2022). Mishev, A., Poluianov, S., Usoskin, I., 2017. Assessment of spectral and angular characteristics of sub-GLE events using the global neutron monitor network. J. Space Weather Space Clim. 7 (1). https://doi.org/ 10.1051/swsc/2017026 (Accessed: October 6, 2022). Mishev, A., Usoskin, I., 2020. Current status and possible extension of the global neutron monitor network. J. Space Weather Space Clim. 10 (17). https://doi.org/10.1051/swsc/2020020 (Accessed: October 6, 2022). Munakata, K., Bieber, J.W., Yasue, S., Kato, C., Koyama, M., Akahane, S., Fujimoto, K., Fujii, Z., Humble, J.E., Duldig, M.L., 2000. Precursors of geomagnetic storms observed by the muon detector network. J. Geophys. Res. 105 (A12), 27457–27468. https://doi.org/ 10.1029/2000JA000064 (Accessed: October 6, 2022). https://doi.org/10.3847/0004-637X/830/2/88 https://doi.org/10.3847/0004-637X/830/2/88 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0060 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0060 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0065 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0065 https://climate.weather.gc.ca/ http://spaceweather.izmiran.ru/dbs/fds/full-list-parameters-eng.pdf http://spaceweather.izmiran.ru/dbs/fds/full-list-parameters-eng.pdf http://refhub.elsevier.com/S0273-1177(23)00448-9/h0080 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0080 https://www.canada.ca/en/health-canada/services/health-risks-safety/radiation/understanding/measurements.html https://www.canada.ca/en/health-canada/services/health-risks-safety/radiation/understanding/measurements.html https://doi.org/10.1016/j.asr.2021.01.011 https://doi.org/10.1016/j.asr.2021.01.011 https://gle.oulu.fi/%23/ http://cr0.izmiran.ru/gmdnet/ http://cr0.izmiran.ru/gmdnet/ https://doi.org/10.1016/B978-0-444-50710-5.X5000-3 https://doi.org/10.1016/B978-0-444-50710-5.X5000-3 https://doi.org/10.1134/S1063778819660220 https://doi.org/10.1007/978-1-4020-5604-8 https://doi.org/10.5194/angeo-23-3149-2005 https://doi.org/10.5194/angeo-23-3149-2005 http://store.icao.int/products/annex-3-meteorological-service-for-international-air-navigation http://store.icao.int/products/annex-3-meteorological-service-for-international-air-navigation http://store.icao.int/products/annex-3-meteorological-service-for-international-air-navigation https://store.icao.int/en/manual-on-space-weather-information-in-support-of-international-air-navigation-doc-10100 https://store.icao.int/en/manual-on-space-weather-information-in-support-of-international-air-navigation-doc-10100 https://store.icao.int/en/manual-on-space-weather-information-in-support-of-international-air-navigation-doc-10100 https://www.ncei.noaa.gov/products/international-geomagnetic-reference-field https://www.ncei.noaa.gov/products/international-geomagnetic-reference-field http://spaceweather.izmiran.ru/eng/dbs.html http://spaceweather.izmiran.ru/eng/dbs.html https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2 https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2 https://doi.org/10.5194/angeo-28-479-2010 https://doi.org/10.1051/swsc/2021028 https://doi.org/10.1029/2008JA013229 https://doi.org/10.1007/978-94-017-1187-6_8 https://doi.org/10.1007/978-94-017-1187-6_8 https://doi.org/10.1086/376408 https://doi.org/10.1007/s11207-019-1461-3 https://doi.org/10.1016/j.jenvrad.2018.04.023 https://doi.org/10.1016/j.jenvrad.2018.04.023 https://doi.org/10.1097/HP.0000000000001098 https://doi.org/10.1097/HP.0000000000001098 https://doi.org/10.1016/j.jenvrad.2022.107009 https://doi.org/10.1016/j.jenvrad.2022.106811 https://doi.org/10.1016/j.jenvrad.2022.106811 https://doi.org/10.1126/sciadv.abi8065 https://doi.org/10.1007/978-3-319-09429-8 https://doi.org/10.1007/978-3-319-09429-8 https://doi.org/10.1051/swsc/2017026 https://doi.org/10.1051/swsc/2017026 https://doi.org/10.1051/swsc/2020020 https://doi.org/10.1029/2000JA000064 https://doi.org/10.1029/2000JA000064 C. Liu et al. Advances in Space Research 72 (2023) 5607–5625 Munakata, K., Kozai, M., Evenson, P., Kuwabara, T., Kato, C., Tokumaru, M., Rockenbach, M., Dal Lago, A., De Mendonça, R. R.S., Braga, C.R., 2018. Cosmic-Ray Short Burst Observed with the Global Muon Detector Network (GMDN) on 2015 June 22. Astro- phys. J. 862 (2), 9. https://doi.org/10.3847/1538-4357/aacdfe (Accessed: October 6, 2022). NCEP, 1994. National Centers for Environmental Prediction/National Weather Service/NOAA/U.S. Department of Commerce. NCEP/ NCAR Global Reanalysis Products, 1948-continuing. Research Data Archive at NOAA/PSL: https://psl.noaa.gov/data/gridded/data.ncep. reanalysis.html. (Accessed: October 6, 2022). NERC, 2019. TPL-007-4 – Transmission System Planned Performance for Geomagnetic Disturbance Events, North American Reliability Cor- poration Standard, NERC Std. https://nercipedia.com/active-stan- dards/tpl-007-4-transmission-system-planned-performance-for-geo- magnetic-disturbance-events/. (Accessed: October 6, 2022). NMDB, 2022. The Neutron Monitor Database. https://www.nmdb.eu/ nest/. (Accessed: October 6, 2022). Olah, L., Varga, D., 2017. Investigation of soft component in cosmic ray detection. Astropart. Phys. 93, 17–27. https://doi.org/10.1016/j.as- tropartphys.2017.06.002 (Accessed: October 6, 2022). Papailiou, M., Mavromichalaki, H., Belov, A., Eroshenko, E., Yanke, V., 2012. The asymptotic longitudinal cosmic ray intensity distribution as a precursor of Forbush decreases. Sol. Phys. 280, 641–650. https://doi. org/10.1007/s11207-012-9945-4 (Accessed: October 6, 2022). Papailiou, M., Abunina, M., Mavromichalaki, H., Belov, A., Abunin, A., Eroshenko, E., Yanke, V., 2021. Precursory signs of large Forbush decreases. Sol. Phys. 296 (100). https://doi.org/10.1007/s11207-021- 01844-y (Accessed: October 6, 2022). Patrignani, C., Agashe, K., Aielli, G., et al., 2016. Particle data group. Chin. Phys. C 40 (100001), 2016. 5625 Rockenbach, M., Dal Lago, A., Gonzalez, W.D., Munakata, K., Kate, C., Kuwabara, T., Bieber, J., Schuch, N.J., Duldig, M.L., Humble, J.E., Al Jassar, H.K., Sharma, M.M., Sabbah, I., 2011. Geomagnetic storm’s precursors observed from 2001 to 2007 with the Global Muon Detector Network (GMDN). Geophys. Res. Lett. 38 (16). https://doi. org/10.1029/2011GL048556 (Accessed: October 6, 2022). Sato, T., Yasuda, H., Niita, K., Endo, A., Sihver, L., 2008. Development of PARMA: PHITS-based analytical radiation model in the atmo- sphere. Radiat Res. 170 (2), 244–259. https://doi.org/10.1667/ RR1094.1. Sato, T., Niita, K., Iwamoto, Y., Hashimoto, S., Ogawa, T., Furuta, T., Abe, S., Kai, T., Matsuda, N., Okumura, K., Iwase, H., Sihver, L., 2017. Recent improvements of particle and heavy ion transport code system: PHITS. EPJ Web Conf. 153, 06008. Simpson, J.A., Fonger, W., Treiman, S.B., 1953. Cosmic radiation intensity-time variations and their origin: I. Neutron intensity varia- tion method and meteorological factors. Phys. Rev. 90, 934. Smart, D.F., Shea, M.A., 2009. Fifty years of progress in geomagnetic cutoff rigidity determinations. Adv. Space Res. 44 (10), 1107–1123. https://doi.org/10.1016/j.asr.2009.07.005 (Accessed: October 6, 2022). Trichtchenko, L., Holmlund, K., 2021. Space Weather, Extending the Borders Beyond the Earth, WMO Bulletin Vol. 70 (2). https://public. wmo.int/en/resources/bulletin/space-weather-extending-borders- beyond-earth. (Accessed: October 6, 2022). Ullrich, T., Xu, Z., 2007. Treatment of Errors in Efficiency Calculations. arXiv:physics/0701199v1, https://doi.org/10.48550/arXiv.physics/ 0701199. Ziegler, J.F., 1996. Terrestrial cosmic rays. IBM J. Res. Dev. 40 (1), 19–39. https://doi.org/10.1147/rd.401.0019 (Accessed: October 6, 2022). https://doi.org/10.3847/1538-4357/aacdfe https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.html https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.html https://nercipedia.com/active-standards/tpl-007-4-transmission-system-planned-performance-for-geomagnetic-disturbance-events/ https://nercipedia.com/active-standards/tpl-007-4-transmission-system-planned-performance-for-geomagnetic-disturbance-events/ https://nercipedia.com/active-standards/tpl-007-4-transmission-system-planned-performance-for-geomagnetic-disturbance-events/ https://www.nmdb.eu/nest/ https://www.nmdb.eu/nest/ https://doi.org/10.1016/j.astropartphys.2017.06.002 https://doi.org/10.1016/j.astropartphys.2017.06.002 https://doi.org/10.1007/s11207-012-9945-4 https://doi.org/10.1007/s11207-012-9945-4 https://doi.org/10.1007/s11207-021-01844-y https://doi.org/10.1007/s11207-021-01844-y http://refhub.elsevier.com/S0273-1177(23)00448-9/h0265 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0265 https://doi.org/10.1029/2011GL048556 https://doi.org/10.1029/2011GL048556 https://doi.org/10.1667/RR1094.1 https://doi.org/10.1667/RR1094.1 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0285 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0285 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0285 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0285 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0290 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0290 http://refhub.elsevier.com/S0273-1177(23)00448-9/h0290 https://doi.org/10.1016/j.asr.2009.07.005 https://public.wmo.int/en/resources/bulletin/space-weather-extending-borders-beyond-earth https://public.wmo.int/en/resources/bulletin/space-weather-extending-borders-beyond-earth https://public.wmo.int/en/resources/bulletin/space-weather-extending-borders-beyond-earth https://doi.org/10.48550/arXiv.physics/0701199 https://doi.org/10.48550/arXiv.physics/0701199 https://doi.org/10.1147/rd.401.0019 Space weather monitoring with Health Canada’s terrestrial �radiation monitoring network 1 Introduction 2 Fixed point surveillance network 2.1 RS250 detector 2.2 Cosmic ray channel 2.3 Cosmic radiation response difference between the FPS and GNMN/GMDN detectors 2.4 Station selection for FD studies 3 Atmospheric and instrumental effects 3.1 Barometric effect 3.2 Temperature effect 3.3 Instrumental effect 3.4 Fitting and correction procedures 4 Other data used in this work 4.1 Global neutron and muon monitoring data 4.2 Forbush decrease event list, information and selection 4.3 Meteorological and atmospheric profile data 5 Results and discussions 5.1 Observation of long-term solar-cycle variations in the FPS network 5.2 Observation of Forbush Decrease (FD) effects in FPS – Individual event illustration 5.3 Observation of the ground level enhancement 69 event in the FPS network 5.4 Comparison of the FD effects between the FPS, GNMN and GMDN data – Data ensemble effects 5.5 Implication for space weather monitoring and forecasting 6 Conclusions Declaration of Competing Interest Acknowledgments Appendix A Barometric effect Appendix B Temperature effect Appendix C Meteorological and atmospheric profile data References