The First Ground Level Enhancement Seen on Three Planetary Surfaces: Earth, Moon, and Mars

On 28 October 2021, solar eruptions caused intense and long‐lasting solar energetic particle (SEP) flux enhancements observed by spacecraft located over a wide longitudinal range in the heliosphere. SEPs arriving at Earth caused the 73rd ground level enhancement (GLE) event recorded by ground‐based neutron monitors. In particular, this is also the first GLE event seen on the surface of three planetary bodies, Earth, Moon, and Mars, by particle and radiation detectors as shown in this study. We derive the event‐integrated proton spectrum from measurements by near‐Earth spacecraft and predict the lunar and martian surface radiation levels using particle transport models. Event doses at the lunar and martian surfaces of previous GLE events are also modeled and compared with the current event. This statistical and comparative study advances our understanding of potential radiation risks induced by extreme SEP events for future human explorations of the Moon and Mars.

10.1029/2023GL103069 2 of 10 As our Moon is in the vicinity of Earth, it is not surprising that during the GLE73 event, both its surface and orbital radiation detectors, which are the lunar Lander Neutron and Dosimetry (LND, Wimmer-Schweingruber et al., 2020) Experiment aboard the Chang'E4 Lander and the Cosmic Ray Telescope for the Effects of Radiation (CRaTER, Spence et al., 2010) in lunar orbit, detected significant enhancements of the lunar radiation level. Additionally, it is intriguing that Mars, separated by 191° longitudinally from Earth, also registered a GLE event on its surface as measured by the Radiation Assessment Detector (RAD, Hassler et al., 2012) on NASA's Curiosity rover. The exact origin of relativistic particles spreading over such a large heliospheric longitude during this and similar events is still under debate, with some suggesting that the particles are accelerated at the CME-driven shock (Papaioannou et al., 2022), while others emphasize the magnetic reconnection and confining process during the eruption (Klein et al., 2022).
In this letter, we focus on the event's radiation impact on the Moon and Mars. We compare and analyze the radiation dose data from Earth, Moon, and Mars (Section 2). We further study the event flux and spectrum (Section 3) which is used to model the lunar and martian surface radiation levels that can be compared to the in-situ measurements (Section 4). We also model the impact of past GLEs on the lunar and martian radiation environment (Section 4) to assess the potential radiation risks induced by extreme SEP events for future human explorations of the Moon and Mars.

Radiation Dose Measurements at Earth, Moon, and Mars
First we show the event dose rate measured by various detectors in the vicinity of Earth, Moon, and Mars. Dose rate (in units of J/kg/s or Gray/s) is the radiation energy deposited by incoming particles per unit time to the absorber of unit mass and is often measured in silicon detectors. Sometimes when evaluating the biological effect, dose in silicon is converted into dose in water with a conversion factor of about 1.33 accounting for the comparatively larger ionization potential of silicon. Note that, however, variation of this factor can be expected when accounting for detector size, shape, material, and the varying radiation source .
The LND on the backside of the lunar surface consists of a stack of 10 dual-segment (inner and outer segments) silicon solid-state detectors (SSDs), A-J, and the dose rate is measured in the second topmost detector B . LND only functions during local lunar daytime and has to be switched off during the cold nighttime. Therefore, its measurements are continuous for about 14 Earth days followed by 14 days of gap time. The GLE73 event is hence only seen during its declining phase by LND, shown in red in Figure 1a, while the event onset and peak (for nearly 2 days) are unfortunately not covered by LND.
Meanwhile, the CRaTER instrument orbiting the Moon captured the entire duration of the event, albeit with lower cadence during the peak time. CRaTER has three pairs of SSDs, namely D12, D34, and D56, separated by the thick tissue equivalent plastic (TEP) which simulates soft body tissue (muscle) shielding of the incoming radiation (Spence et al., 2010). During normal operational mode, D12 points toward deep space, D56 points to the lunar surface while D34 is in the middle. The CRaTER team routinely corrects its data from the orbit to the surface considering that the shielding of primary radiation by the Moon body itself varies with the orbital height and should be 2π on the lunar surface. Note that this assumption may face challenges when the surface albedo radiation from the Moon plays an important role. Figure 1a shows the dose rates measured by the different CRaTER detectors (corrected to the surface) and LND "B" detector. One can see that the dose rates recorded by the different CRaTER detectors agree very well before the SEP onset when background galactic cosmic rays (GCR) were dominating, but they differ significantly throughout most of the event duration with D12 measurement much higher and D34 detecting the lowest values. This is likely because SEPs, in comparison to GCRs, have generally lower energies and are more sensitive to the TEP shielding structure of the instrument. D12 sees more SEPs from deep space and D56 sees lunar albedo particles from the surface and some SEPs from the side (deep space), while D34, being sandwiched in the middle, is shielded against low-energy particles from both sides.
Due to the notable discrepancy, we cannot rely on the CRaTER data "corrected to the surface" to be a measure of the lunar surface radiation during the event. In order to reconstruct the surface radiation level when LND was switched off, we correlate the LND and CRaTER data during the SEP event (marked in the gray shaded area) and find correlations that can further predict the lunar surface dose rate. The procedure is shown in Figures 1b  and 1c. The surface radiation predicted by the D34 & LND correlation and the D56 & LND correlation, that is, "Surf_f2" and "Surf_f3" in (c), agree very well, while the correlation of D12 & LND in the shaded region results in a relatively poor prediction of "Surf_f1" near the event onset and end resulting in a much lower GCR background and a lower dose rate compared to the actual LND measurement at the declining phase. Therefore, we use the averaged D34 & LND and D56 & LND prediction for time periods when LND data are missing.
Near Earth, there are also various instruments monitoring the radiation environment, mounted either on-board satellites or the International Space Station. Here we use the RAdiation Measurement In Space (RAMIS) instrument data onboard the German Eu:CROPIS satellite (Hauslage et al., 2018) circling at an ∼600 km polar orbit since December 2018. RAMIS uses a silicon detector telescope with two detectors D1 and D2. Similar to CRaTER in lunar orbit, RAMIS measures the deep space radiation where no planetary shielding is present and Earth's albedo radiation within the view angle of Earth itself, with the former making a dominant contribution to the top detector D1. Selected data from D1 are collected at high latitude (polar regions) with the vertical geomagnetic cutoff rigidity close to zero, as plotted in green in Figure 2a. It shows a very similar temporal evolution, with slightly smaller values, compared to the lunar surface dose rate (derived values in empty gray circles and LND-measured values in gray dots). This is likely because RAMIS D1 is located behind 3 mm of Aluminum shielding which stops vertical incident protons below about 24 MeV, while LND B detector is only shielded by the 0.5 mm-thick A detector from the top (although the shielding of LND from the side and the bottom by the lander itself is nonuniform and can be much higher). The lunar orbital measurement by CRaTER D12 (now "uncorrected" back to the orbit) is shown in brown as a comparison. As it is the least shielded instrument directly exposed to SEPs coming from deep space which occupies most of the instrumental solid angle, its dose rate is much higher than LND and RAMIS measurements and also shows more significant bumpy structures related to the low-energy component of the SEP flux as later shown in Figure 3.
At Mars' orbit, the Liulin-MO dosimeter (Semkova et al., 2018) onboard the ExoMars Trace Gas Orbiter (TGO) has been investigating the radiation conditions at a 400-km circular orbit since May 2018. Liulin-MO contains two dosimetric telescopes, AB and CD, arranged at perpendicular directions. During nominal status, the axes of BA and DC are both 90° from the nadir. A and B (or C and D) have different and complementary energy response ranges so that together they measure the dose rate. In Figure 2, we show only the AB dose rate which is less noisy than CD data. The Mars orbit dose rate temporal evolution shows very similar profiles to that at Earth orbit, except for a slight delay of the onset and peak time. This is intriguing considering that the flare/CME was directed toward Earth in longitude (West 0°-2° and South 26°-28° as seen from Earth) while Mars was located at the backside of the Sun from the eruption site ( Figure 2c).
Detailed analysis considering the CME evolution ( Figure 2c and more details in Li et al., 2022) and the in-situ solar wind speed at Earth/Mars shows that both planets could have had a direct magnetic connection to the potential acceleration source (the shock front) which allows for the most efficient particle transport. It is shown that during the early eruptive phase when acceleration was most efficient, Earth and Mars were nearly symmetrically connected to the east and west sides of the shock which may have had similar effectiveness of acceleration assuming lateral-symmetrical expansion of the shock (e.g., Veronig et al., 2018). This could explain why the dose rate profiles at two locations are rather similar. It is also interesting to note that if Mars were much closer to the Sun (the current distance is 1.6 AU), its magnetic footpoint connected by a Parker spiral to the Sun should be further away (longitudinally) from the eruption center and it might detect much less SEP radiation enhancement.
On Mars' surface, RAD has been detecting energetic particle radiation as part of the Mars Science Laboratory (MSL) mission which landed the Curiosity rover in Gale crater in August 2012 . From top to bottom, RAD consists of three SSDs (A, B and C, each having a thickness of 300 μm), and two different scintillators D and E. In particular, dose rate is measured concurrently in two active dosimeters, that is, detectors B and E. To ease the comparison with other silicon detectors, we adopt "B" measurement as plotted in red in Figure 2. In comparison with other measurements, the Mars' surface radiation dose is much smaller during the SEP event and only shows an enhancement during the first 1.5 days of the event. This is because with the martian atmospheric shielding, only protons with energy above about 150 MeV can penetrate to the martian surface and contribute to RAD's measurement . Since these high-energy SEPs are mainly accelerated during the initial  10.1029/2023GL103069 6 of 10 eruption phase, they are primarily detected at the early phase of the SEP event (also see high-energy components in Figure 3b). Figure 2b shows SEP accumulated dose (with the GCR background subtracted and converted to dose in water). Lunar orbit dose is the highest while martian surface dose is the lowest for this event. The radiation reduction from Mars orbit to surface is mostly due to the shielding of the atmosphere as explained earlier. However, it is nontrivial to compare and explain the discrepancies between different orbital measures at 3 locations due to the differences in both planetary shielding and albedo contributions as each orbit scenario corresponds to a different ratio of planetary shielding of the primary radiation and each planet has a different albedo radiation contribution to the orbital measurement. Besides, SEPs may have strong spatial anisotropies which differ at different locations especially during its initial phase (e.g., Dresing et al., 2014) and SEP detection at planet's orbit is constrained by both the direction of the incoming SEPs and the shielding condition around detector.

Particle Measurements at Earth
Various Earth-ground NMs have detected the GLE73 event as shown in Figure 3a with 5-min data (from the neutron monitor database, https://www.nmdb.eu/data/) normalized to the pre-event background level of each NM. Each station's name and its geomagnetic cutoff rigidity are marked in the legend. It is clearly shown that the relative enhancement is larger for stations with smaller cutoff rigidities which allow more SEPs to enter the Earth's magnetosphere. More detailed analysis of the GLE73 event can be found in Papaioannou et al. (2022) and Mishev et al. (2022).
To reveal the properties of relativistic SEPs associated to GLE events, the identification of the primary SEP spectrum is necessary. We therefore load energy-dependent spacecraft data from particle detectors onboard the Advanced Composition Explorer (ACE, Stone et al., 1998), Wind (Acuña et al., 1995), Solar and Heliospheric Observatory (SOHO, Domingo et al., 1995) and the Geostationary Operational Environmental Satellite (GOES, e.g., Kress et al., 2020) as shown in Figure 3b. Note that the last channel of GOES was originally an integral channel for protons above 500 MeV. Here in order to derive the energy-differential flux, we assume an upper bound of 1.6 GeV for this channel which is the upper energy of this event constrained by ground-based NM detection (Papaioannou et al., 2022). The geometric mean of the bin edges that is 894 MeV is approximated as the effective energy of this channel. Therefore, flux of this channel is more reliable during the SEP event than during the quiet time when the dominating flux is from GCRs which have a much wider energy range.
The temporal profile of proton flux shown in Figure 3b has many complex structures related to solar eruptions and the pass-by of heliospheric disturbances which deserve detailed investigations in another study. Here we include some vertical lines marking the time of a few relevant events analyzed by Li et al. (2022) who used both the remote-sensing and in-situ data throughout this period and found eight pairs of flares and CMEs. We mark the onset times of Flare2 (associated with GLE73) and Flare4 which had a direct magnetic connection to Earth with corresponding SEPs arriving after the events. The identified in-situ arrivals of CME2 and CME8 (orange lines) are followed by enhanced flux of protons with energy below 10 MeV, especially upon CME8's arrival which corresponds to a much stronger in-situ shock (Li et al., 2022). These so-called energetic storm particles (ESPs) are believed to be accelerated by interplanetary shocks as they propagate outward (C. M. Cohen, 2006).
We also note that right after the onset of Flare2, the flux enhancement at proton energies below about 6 MeV (by ACE/Electron and Proton Alpha Monitor and Wind/3DP) is likely a contamination of secondary particles from the interaction of high-energy SEPs with spacecraft and instrument structure. Therefore the fluxes of these channels before 29 October 2021 12:00 UT are excluded from the following spectral analysis.
To compare this event spectrum with previous ones, we also plot the event spectra summarized by Raukunen et al. (2018, 65 spectra in Table 2) using spacecraft and ground-based data. Each event spectrum has been fitted by a band function as plotted in Figure 4a which shows that GLE73 has an event-integrated flux relatively low as compared to the previous events, especially at energy below 10s of MeV.

Modeling GLE-Induced Radiation Levels on the Moon and Mars
The GLE event spectra shown above can be used to model SEP induced radiation doses at the Moon and Mars. Since Moon and Earth are close neighbors in the heliosphere, it is reasonable to assume that these spectra derived at Earth can be directly applied to the Moon. Alternatively, since SEP flux distribution in the heliosphere is highly variable and may depend on both the radial and longitudinal distances from the flare/CME site (e.g., Lario et al., 2013), it is nontrivial to derive the SEP flux from Earth to Mars with a different heliospheric location. As the relative longitudinal separation of the flare/CME site to either Mars or Earth is arbitrary and thus statistically equivalent for many events, we here scale the flux from Earth to Mars only considering the radial gradient. In a simple spherical coordinate where particles expand outwards from the spherical center, a simple scaling factor of 1/r 2 can be applied to the particle flux, that is, the event fluence (with unit of counts/MeV/cm 2 ) at Mars (r ∼ 1.5 AU from the Sun) is about 1/1.5 2 times of that at Earth (at 1 AU Solar distance).
Since our Moon does not have a global magnetic field or an atmosphere, SEPs can reach the lunar surface directly and also interact with the lunar soil to generate so-called upward "albedo" radiation (e.g., Xu et al., 2022). Mars also lacks a global intrinsic magnetosphere but instead has a thin atmosphere, where SEPs can directly propagate through or lose part of its energy and generate secondaries. When reaching the surface, these particles may also interact with the martian soil to induce the surface albedo radiation (e.g., Guo et al., 2021). The interaction of particles with the lunar or martian environment has been studied via particle transport modeling through the planetary atmosphere and regolith (e.g., Banjac et al., 2018;Kim et al., 2014;Matthiä et al., 2016Matthiä et al., , 2017Mesick et al., 2018;Naito et al., 2020). Here, we use the state-of-the-art lunar radiation model "REDMoon" (Dobynde & Guo, 2021) and martian radiation model "AtRIS/Mars"  to calculate the induced radiation field by these GLE events on the surfaces of the Moon and Mars, respectively.
For each event, first the spectra of different particles (including primary and secondary protons, neutrons, electrons, gammas etc.) in both the upward and downward directions on the planetary surface are obtained. Then they 8 of 10 are assumed to impinge onto a 0.5 mm silicon slab which is chosen to be comparable to aforementioned radiation detectors, while the rest of the specific instrument/spacecraft is not considered. The total event dose in the silicon slab is then converted to dose in water using the silicon-water conversion factor 1.33 to ease the assessment of radiation effects on humans.
The surface pressures on Mars may differ significantly from one location to another resulting in different atmospheric shielding conditions for incoming energetic particles (J. Zhang et al., 2022). Therefore we perform six sets of modeling processes under martian surface pressures of 305, 530, 750, 975, and 1,200 Pa (J. Zhang et al., 2022, Figure 2) to assess the variation of the SEP radiation on different martian locations (or altitudes). We take 750 Pa as the "reference pressure" since it is closest to the pressure (742 Pa) during the GLE73 event observed by MSL in Gale crater. The modeled result of each event is shown in Figure 4b with The modeled and measured GLE73 doses are slightly different. For lunar surface, the modeled event dose is about 11 mGy while the measurement reaches about 17 mGy. But it is difficult to explain this since the lunar surface data during the first 2 days of the event were derived using CRaTER data and this procedure carries unknown uncertainties. On the other hand, the Mars surface measurement is 0.288 mGy in RAD "B" (converted to water) while the modeled dose under 750 Pa is 0.159 mGy ("GLE73 Model1" in Figure 4b). This is likely because that the general 1/r 2 scaling of the SEP fluence is not accurate for this particular event. As already shown by the comparable doses measured at Earth and Mars orbit, if we rescale the SEP flux by the dose ratio (9186/10474, Figure 2) instead of 1/r 2 , we predict the martian surface dose to be 0.315 mGy ("GLE73 Model2" in Figure 4b), a value much more comparable to the measured 0.288 mGy.

Summary and Conclusion
We analyzed the first GLE event detected at three planetary bodies: Earth, Moon, and Mars with a focus on the radiation dose rate measurement. We derived the event-integrated SEP spectra at Earth and used particle transport models to calculate the lunar and martian surface radiation levels which are compared with measurements. The modeling approach is also applied to previous GLEs and a statistical correlation between the lunar and martian surface radiation dose is derived. We also showed the significant atmospheric shielding effect on SEPs so that higher altitudes on Mars (with lower pressures) can have much higher event doses.
Both the event-spectrum and the radiation impact of the GLE73 are modest compared to previous GLEs, which may be due to the fact that (a) the parent CME/flare acceleration is not as efficient as for other events with higher total doses, and/or (b) the magnetic connectivity between the observer and the SEP source region is suboptimal (e.g., C. M. S. Cohen et al., 2021, and references therein). More multi-spacecraft observations of extreme types of SEPs would be helpful to better understand and predict the spatial distribution of the radiation environment in the inner-heliosphere during the event.
Our results suggest that for future astronauts going to the Moon or Mars, extreme SEP events (GLE-type) can induce significant radiation risks. Specifically, Acute Radiation Syndrome (ARS, https://www.cdc.gov/nceh/ radiation/emergencies/arsphysicianfactsheet.htm) can occur when large amount of dose, above ∼700 mGy, is delivered to the entire body within a short period of time. None of the events on Mars have passed this threshold, but 12 out of the total 67 events over the course of 66 years (on average one event every 5.5 years) have exceeded this level on the Moon. However, more detailed modeling including shielding around astronauts by a spacesuit shield or even a habitat and also human body geometries should be considered for an accurate quantification of the radiation impact on future Lunar explorers.