Introduction
Understanding the habitability of planets in the Milky Way includes evaluation of exposure to high-energy photon radiation from supernovae (SNe) and γ ray bursts (GRBs) (Piran and Jimenez, Reference Piran and Jimenez2014; Gowanlock, Reference Gowanlock2016). The predicted effects of hard photon radiation on Earth include temporarily elevated levels of cosmogenic isotopes such as 14C, 10Be and 36Cl, ionization of N species in the atmosphere, depletion of the ozone layer, increased UV radiation at the Earth's surface and delivery of fixed N to the Earth's surface. There is a rich literature describing these potential effects (Terry and Tucker, Reference Terry and Tucker1968; Ruderman, Reference Ruderman1974; Clark et al., Reference Clark, Mccrea and Stephenson1977; Rood et al., Reference Rood, Sarazin, Zeller and Parker1979; Thorsett, Reference Thorsett1995; Scalo and Wheeler, Reference Scalo and Wheeler2002; Gehrels et al., Reference Gehrels, Laird, Jackman, Cannizzo and Mattson2003; Melott et al., Reference Melott, Lieberman, Laird, Martin, Medvedev, Thomas, Cannizzo, Gehrels and Jackman2003, Reference Melott, Thomas, Hogan, Ejzak and Jackman2005; Thomas et al., Reference Thomas, Melott, Jackman, Laird, Medvedev, Stolarski, Gehrels, Cannizzo, Hogan and Ejzak2005; Galante and Horvath, Reference Galante and Horvath2007; Martín et al., Reference Martín, Galante, Cárdenas and Horvath2009; Thomas, Reference Thomas2009; Horvath and Galante, Reference Horvath and Galante2012; Gowanlock, Reference Gowanlock2016). If sufficiently intense, the radiation events can also affect other solar system atmospheres and surfaces (Duggan et al., Reference Duggan, McBreen, Hanlon, Metcalfe, Kvick and Vaughan2001; Scalo and Wheeler, Reference Scalo and Wheeler2002; Fox et al., Reference Fox, Eigenbrode, Pavlov and Lewis2017).
In this regard, records of the Earth's paleoenvironmental history during the late Quaternary Period (40 000 years BP to present) allow testing of hypotheses concerning the possible terrestrial effects (Brakenridge, Reference Brakenridge1981). On Earth, marine, lake and ice cores, speleothems, and tree ring records provide opportunities to locate and measure the traces, including through the use of cosmogenic isotopes produced in the atmosphere. At present, even the γ radiation from distant galactic magnetar flares and from GRBs in other galaxies creates measurable terrestrial ionosphere changes (Fishman and Inan, Reference Fishman and Inan1988; Inan et al., Reference Inan, Lehtinen, Moore, Hurley, Boggs, Smith and Fishman2007; Tanaka et al., Reference Tanaka, Terasawa, Yoshida and Hayakawa2008) and neutron thermalization effects from lightning-produced γ are also observed (Carlson et al., Reference Carlson, Lehtinen and Inan2010). Using tree rings, it is possible to assay relatively small changes in the cosmogenic isotope 14C on an annual or near-annual basis over thousands of years (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Miyake et al., Reference Miyake, Nagaya, Masuda and Nakamura2012; Miyake et al., Reference Miyake, Masuda, Nakamura, Kimura, Hakozaki, Jull, Lange, Cruz, Panyushkina, Baisan and Salzer2016). Thus, SN radiation events of different sizes can be detected and measured: if they can be separated from the changes produced by other environmental variables.
To understand probable exposure history from SNe, the expected emissions, distances and ages must be estimated or measured. In this regard, the sizes and energy cross-sections of γ radiation from ‘typical’ SNe in other galaxies, of various classes, and from GRBs are increasingly constrained, through theory, modelling and direct observations (Matz et al., Reference Matz, Share, Leising, Chupp, Vestrand, Purcell, Strickman and Reppin1988; Churazov et al., Reference Churazov, Sunyaev, Isern, Bikmaev, Bravo and Al2015; Wang et al., Reference Wang, Huang and Zhuo2019a, Reference Wang, Fields and Lien2019b). Also, detailed and nearly complete inventories of in-galaxy prehistoric SNe as detected by their remnant nebulae (SNRs) and/or remnant compact objects are available (Farrand and Safi-Harb, Reference Farrand and Safi-Harb2012; Zhu and Tian, Reference Zhu and Tian2013; Green, Reference Green2019). The ages and distances of most galactic SNRs in this time period are constrained by observations, and in many cases, the sizes of progenitor stars and the total explosion energies have been determined.
SNRs are strong radio emitters and most objects within the Milky Way are believed to have already been detected: although newly discovered ones are occasionally reported (Foster et al., Reference Foster, Cooper, Reich, Kothes and West2013; Gao et al., Reference Gao, Reich, Reich, Hou and Han2020). Thus, the galactic inventory may be nearly complete for the past ~40 000 years (Green, Reference Green2019). The remnant nebulae expand through time and eventually merge into the interstellar medium: 40 000 years appears to be a practical maximum age limit for comparisons of known SNe to observed terrestrial responses. There can now be assembled a list of candidate events that may have been sufficiently intense to have effected terrestrial environmental changes. Their Earth-incident γ fluences and ages provide a chronology of late Quaternary solar system exposure, and this can be compared to terrestrial records of relevant changes. For the latter objective, this paper emphasizes the cosmogenic and radiogenic isotope 14C as a sensitive indicator.
Methodology
Presently, any SNe causation for significant terrestrial changes during this time period is unproven. Some workers conclude that massive solar energetic particle events (SEPs) must instead be invoked to explain observed abrupt increases in atmospheric 14C production (Usoskin et al., Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltsov, Solanki and Wacker2013; Dee et al., Reference Dee, Pope, Miles, Manning and Miyake2016). This even though the most intense solar flare in history, the Carrington corona mass ejection, left no 14C traces (Jull et al., Reference Jull, Panyushkina, Molnár, Varga, Wacker, Brehm, Baisan, Salzer and Tegel2020). Also, in at least two cases, possible SNe causation for measured abrupt 14C changes has been dismissed because no possibly associated SNRs were thought to exist (Dee et al., Reference Dee, Pope, Miles, Manning and Miyake2016). Plausible SN candidates are now available for both (see text below and Zhu and Tian, Reference Zhu and Tian2013). This emphasizes the need for a comprehensive chronology of nearby SNe events and expected emission fluences during this time period.
There may be no reason to exclude either solar or SNe causation for changes in the detailed late Quaternary records of the cosmogenic isotopes: ‘It is clear that there are several types of rapid events in the Δ14C record’ (Jull et al., Reference Jull, Panyushkina, Molnár, Varga, Wacker, Brehm, Baisan, Salzer and Tegel2020). In this regard, predictive capability is an important criterion for the utility of any causal hypothesis. If secure linkages are established for even a few SNe/terrestrial response events, then, given the reservoir of dozens of known SNe, this causal understanding can help guide further work and including investigation of other predicted effects. This paper provides evidence for a hypothesis that offers some explanatory power and also utility for prediction. Thus, if: (1) an SN is known to have occurred, within uncertainty, at a particular time; (2) its emitted energy was likely of sufficient size and spectral characteristics to have caused a predicted terrestrial response, and (3) there is documentation in terrestrial records at that time of such a response, then the hypothesis of SN causation is viable and must be retained to guide further work.
This paper: (1) inventories all nearby SNRs that are <40 000 years in age, (2) compiles the relevant observational statistics including distances and ages, (3) constrains the probable emission energies from recent SN theory, modelling and observational results, and (4) compares the resulting solar system radiation history to relevant terrestrial records that could record the predicted changes.
Locating nearby supernovae
Three comprehensive radio and high-energy catalogues of SNRs (Safi-Harb et al., Reference Safi-Harb, Ferrand and Matheson2012; Pavlovic et al., Reference Pavlovic, Dobardzic, Vukotic and Urosevic2014; Green, Reference Green2019) were interrogated to identify objects <5 × 104 years in age and <1.5 kpc in distance, providing 18 objects (Table 1). Although some ages for radio SNRs are based on radio surface brightness/remnant diameter (Σ–D) and D-age relations, these are calibrated using measured radial expansion velocities and other methods. Their precision is known to be low (±25%) (Pavlovic et al., Reference Pavlovic, Dobardzic, Vukotic and Urosevic2014). Thus, these were used only when other more-direct estimates were not available. Where uncertainty values are published, they are included in Table 1; otherwise an uncertainty of ±25% (standard error) is assumed. All uncertainties are carried through to incident fluence (energy/unit area) and also 14C production calculations (see following sections).
Table 1. Distances, Earth-incident γ (using 4 × 1049 erg total γ SN emission), ages, predicted 14C production and measured 14C rise (reported as + Δ14C) within the time intervals for 18 of the closest SNe
14 C production is based on 130 atoms per SN-generated γ erg. The γ fluence is divided by 4 to adjust the distance-corrected total SN γ/cm2 to a spherical Earth. The energies and production ranges are based on the distance uncertainties. The Δ14C results use the higher temporal resolution data when available, as described in the text. Ages are years before present (BP, before CE 1950). All errors are expressed as standard errors; a ±25% error is assumed if not otherwise provided.
There are also approximately 20 poorly-constrained SNR compact objects in the high-energy compilation (Safi-Harb et al., Reference Safi-Harb, Ferrand and Matheson2012) for which no distances or ages are available. These were excluded for the purpose of this paper and because it is unlikely that any of these are close, relatively young and of importance to the present analysis. For convenient reference, the common names of the SNRs are also provided in Table 2.
Table 1, as described further below, adopts 4 × 1049 erg as a reasonable ‘typical’ total γ (mainly 70 MeV to 10 GeV): over 3 years, including prompt relativistic shock breakouts, any intercepted jetted emission and sustained emission. This is for comparison of SNe events of varying distances. However, intrinsic intensities, duration and γ cross-sections of Milky Way SNe may vary significantly. In this regard, the distances of the SNE of most interest vary from 0.25 to 1 kpc, which using the inverse square law provides a 16× luminosity factor if γ from each event was identical. The intrinsic energy uncertainties may be reduced in the future by more observation and analysis of the individual candidate SNR characteristics, such as calculated total explosion energies and progenitor star masses; a more detailed consideration is provided below.
The distances in Table 1 are based on a variety of observational methods: proper motions, shock and radial velocities, HI absorption and polarization, kinematic spectral line observations and association with star fields measured via parallax. Accuracies vary; for kinematic distances, the uncertainties may be <30%; for distances from X-ray fitting, they may be >50% (Zhu and Tian, Reference Zhu and Tian2013). Where uncertainties are not available for particular objects, a ±25% value is assumed and represents an approximate average of the published uncertainties (Table 1).
Available age estimates are partially dependent on measured or estimated distances. Those in Table 1 include ages from empirical SNR radio surface brightness/remnant diameter Σ–D and D–age relations, and also from measured radial expansion velocities; the latter may be more accurate. Where uncertainty values are published, they are included in Table 1. The most recent observational findings for the SNe are used in each case.
Constraints over typical SNe γ emission energies
Varying intrinsic emission energies may strongly affect the accuracy of the predicted terrestrial SN γ fluences provided in Table 1. Related research using the new satellite observatories as applied to extragalactic SNE is rapidly expanding and is critical to understanding terrestrial and solar system exposure.
SNe are diverse and created by different types of explosions (e.g. those associated with Type 1A white dwarf binaries and Type II massive or supermassive progenitor stars). Such diversity also occurs for their observed γ emissions and associated spectral cross-sections: although these can be remarkably uniform within some classes, allowing their use as ‘standard candles’ (Hamuy and Pinto, Reference Hamuy and Pinto2002; Kasen and Woosley, Reference Kasen and Woosley2009; Cano, Reference Cano2014). The γ evolution in the first hours to several years after the explosion is known to vary also; in particular, some events fade and then re-emerge to produce large amounts of γ radiation months to several years after the initial explosion (Wang et al., Reference Wang, Huang and Zhuo2019a).
Many decades prior to any direct γ observations, SNe theory for core-collapse events already predicted prompt X- and γ radiation. Peak luminosities of 1.9 × 1045 erg s−1 were calculated, and the total hard γ energy from SNe was estimated to vary between 1047 and 1050 erg, radiated over a period of months (Colgate, Reference Colgate1975; Klein and Chevalier, Reference Klein and Chevalier1978). Observational data for extragalactic objects later indicated that Type II SNe total explosion energies vary from 0.5 to 4.0 × 1051 erg (Kasen and Woosley, Reference Kasen and Woosley2009). Perhaps an average of 0.01 of such energy is emitted as γ (Miyake et al., Reference Miyake, Nagaya, Masuda and Nakamura2012). Varying SNe γ emissions depend on the mass and type of the progenitor star, metallicity, rotation velocity, the type of event and whether a binary system was involved (Kann et al., Reference Kann, Schady, Olivares, Klose, Rossi, Perley, Krühler, Greiner, Nicuesa Guelbenzu, Elliott, Knust, Filgas, Pian, Mazzali, Fynbo, Leloudas, Afonso, Delvaux, Graham, Rau, Schmidl, Schulze, Tanga, Updike and Varela2019).
Verified γ-emitting SNe now include: Type Ia white dwarf explosions (Churazov et al., Reference Churazov, Sunyaev, Isern, Bikmaev, Bravo and Al2015), Type Ib, Ic and II massive star core-collapse events (Podsiadlowski, Reference Podsiadlowski, Oswalt and Barstow2013), core-collapse hypernovae (Pian et al., Reference Pian, Mazzali and Starling2006), superluminous SNe (Moriya et al., Reference Moriya, Sorokina and Chevalier2018), many or most long GRBs (Gehrels and Mészáros, Reference Gehrels and Mészáros2012; Cano, Reference Cano2014; Cano et al., Reference Cano, Wang, Dai and Wu2017) and subluminous long GRBs (Nakar and Sari, Reference Nakar and Sari2012). Prompt isotropic relativistic shock breakout γ emissions of 1048 and 1048–50 erg from ‘standard’ long GRBs (including those with beamed emissions) and also from subluminous GRBs are predicted by theory and agree with some observational data (Nakar and Sari, Reference Nakar and Sari2012). Intense γ emission over weeks to up to several years has now been observed from Type II SNe and can be sustained by shock-induced emission from a circumstellar medium (Wang et al., Reference Wang, Huang and Zhuo2019a). Finally, observations in the γ domain of the nearby core-collapse SN 1987 in the large Magellanic Cloud tested some of the relevant theory for γ in the months after the initial explosion (Matz et al., Reference Matz, Share, Leising, Chupp, Vestrand, Purcell, Strickman and Reppin1988; Chevalier, Reference Chevalier1992).
During a SN, an initial shock breakout may produce either beamed or isotropic γ emission or both (non-beamed emission may occur after the brief collimated burst). Observations with γ and X-ray observatories and optical telescopes demonstrate that many long (10–300 s) GRBs are a special class of extra-galactic, supermassive star SN with beamed γ emission reaching isotropic-equivalent energies of 1053 erg (Gehrels and Mészáros, Reference Gehrels and Mészáros2012). At least one late Quaternary galactic SNR exhibits characteristics compatible with origin as a GRB (Lopez et al., Reference Lopez, Ramirez-RUIZ, Castro and Pearson2013). Also, XRFs (e.g. SN 2006j) produce prompt X-ray flashes. Such objects are half as luminous as some GRB-associated optical SNe; they attain total energies smaller than GRBs but greater than typical SNe, and may be isotropic radiators of both γ and X-rays (Pian et al., Reference Pian, Mazzali and Starling2006). Optically, some ‘superluminous SNe’ exhibit peak brightness approximately 10–100 times that of more common SNe (Moriya et al., Reference Moriya, Sorokina and Chevalier2018). Their γ emissions may be larger as well; their precursor stars may be exceptionally massive. Some of the SNRs in Table 1 could represent events like these, though they are infrequent. Finally, for Type Ia (binary white dwarf) SNe, sometimes used as ‘standard candles’, observation of an extragalactic example indicates γ luminosities of 11 ± 1 × 1041 erg s−1 on day 73 and 6.5 ± 0.6 × 1041 erg s−1 on day 96 (Churazov et al., Reference Churazov, Sunyaev, Isern, Bikmaev, Bravo and Al2015). A year of such emission would provide a total γ close to 1049 erg (depending on the size of the earliest emission).
Theory, modelling and observation using the orbital high-energy observatories are also helping constrain the time evolution of SNe γ emission. In part, the work is designed to facilitate early SNe detection. A significant fraction of the photon energy of a Type Ia SN emerges in hard X-rays and γ lines; total explosion energies exceed 1051 erg. The ejecta mass and expansion rate of Type Ia differ greatly from core-collapse SN, creating an early and strong signal in γ (Wang et al., Reference Wang, Fields and Lien2019b). Depending on models used, the peak γ flux occurs ~60–90 days after explosion. For Type II events, γ depends on the level of the two-photon annihilation process (Cristofari et al., Reference Cristofari, Renaud, Marcowith, Dwarkadas and Tatischeff2020); multi-GeV and some TeV γ are expected in the first several days, followed by an interval of obscuration and no detection, and then re-emerging during a period of intense emission from ejecta–wind interaction (Wang et al., Reference Wang, Huang and Zhuo2019a). Hard γ is emitted as the shock wave interacts with soft photons from the SN photosphere through pair production, thereby temporarily suppressing the γ leaving the system. In the case of SN 1993J, γ attenuation was calculated at 10 orders of magnitude in the first few days after the SN explosion: γ would be detectible if observations are performed either earlier than 1 day, or later than 10 days after the explosion, when γ attenuation decreases to about 2 orders of magnitude (Cristofari et al., Reference Cristofari, Renaud, Marcowith, Dwarkadas and Tatischeff2020).
Total explosion energies, chemistry/spectra, masses and existence or absence of associated compact objects such as pulsars are obtained from SNR observations to constrain the probable precursor star characteristics and the size, type and physics of the explosion. Many GRBs and superluminous SNe may be hypernovae and produce black holes (Podsiadlowski, Reference Podsiadlowski, Oswalt and Barstow2013): unusually energetic core-collapse SNe with supermassive star progenitors. Prompt emission in γ may be from successful or failed shock breakout (Nakar and Sari, Reference Nakar and Sari2010), some days prior to initiation of the optical event. Then, as the explosion evolves, γ radiation again emerges, and is sustained over a period of several years dependent on characteristics of the expanding shell (Matz et al., Reference Matz, Share, Leising, Chupp, Vestrand, Purcell, Strickman and Reppin1988).
The spectra for SN 1987A, the well-observed and relatively nearby core-collapse SN in the Large Magellanic Cloud, provides an example of sustained emission. Hard photon emission was observed between 0.02 and 2 MeV over 500 days; the measured total γ energy was 1046 erg (any initial emission was not monitored), and total SN energy was 1.4 × 1051 erg (Pinto and Woosley, Reference Pinto and Woosley1988; Chevalier, Reference Chevalier1992). However, the progenitor for 1987A was a blue supergiant with an initial mass of about 20 M⊙ instead of the more typical red supergiant, and the SN was fainter optically than typical Type II SNe at maximum by an order of magnitude (Chevalier, Reference Chevalier1992).
Modelling of possible terrestrial γ effects from a more typical event (Gehrels et al., Reference Gehrels, Laird, Jackman, Cannizzo and Mattson2003) used 1047 erg γ total, a spectral distribution binned into 66 logarithmic intervals 0.001–10 MeV, and a red supergiant progenitor of 15 M⊙. For comparison, the Vela X Y Z SN (Table 1) progenitor was 30 M⊙ (Sushch and Hnatyk, Reference Sushch and Hnatyk2014). In the modelling, the Type II SN γ luminosity peaks at 340 days and is within a factor of 10 of the peak for 500 days. Also, γ emission in Gev energies has recently been observed in other Type II SNe and with a different evolution through time. Thus, between 0.2 and 500 GeV, over a 3 years emission time, the total energy released in γ-rays was 1051 erg, and the γ commenced 300 days after the explosion and lasted for another 850 days. This was a luminous but not particularly superluminous SN (Yuan et al., Reference Yuan, Liao, Xin, Li, Fan, Zhang, Hu and Bi2018).
In summary, very large quantities of γ output from SNe were earlier predicted and are now firmly established from advanced modelling and direct observation. There is exceptional variability among the different classes of objects, and theory development is underway to explain the new data. Spectra have been obtained for SN objects from different classes and the flux trajectories, over timescales from seconds to several years, are increasingly being observed. This research allows an increasingly accurate evaluation of the implications for astrobiology. Some of the predicted terrestrial effects (e.g. on cosmogenic isotopes, atmospheric photochemistry and climate) will depend on the energy spectra and duration, but a total fluence estimate, even without such relevant information, remains useful (Table 1 and Fig. 1). At the temporal resolution of most terrestrial records (yearly to decadal, at best), the SN events are short, and it is the cumulative geochemical and geophysical results over each event that may be observable.
Fig. 1. Top: Intensity of SNe γ (total fluence incident on Earth) for 18 of the closest events. Bottom: Expanded view for the past 12 000 years. Data from Table 1, this paper. Ages are years before present (BP, before CE 1950).
Potential solar system effects
The terrestrial effects of SNe radiation may include the interactions of γ at high energies with the upper atmosphere, Compton scattering creating a γ-initiated ionization cascade similar to cosmic ray protons and re-emission of γ at lower energies (including within the troposphere at energies where cosmogenic isotopes are generated), ionization of atmospheric gasses, and altered atmospheric and biological processes: as the energy is re-radiated (Scalo and Wheeler, Reference Scalo and Wheeler2002). Like particle cosmic rays, SNe hard photon effects on solar system planetary bodies are mediated by any atmosphere, which shields the surfaces, but absorbs, scatters and re-emits the radiation (Scalo and Wheeler, Reference Scalo and Wheeler2002; Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013).
In this regard, relatively steady levels of radioactive 14C are maintained in Earth's upper atmosphere by 14N(n,p)14C from incoming galactic and solar cosmic ray particles. However, γ photons from SNe can also produce 14C, 10Be and 36Cl, ionize N by photonuclear reactions, and initiate neutron cascades (Lingenfelter and Ramaty, Reference Lingenfelter, Ramaty and Olsson1970; Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Miyake et al., Reference Miyake, Nagaya, Masuda and Nakamura2012). Thermalized neutron yields from γ photons reach a maximum at about 23 MeV (from absorption around the giant dipole resonance for N and O nuclei), then rises again at MeV > 60 (Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013). In contrast, 10Be and 36Cl maximum production could occur after additional progress of the energy through the atmosphere and γ emission at lower energies. Also, O3, an important greenhouse gas and solar UV shield, may be depleted by the ionizing radiation, and catalytic reactions producing NOx species initiated (Ruderman, Reference Ruderman1974; Gehrels et al., Reference Gehrels, Laird, Jackman, Cannizzo and Mattson2003; Thomas et al., Reference Thomas, Melott, Jackman, Laird, Medvedev, Stolarski, Gehrels, Cannizzo, Hogan and Ejzak2005).
The terrestrial 14C record provides perhaps the best opportunity to test for SN effects. Since its development in the middle part of the last century, radiocarbon dating has proceeded based on the relatively constant production of 14C in Earth's atmosphere by cosmic rays: as calibrated using 14C assays on wood cellulose securely dated to actual year BP by dendrochronology (Stuiver, Reference Stuiver1961). Radiocarbon production is, however, modulated by other factors, including the Earth's variable magnetic field and the Sun. There are short-lived ‘wiggles’ in the 14C production rate and atmospheric concentration: most are attributed to solar cycle variations (Eastoe et al., Reference Eastoe, Tucek and Touchan2019). However: ‘there are not many factors’ that can induce abrupt elevations (sharp rises within 1–2 years) of 14C (Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013). Such rapid-increase anomalies in the Earth's atmospheric 14C production have been observed in 14C assays of individual tree ring records, although some actually extend over several years (Table 3). These are commonly attributed to (possibly) photon radiation from SNe or GRBs, or to cosmic ray particle radiation from exceptionally large solar flares or extended periods of high solar activity (Miyake et al., Reference Miyake, Nagaya, Masuda and Nakamura2012; Jull et al., Reference Jull, Panyushkina, Miyake, Masuda, Nakamura, Mitsutani, Lange, Cruz, Baisan, Janovics, Varga and Molnár2018). The viability of the first hypothesis is now further considered.
Table 3. Rapid 14C increases (>7‰ Δ14C rise within 5 years, decline to near the previous level within 20 years) so far identified by various workers from tree rings sampled at high resolution (2 years or higher)
The historical CE 1006 example
Terrestrial radiocarbon production from SN γ in the 70 MeV to 10 GeV range is, in principle, measurable: for the historical SN 1006, if SN γ energies reach to 1050 erg (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995). Some prehistoric SN are much closer (Table 1). In order to constrain SNe γ requirements for 14C production and its recording in terrestrial records, the relation between the observed 14C changes (e.g. in tree ring wood), commonly expressed as Δ14C (Stuiver and Polach, Reference Stuiver and Polach1977), and 14C production in the atmosphere must be modelled. Carbon cycle modelling, including the pathways of atmospheric 14C and its incorporation into terrestrial records, is increasingly comprehensive (Kanu et al., Reference Kanu, Comfort, Guilderson, Cameron-Smith, Bergmann, Atlas, Schauffler and Boering2016). However, the purpose here is restricted to the identification of important candidate SNe that may have affected global production. Therefore, the published results of relatively simple four- or five-box models can be used to compare SN-predicted increases in 14C production with observed biosphere Δ14C changes (Table 1).
Note that, unlike particle cosmic radiation, Earth-incident γ is not affected by the geomagnetic field, and also that much of any SN 14C (and several other cosmogenic isotopes) should be produced in the troposphere (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995). The complex atmospheric changes that may be initiated could also themselves affect the resulting 14C record (Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013): for example, how quickly 14C is mixed between the northern and southern hemispheres, even under steady-state conditions, is still being investigated. As well, the tree ring chronologies themselves commonly exhibit regional offsets by <10 years, for reasons which may include dating/laboratory error but could also reflect actual regional differences whose causes are not yet understood (Pearson et al., Reference Pearson, Wacker, Bayliss, Brown, Salzer, Brewer, Bollhalder, Boswijk and Hodgins2020).
The possibility of SNe affecting Earth's 14C production has been investigated intermittently for several decades. Most of the relatively nearby objects in Table 1 have not been considered; instead, the historical SNe have been the focus (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Dee et al., Reference Dee, Pope, Miles, Manning and Miyake2016). For example: in California tree ring records, a 5–9.5‰ Δ14C rapid-increase tree ring anomaly commences at 942 BP and is followed by gradual decay over a decade (Tables 1 and 3). This is 2 years after the historic SN 1006 (Lingenfelter and Ramaty, Reference Lingenfelter, Ramaty and Olsson1970; Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995). For SN 1006, the authors used a distance of 1.3 kpc to calculate Earth-incident γ, and an intrinsic energy of 1 × 1049 erg in γ > 10 Mev. This produces 1.4 × 104 erg cm−2 at Earth and yields approximately 650 thermalized neutrons per erg available to produce 14C by 14N(n,p)14C (Lingenfelter and Ramaty, Reference Lingenfelter, Ramaty and Olsson1970). Thus, 0.9 × 107 thermal neutrons generate the SN-related 14C (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995). If this arrives in 1 year, the 14C production is 0.3 a cm−2 s−1, as compared to the annual steady-state production by cosmic rays of 1.64 a cm−2 s−1 (Kovaltsov et al., Reference Kovaltsov, Mishev and Usoskin2013).
The observed tree ring anomaly, which decays over 9 years, was fit via carbon cycle box modelling to a 1 year only, 2.5× increased 14C production rate (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995). A subsequent tree ring search for the same 14C anomaly from another geographic location (Japan) was successful; a +5‰ Δ14C increase was measured, which is more compatible with causation from this relatively distant SN (Menjo et al., Reference Menjo, Miyahara, Kuwana, Masuda, Muraki and Nakamura2005): the predicted effect on 14C is small. Although this result was not replicated at a location in England (Dee et al., Reference Dee, Pope, Miles, Manning and Miyake2016), recent remeasurements of the California wood verify the earlier results; also other, smaller (<5‰) positive (step-wise) anomalies were observed at CE 1063, 1167, 1297, 1429 and 1448–1450 (Eastoe et al., Reference Eastoe, Tucek and Touchan2019).
The distance of the SN 1006 SNR (G327.6+14.5, pulsar PKS 1459-41) has since been revised to 1.56 kpc (Jiang and Zhao, Reference Jiang and Zhao2007). Also, a lower, 20–55 a erg−1 mean yield (adjusted to a spherical Earth target) is used by Pavlov et al. (Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013) for γ entering the atmosphere from a hypothetical GRB with typical spectral parameters. Other recent studies use a production rate of 130 a erg−1 (Usoskin et al., Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltsov, Solanki and Wacker2013) for γ-related production, once divided by 4 this is ~32 a erg−1. This value is used in Table 1 for all events for uniform comparison purposes. For SN 1006, using a 4 × 1049 γ energy, the 130 a erg−1 production rate, the reduction by ¼ to adjust to a spherical Earth and the revised distance, the predicted changes in 14C again produce a small anomaly for SN 1006 (Table 1). The calculated extra production of 0.1–0.3 a cm−2 s−1 is still much less than the modelled need for 2.1 a cm−2 s−1 for the 9‰ Δ14C, and also less than that needed for the 5‰ Δ14C increase measured at the second site (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020). Damon et al. (Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995) address this problem by inferring that the SN must have emitted 1050 erg in γ, which does fall within the known range of variability in SN γ emissions (see above).
In regard to the slightly lagged response of tree ring 14C to this historic SNe, ring-based 14C concentrations reflect tropospheric conditions during wood formation, and some time lag is expected; such lags are variously accommodated by different box models (Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013). However, the lag, in this case, is also in agreement with a new understanding about the potential duration of SN γ emission. SN 1006 is considered to be a Type 1a white dwarf event, which re-brightened (Jiang and Zhao, Reference Jiang and Zhao2007); not all of the relevant γ was likely to have been produced in 1 year. The observed 14C anomalies at 942 BP, 2 years later (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Menjo et al., Reference Menjo, Miyahara, Kuwana, Masuda, Muraki and Nakamura2005), fit what is now known about the probable evolution of γ radiation from an SN of this type.
Comparison of prehistoric SNe γ to the 14C record
At a distance of 1.56 kpc, SN 1006 is more distant than many other known SNe during late Quaternary time. Previous examinations of the capability of SNe to affect Earth's atmosphere have mainly focused on the historical SNe; none of these were exceptionally close to the Earth. However, there are 18 SNe with distance estimates ≤1.5 kpc and ages less than 5 × 104 BP (Table 1, Fig. 1). One is as close as ~0.25 kpc.
The IntCal13 (Reimer, Reference Reimer2013) and separately reported records with much better temporal resolution can be used to examine the case for predicted effects. Dendrochronologically-dated tree rings provide the assayed carbon for the younger, <12 500 BP, part of IntCal13. Other materials sample 14C in the upper mixed ocean (marine corals and foraminifera), soil water (speleothems) or lake biota (Southon et al., Reference Southon, Noronha, Cheng, Edwards and Wang2012). The complete IntCal13 temporal coverage is 50 000 BP to present, with much loss of temporal resolution and attenuation of any brief pulses of atmospheric 14C in the older part of coverage. Thus, the temporal sampling is 20 years from 26 000 to 15 020 BP, 10 years from 15 000 to 12 500 BP, and 5 years from 12 495 to 0 BP (Reimer, Reference Reimer2013). IntCal13 may not detect short-term variations lasting less than 500 years earlier than 15 000 BP at all, unless they are exceptionally large. IntCal13 smooths and attenuates any abrupt changes, even if global, and even in the youngest portions of its coverage. It is used here only for a common comparison of different possible events and more detailed information is provided where available.
With these constraints, the closest eight late Quaternary SNe are evaluated individually, in order of distance, compared to the 14C record (see also Tables 1 and 3). The more temporally detailed assays are also used. The ages provided for each event are those of the initiation of the 14C anomaly and are in calendar years BP. The SNR's catalogue number and name are in each case paired with the possible 14C anomaly, by age.
Example 1: G263.9-03.3, Vela X Y Z and 12 740 BP
The nearest of late Quaternary SNe, the older Vela SN core collapse of a 30 M⊙ star occurred (within uncertainty) at the same time as a large and sudden positive global 14C anomaly in the IntCal13 record (Tables 1 and 3, Fig. 2(a)). Vela's distance of 0.25 ± 0.03 kpc is from Ca II absorption line spectra (Cha et al., Reference Cha, Sembach and Danks1999) and there is an independent distance estimate of 0.29 + 0.019, −0.017 kpc from VLBI parallax measurements on the associated pulsar (Dodson et al., Reference Dodson, Legge, Reynolds and Mcculloch2003). Caraveo et al. (Reference Caraveo, De Luca, Mignani and Bignami2011) also obtain a distance of 0.29 + 0.076, −0.050 kpc from Hubble parallax measurements. The age is estimated as 14 500 ± 1500 years from shock velocity considerations (Wallerstein and Silk, Reference Wallerstein and Silk1971; Cha et al., Reference Cha, Sembach and Danks1999); the pulsar characteristic age is 11 400 years (these are commonly considered minimum ages). The hydrodynamical age from modelling of the radio SNR is 7000–12 000 years (Sushch and Hnatyk, Reference Sushch and Hnatyk2014). Vela's proximity indicates that a 14C isotope signal should be detectible.
Fig. 2. Radiocarbon variation at the times of nearby prehistoric SNe. (a) A steep 25.3‰ rise in Δ14C occurs in IntCal13 at 12 745–12 640 BP. (b) A 21‰ IntCal13 Δ14C rise occurs at 22 500 BP, but the coarse temporal resolution is inadequate to reveal brief anomalies. (c) IntCal13 shows a steep rise in Δ14C of 15.3‰ at 7440–7410 BP. Single-year tree ring data show a 20‰ rise from 7431 to 7421 (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017). (d) At least three rapid-increase 14C anomalies may be illustrated in this IntCal13 plot spanning 5500–2500 BP: one at 5340 BP, one at 4880 BP and one at 2765 BP. (e) At 2765–2735 BP, Δ14C rises by 11.4‰ in 30 years, at the approximate time of the Vela Jr. SN. (f) Between 10 255 and 10 220 BP, Δ14C rises by 12.2‰ at the approximate time of the Boomerang SN. (g) IntCal13 data for 15 000–13 000 are without detailed time resolution and can reveal no short-lived anomalies. (h) Floating tree-ring chronologies with closer temporal sampling, however, document a strong and short-lived Δ14C increase marked by younger radiocarbon dates just after 14 697 BP (Adolphia et al., Reference Adolphia, Muscheler, Friedrich, Güttlere, Talamo and Kromerc2017) and compatible with the age and probable γ intensity of the Cygnus Loop SNR.
In floating chronology tree ring records with close-interval sampling, a +30‰ Δ14C increase occurs within 60 years starting at 12 740 BP, and reaches +45‰ after another 40 years (Hua and Al, Reference Hua and Al2009) (their Fig. 7; not illustrated here). The (smoothed) IntCal13 curve instead provides a +25.3‰ Δ14C from 12 745 to 12 640 BP and increasing another 5.4‰ to 12 515 BP (Fig. 1(a)). These latter results are based on decadal samples. Carbon cycle considerations and lags in cross-hemisphere atmospheric mixing suggest that several years would be required for full incorporation of the pulse into the global atmosphere and thence into 14C-recording materials such as tree rings. Also, for a nearby event, the predicted sustained emission several years after the explosion may be significant. After any sharp increase, a decay period is expected as the excess 14C is assimilated and merges with the normal annual production values for that time interval.
The timing of the Δ14C increase is synchronous with abrupt terrestrial climatic changes at the onset of the Younger Dryas Stadial: an interval of sharply cooler temperatures, especially at temperate to high northern latitudes (Hughen et al., Reference Hughen, Southon, Lehman and Overpeck2000). However, a system-internal mechanism has been suggested for the Younger Dryas increase: temporary cessation of North Atlantic Ocean deep/shallow water circulation, which can possibly explain the elevated 14C. However, then a triggering cause must still be located (Siegenthaler et al., Reference Siegenthaler, Heimann and Oeschger1980). The rapid changes starting at this time agree with predictive modelling of cooler temperatures and other atmospheric changes possibly caused by a γ radiation event (Thomas et al., Reference Thomas, Melott, Jackman, Laird, Medvedev, Stolarski, Gehrels, Cannizzo, Hogan and Ejzak2005). Thus, sharply increased NOx-induced atmospheric opacity and reduction of O3, which is an important greenhouse gas, both favour cooler temperatures. These are numerically modelled to occur if the results of Melott et al. (Reference Melott, Thomas, Hogan, Ejzak and Jackman2005) are applied: the 108 erg of GRB kev γ at a distance of 2 kpc used for modelling by Melott for a major extinction-causing GRB (at the end of the Ordovician Period) is 6.4 × 107 erg for the 8× closer, non-GRB (4 × 1049 erg) Vela, and thus comparable to the predicted 1.1–1.7 × 106 erg for this actual SN (Table 1). As well, a major mammalian extinction did occur at the start of the Younger Dryas (Barnosky et al., Reference Barnosky, Koch, Faranec, Wing and Shabel2004; Faith and Surovell, Reference Faith and Surovell2009; Brakenridge, Reference Brakenridge2011).
Causation of the steep rise of 14C at the start of the Younger Dryas remains controversial (Olivier et al., Reference Olivier, Stocker and Muscheler2001); a variety of Earth system-internal changes occurred then, and oceanic circulation can clearly affect atmospheric 14C concentrations. However, the very rapid and large 14C increase, observed in both tree rings and varved marine sediments (Hughen et al., Reference Hughen, Southon, Lehman and Overpeck2000), has remained difficult to explain or model with only changes in oceanic circulation. From Renssen et al. (Reference Renssen, Van Geel, Van Der Plicht and Magny2000):
‘The second feature in the Cariaco basin Δ14C record not replicated by our model is the rapidity of the Δ4C increase at the onset of the Younger Dryas…If the rapid Δ14C increase at the onset of the Younger Dryas observed in the Cariaco basin record is a faithful reflection of a Δ14C change in the atmosphere at that time, the previous concern to explain the early Δ14C drawdown during the Younger Dryas should be substituted by a new concern to explain this increase.’
The SN-predicted increased global production of 14C, 4.4–7.1 a cm−2 s−1, is much higher than the average steady-state value of 1.64 a cm−2 s−1 from cosmic rays. Also, this was an exceptional Type II SN event, with a high precursor mass and a total explosion energy of 1.4 × 1051 erg (Sushch and Hnatyk, Reference Sushch and Hnatyk2014). The timing, proximity and energy of this prehistoric SN, when compared to the rapid increase, size, global extent and timing of the 14C anomaly, support a possible cause and effect relationship. If indeed climate was affected, at a time of continental deglaciation, then Earth system-internal changes, already underway, may also have been affected, and further perturbed the carbon cycle.
Example 2: G330.0+15.0, Lupus Loop and 22 500 BP
This SNR may be, within large uncertainty limits, nearly as close (0.32 kpc) as Vela X Y Z, but it is older and also of uncertain age: 23 000 ± 8000 BP (Table 1, Fig. 1) (Safi-Harb et al., Reference Safi-Harb, Ferrand and Matheson2012). The possible, though uncertain, proximity merits discussion here of its potential terrestrial effect.
In the IntCal13 record, greatly smoothed for this time period, there is a relatively steep 21‰ Δ14C increase over 140 years at 22 500 BP (Fig. 1(b)). Unlike the case for Vela, there is no close-interval 14C sampling available. A revised Σ–D relation estimates the SNR distance at 0.5 kpc (Pavlovic et al., Reference Pavlovic, Dobardzic, Vukotic and Urosevic2014); an age of 50 000 BP may also be consistent with the X-ray observations. Without more narrow SN age constraints and without detailed 14C sampling, no confirmation or falsification of the predicted (Table 1) 14C effects from this SN is possible. The smoothed Intcal13 data in Fig. 2(b) are, however, compatible with an SN signal of the expected size and at an appropriate time. Modulation by geomagnetic or solar activity changes is also possible because the changes may in reality indeed be as gradual as indicated in IntCal13.
Example 3: G114.3+00.3, S165 and 7431 BP
IntCal13 results show a steep rise of 15.3‰ Δ14C between 7440 and 7410 BP and are provided in Fig. 2(c), but detailed 14C assays for this period from Bristlecone Pine are also available with a 1–2 years resolution. They demonstrate a large increase (20‰) over 10 years, from 7431 to 7421 BP (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017). Recently, the distance of S165 was revised downward to ~0.7 kpc and with an age of approximately 7700 years. The total energy is unusually high and estimated at 5 × 1051 erg as a Type II event (Yar-Uyaniker et al., Reference Yar-Uyaniker, Uyaniker and Kothes2004). The distance is from associated patches of H I and H II emission (Safi-Harb et al., Reference Safi-Harb, Ferrand and Matheson2012), and there is also a central pulsar.
The rapid rise, magnitude and duration of the 14C anomaly is compatible with causation by this close and potentially powerful SN γ source. However, abnormal solar activity has instead been invoked (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017; Jull et al., Reference Jull, Panyushkina, Miyake, Masuda, Nakamura, Mitsutani, Lange, Cruz, Baisan, Janovics, Varga and Molnár2018). Yet, S165 appears to be of appropriate age and distance to be associated with this terrestrial 14C anomaly. The estimated S165 elevation of the 14C production rate over 1 year is 0.3–2.8 a cm−2 s−1 (range due to the distance uncertainty), whereas box modelling and comparison to solar modulation effects of the 14C anomaly indicate a total 14C increase in production between 6.0 ± 2.4 and 10.5 ± 3.0 a cm−2 s−1 (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017). Predicted SN effects are significant compared to annual steady-state production, and should be detectible, but they are smaller than what the tree-ring assays and carbon cycle modelling indicate is required. Modelling has, however, not yet been performed to evaluate a ~3–4 years long receipt of SN γ, as compared to a single solar superflare or a prolonged period of higher solar activity (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017). Also, as noted, the emitted energy was likely higher than used in Table 1. SN causation for the observed anomaly is viable, given that a relatively nearby and high-energy candidate SN has been identified, and that its emissions may have been intense and prolonged over several years.
Example 4: G266.2-1.2, Vela Jr. and 2765 BP
The IntCal13 radiocarbon chronology shows a Δ14C rise of 20‰ (Fig. 2(d) and (e)) at 2765 BP. This anomaly has recently been further investigated by more detailed tree ring 14C data demonstrating a rapid rise of approximately 13‰ at 2765–2749 BP (Jull et al., Reference Jull, Panyushkina, Miyake, Masuda, Nakamura, Mitsutani, Lange, Cruz, Baisan, Janovics, Varga and Molnár2018). Vela Jr. is a shell-type SNR in the same line of sight as G263.9-03.3 Vela X Y Z. Its age is estimated at 3800 ± 1400 years and distance at 0.75 ± 0.25 kpc (Allen et al., Reference Allen, Chow, Delaney, Filipović, Houck, Pannuti and Stage2015). No associated pulsar or other compact object has so far been observed: this may indicate a supermassive star precursor.
The SN age is between 2400 and 5100 years if it is expanding into a uniform ambient medium; if it is instead expanding into the material shed by a steady stellar wind, then the age may be as much as 50% older. Thus, SN causation for the brief, rapid-increase 14C anomaly at 2765 BP is plausible, but without a tighter age constraint (Table 1, Fig. 1), the correlation in time is not secure. The 14C anomaly exhibits a relatively slow rise (10 years) (Table 3): compatible with Vela Jr. if it was accompanied by extended γ emission, but other causation is inferred (Jull et al., Reference Jull, Panyushkina, Molnár, Varga, Wacker, Brehm, Baisan, Salzer and Tegel2020). Vela Jr., given its probable energy and proximity, likely did leave a trace in the terrestrial 14C record; better understanding of its age can test the possible specific connection to 14C changes.
Example 5: G160.9+02.6, HB9 and 5340 BP
IntCal13 results indicate a +18‰ 14C anomaly commencing at 5340 BP (Table 1 and Fig. 2(d)), but a rise that is not as steep as others, to ~5280 BP. However, a detailed tree ring study (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2017) concludes an abrupt rise (+9‰ in 1 year) in 5372 BP with a decay period of about 10 years. This result, from a floating tree ring chronology and buried logs, has not been validated by other workers (Jull et al., Reference Jull, Panyushkina, Molnár, Varga, Wacker, Brehm, Baisan, Salzer and Tegel2020; Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020); a dating error may be involved. HB9 is a radio SNR with an associated magnetar/pulsar compact object. The age is estimated at 4000–7000 BP based on the Sedov equation and evaporative cloud modelling (Leahy and Tian, Reference Leahy and Tian2007). It is nearly as close as Vela Jr., at 0.80 ± 0.40 kpc. HB9 may be of appropriate distance and age to be compatible with this brief 14C anomaly; its emission energy was also likely to have been unusually high.
Example 6: G106.3+02.7, Boomerang and 10 255 BP
At this time, a 12‰ rise in Δ14C occurs within 35 years in the IntCal13 results (Fig. 2(f)). The steep rise is followed by a more gradual but sustained rise to 10 145 BP (another 10‰), possibly of different causation. The Boomerang SNR is ~10 000 years in age and is at a distance of ~0.8 kpc. Table 1 provides another SN, G89.0+4.7, with an appropriate age (9900 ± 5100 BP) but at a greater distance (1.25 ± 0.45 kpc). The larger distance implies a significantly smaller 14C effect. The Boomerang SN is more consistent with the measured anomaly at 10 255 BP (Table 1). Single-year tree ring analysis is needed to further constrain the characteristics of 14C through this time interval. As the case for the Lupus Loop example, the data presented here are compatible with SN causation, but better age and distance estimates for the SN are needed.
Example 7: G107.5-1.5 and 4880 BP
The IntCal13 record includes a brief, positive (17.6‰ rise in Δ14C) anomaly at 4880–4820 BP (Fig. 2(d)), but the existence of an appropriate rapid-rise anomaly cannot be determined from these data. However, the G107.5-1.5 SN is also close to the Earth, at 1.1 kpc, and may be of appropriate age: 4500 ± 1500 (Kothes, Reference Kothes2003). As for Boomerang and Lupus Loop, single-year tree ring analysis is needed to further constrain the characteristics of 14C through this time interval.
Example 8: G074.0-08.5, Cygnus Loop and 14 722 BP
The IntCal13 record bracketing this time reveals no brief anomalies (Fig. 2(g)), but floating tree ring records document a relatively brief episode of much-increased atmospheric 14C concentration (Adolphia et al., Reference Adolphia, Muscheler, Friedrich, Güttlere, Talamo and Kromerc2017) (compare Fig. 2(g) and (h)). The Hulu Cave speleothem data from China also support a brief (10 years) atmospheric 14C excursion (Southon et al., Reference Southon, Noronha, Cheng, Edwards and Wang2012). The Δ14C increase occurs at the beginning of another short-lived but geographically extensive cold interval in climate history: the Older Dryas Stadial (Mangerud et al., Reference Mangerud, Briner, Goslar and Svendsen2017). The Cygnus Loop exhibits a radio and X-ray shell (Fesen et al., Reference Fesen, Neustadt, Black and Milisavljevic2018a) and its distance is now constrained to 0.74 ± 0.03 kpc (Fesen et al., Reference Fesen, Weil, Cisneros, Blair and Raymond2018b). If the true SN age is close to 15 000 BP, then the Cygnus Loop SN may have caused the brief but significant 14 722 years BP 14C anomaly (Table 1).
More recent SNe events
The other SNe listed in Table 1 are at distances of approximately 1–1.4 kpc, with the addition of the unusually bright SN 1006 at 1.6 kpc (for which a possible 14C signal has been recorded). These more-distant SNe may also be detectable by 14C assays with yearly temporal resolution if the γ emitted was sufficiently energetic. In this regard, two rapid-increase, global and short-lived 14C anomalies have been identified at 1176–1175 BP (CE 774–775) and 957–956 BP (CE 993–994) in single-ring tree ring assays (Miyake et al., Reference Miyake, Nagaya, Masuda and Nakamura2012; Miyake et al., Reference Miyake, Masuda and Nakamura2013; Table 3). New data covering the CE 966–1057 period suggest that the increase in atmospheric 14C previously associated with CE 994 actually occurred in CE 993 (Kudsk et al., Reference Kudsk, Philippsen, Baittinger and Fogtmann-Schulz2019) and this date is provided in Tables 1 and 3.
The shapes of the 14C time series for both events are similar: rapid increase within 1–2 years followed by a decade-long decay that could reflect dampening from the operation of the carbon cycle (Table 3). The magnitude of the younger 14C event is 0.6 of the older, suggesting, if intrinsic energies of associated SNe were similar, that the younger SN was ~1.3× more distant. Plausible candidate SNe would be G347.3-00.5 at 1.3 ± 0.4 kpc and approximately 1840 ± 260 years in age (Tsuji and Uchiyama, Reference Tsuji and Uchiyama2016), and the recently discovered and elongated G190.9-2.2 at 1.0 ± 0.3 kpc (Foster et al., Reference Foster, Cooper, Reich, Kothes and West2013) (Table 1). Both are significantly closer than SN 1006 at ~1.56 kpc.
The ~1 kpc distant G190.9-2.2 remnant is of a similar mean radius and thus age as W49B: an SNR at 8 kpc distance that may record a GRB (Lopez et al., Reference Lopez, Ramirez-RUIZ, Castro and Pearson2013). That event is described by Pavlov et al. (Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013) as having possibly produced the 1176 years BP 14C anomaly. For G190.9-2, and using the 130/4 14C a erg−1 production rate, a d = 1 kpc SN release of 4 × 1049 erg of γ causes a 1 year addition of 0.2–0.6 a cm−2 s−1 (Table 1), whereas box modelling of the needed additional 14C indicates 3.9 a cm−2 s−1 (Miyake et al., Reference Miyake, Masuda and Nakamura2013, Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013, Usoskin et al., Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltsov, Solanki and Wacker2013). If this SN was as close as 0.7 kpc and emitted somewhat more energy, then the predicted 14C production reaches to that estimated by the carbon cycle modelling and the tree ring data. Although SNe causation was ruled out by Pavlov et al. (Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013) because no apparent candidate SNR then existed, the SNR was newly reported and analysed the same year (Foster et al., Reference Foster, Cooper, Reich, Kothes and West2013).
The G190.9-2.2 SNR, like W49b, lacks a pulsar central object, is elongated, and its shape suggests the development of very energetic shock breakout γ accompanying a failed jet; or it may have resembled the SN1987A event. In either case, it appears that this recently discovered SN may be a stronger candidate than W49B as an emitter of radiation sufficient to cause the 14C anomaly. Also, a possible historical sighting occurred in CE 774: a ‘red cross in the sky’ in the Anglo-Saxon Chronicle (Allen, Reference Allen2012; Lovett, Reference Lovett2012). This is compatible with the SN's location in the northern hemisphere sky, and a new, bright, non-point source optical object agrees with the observed complex-ringed appearance of the expanding SN1987A remnant, in early stages of its evolution (Chevalier, Reference Chevalier1992). A bright auroral display is not recorded, although they are reported in other years near this date (Stephenson, Reference Stephenson2015). This recently discovered nearby SNR provides independent evidence of a bright SN at an appropriate time. It may also be compatible with ice core 10Be records through this time interval that are considered by others to be caused by an SEP (Sukhodolov et al., Reference Sukhodolov, Usoskin, Rozanov, Asvestari, Ball, Curran, Fischer, Kovaltsov, Miyake, Peter, Plummer, Schmutz, Severi and Traversi2017). Modelling (Pavlov et al., Reference Pavlov, Vdovina, Vasilyev, Pavlov, Blinov, Ostryakov, Konstantinov and Volkov2013) indicates a higher production of 14C than 10Be from incoming SN γ if it is mainly in the Mev range; but this differential does not rule out significant 10Be generation.
G347.3-00.5 (Table 1) is a possible candidate for causation of the somewhat less intense CE 993 tree ring 14C anomaly. It also was a close SN, very energetic at 1051 erg (Tsuji and Uchiyama, Reference Tsuji and Uchiyama2016), and its age may be compatible with the anomaly date, given the uncertainties with distances and ages. In this regard, however, instead of any visible SN-like object, a survey of historical records from Ireland, Germany and Korea show several reports of possible auroral activity commencing on 26 December 992. For example, ‘On the night of the birth of St. Stephan, at the first cockcrow, light like the Sun shone from the North and many people said the Sun had risen. This continued for a whole hour. Afterwards, the sky was slightly reddened and returned to the normal color’. Similar but less detailed records occur from Ireland and Germany. These are compatible with the occurrence of ‘intense solar activity’ and a solar particle event at this time (Hayakawa et al., Reference Hayakawa, Tamazawa, Uchiyama, Ebihara, Miyahara, Kosaka, Iwahashi and Isobe2017).
These accounts may as well be compatible with SN causation. The location of G347.3-00.5 in Scorpius indicates that the SN would not have been visible at night from northern latitudes in winter except shortly before dawn. Its position was also considered incompatible with Chinese records of a possible SN at CE 393 (Fesen et al., Reference Fesen, Kremer, Patnaude and Milisavljevic2011); that association has been largely ruled out. Also, not only solar particle events can affect the Earth's ionosphere and excite red glow-producing O2 energy transitions. On 27 December 2004, a giant γ-ray flare from SGR 1806-20 (a magnetar; a special class of neutron star) at 12–15 kpc distance created a massive disturbance in the Earth's daytime lower ionosphere. The radiation (~380 s) dramatically increased ionization in the lower ionosphere down to ~20 km altitude for >1 h (Inan et al., Reference Inan, Lehtinen, Moore, Hurley, Boggs, Smith and Fishman2007) and an X-ray afterglow occurred for 16 h. A γ-emitting SN in earliest CE 993 may have produced even stronger (and visible) upper atmospheric effects, including short-lived but major optical emissions from Compton scattering: creating a γ-initiated ionization cascade similar to cosmic ray protons (Scalo and Wheeler, Reference Scalo and Wheeler2002).
As the case for CE 774, it is premature to rule out SN causation of this 14C anomaly. The advantage of the SN hypothesis for both events is that the exceptionally energetic SNe certainly did occur and at approximately the right times. The disadvantage is that it is not yet possible to more tightly constrain their ages and the γ radiation and energy cross-sections received at Earth. In any case, it appears very likely that both rare and very energetic solar particle events and also significant SNe γ events affected the Earth during the past 40 000 years.
Discussion and conclusions
Changes in atmospheric cosmogenic isotope abundance on Earth can be used to monitor the exposure of the Earth and the rest of the solar system to high-intensity radiation events: from SNe, or from solar ‘super-flares’ or coronal mass ejection events. Determining actual causation is relevant to the knowledge of the Earth's radiation history, to the search for life within our own and in other solar systems, and to human space exploration. Yet the physical origins of the rapid-onset 14C anomalies, a type of evidence that appears to monitor radiation events, still remain uncertain (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020). Contrary to Usoskin et al. (Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltsov, Solanki and Wacker2013) and Dee et al. (Reference Dee, Pope, Miles, Manning and Miyake2016), the evidence presented here clearly indicates that SNe radiation events may have had a significant impact during the past 40 000 years.
This paper presents eight candidate SNe, and, in particular, four events where one or more of the predicted effects occur at the correct time on Earth. These are: Vela, at 12 740 years BP; S165 at 7431 BP; Vela Jr. at 2765 BP; and HB9 at 5340 BP. Their ages, distance and expected γ emissions are compatible with observed terrestrial effects that are also predicted from theory and modelling, and with the increasingly abundant and detailed observations of γ from extra-galactic SNe. However, given the variety of SNe objects and their associated released γ radiation, the distances only partly determine the expected terrestrial 14C signals. In this regard, however, the relative sizes of the 14C anomalies do generally agree with a causal connection: the closer SN events are broadly associated with the larger observed isotopic changes (Table 1).
There are significant differences in how three cosmogenic isotopes, 14C, 10Be and 36Cl, may respond to the two types of causality, and some workers have concluded, for particular events, that strong responses for all three favour a particle event. However, Sigl et al. (Reference Sigl, Winstrup and Al2015) found timing differences between the peaks, and Mekhaldi et al. (Reference Mekhaldi, Muscheler, Adolphi, Aldahan, Beer, McConnell, Possnert, Sigl, Svensson, Synal, Welten and Woodruff2015) adjusted 10Be and 36Cl ice-core records to match tree ring 14C peaks in timing, under the assumption that they were responses to the same event (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020). Such data cannot, therefore, yet be used to falsify SNe causation. The 10Be and 36C ice core results also lack the exceptionally tight temporal control needed for an unbiased comparison at these timescales (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020). Finally, from incoming γ energy, enhanced production of all three isotopes can occur as the incoming radiation penetrates the upper atmosphere and is re-emitted at energies relevant to their production (Damon et al., Reference Damon, Kaimei, Kocharov, Mikheeva and Peristykh1995; Scalo and Wheeler, Reference Scalo and Wheeler2002).
There is a reasonable expectation that both SNe radiation events and very large solar flares (or SEPs) have occurred in Earth history. Thus, in the yearly 10Be data covering the last 400 years, there are significant anomalies at about CE 1460, 1605, 1865 and 1890 in the North Greenland Ice Core Project (Wang et al., Reference Wang, Yu, Zou, Dai and Cheng2020); all are unrelated to strong 14C variations but may record flares. Whether the Sun can produce the very large particle events needed to produce the tree ring-recorded changes is still debated: Neuhäuser and Hambaryan (Reference Neuhäuser and Hambaryan2014) calculated the probability for one solar super-flare with energy larger than 1035 erg within 3000 years to be as low as 0.3–0.008. However, surveys of other solar-type stars indicate they may be possible (Dee et al., Reference Dee, Pope, Miles, Manning and Miyake2016). The inventory of possible SNe and terrestrial record linkages presented here provides additional information and can facilitate further testing of the utility of both hypotheses.
There are several approaches that could be employed in such additional work. These include:
(1) Statistical tests of the correlation in time of the SNe and the proposed matching 14C anomalies (perhaps limited to the past 14 000 years, where standard errors are the lowest);
(2) Quantitative examination of the ‘dose-response’ relationship, to identify the extent to which, assuming constant intrinsic γ emissions, the nearer SNe indeed caused stronger terrestrial isotope response;
(3) Further astrophysical observational and modelling studies focused on individual SNR ages, distances, type of event, and observed total and expected γ energies;
(4) Further observations and modelling of extragalactic SNe using the high-energy satellite observatories, in order to better constrain the γ energy cross-sections and durations of ‘typical’ SNe of various types;
(5) Modelling of the atmospheric ionization cascades initiated by the γ emissions, to better constrain predicted line emissions for high-energy photons corresponding to 14C, 10Be and 36Cl production;
(6) Investigations using terrestrial geologic and paleoecological records of the other climate and atmospheric chemistry effects predicted, and especially for the nearest of the surveyed SNe.
In conclusion, the presented data and analysis do not rule out solar flare or solar activity hypotheses (Usoskin et al., Reference Usoskin, Kromer, Ludlow, Beer, Friedrich, Kovaltsov, Solanki and Wacker2013; Jull et al., Reference Jull, Panyushkina, Miyake, Masuda, Nakamura, Mitsutani, Lange, Cruz, Baisan, Janovics, Varga and Molnár2018; Scifo et al., Reference Scifo, Kuitems, Neocleous, Pope, Miles, Jansma, Doeve, Smith, Miyake and Dee2019) for rapid-onset 14C increases. However, SNe causation instead is viable for many of them. SN γ energies adequate to have produced the 14C pulses were much earlier predicted from theory; they have now been directly observed for SNe outside of our Galaxy. The nearest in-galaxy events recorded by SNRs appear to have left traces in terrestrial isotopic and other paleoenvironmental records at the appropriate times and of the predicted relative magnitudes. The very nearest SNe, in particular, are the strongest candidates: for example, the older Vela SN, at ~0.25 kpc, with relatively tight age and distance constraints, a very massive precursor star and a steep rise in 14C at the appropriate time. If SNe causation of the cosmogenic isotope changes did not, however, occur, then the detailed isotopic records now being produced, from tree rings, ice cores, speleothems and other sources, may indeed provide information only about intense flares of solar cosmic radiation (Miyake et al., Reference Miyake, Jull, Panyushkinad, Wackere, Salzerd and Baisand2017) and/or other environmental changes. If SNe instead caused some or all of these events, then part of the solar system radiation hazard to life is from hard photons from nearby SNe, and some events may have produced recorded paleoecological changes. In this case, the hazard to life is susceptible to further testing and quantification by investigation of the array of specific SNe events that are known to have occurred within our Galaxy and relatively near the solar system.
Acknowledgement
All data sources are provided in the references cited. D. Green, R. Sternberg, M. Sorensen, M. Willmes and five anonymous reviewers provided useful comments on earlier versions of this paper. The work was partially supported by faculty research funds provided by the University of Colorado.
