ON A SOLAR ORIGIN FOR THE COSMOGENIC NUCLIDE EVENT OF 775 a.d.

2.1.1. High-energy Branch

Seventy-one GLEs have been observed since 1956 (Cliver et al. 1982; Smart & Shea 1991; Cliver 2006; Gopalswamy et al. 2013). The standard measure of GLE intensity is the percentage increase above background for a favorably located neutron monitor with a geomagnetic cut-off rigidity of ∼1 GV. Figure 1 is a histogram of GLE intensities. The 1956 February 23 GLE was the largest ever observed with a reported increase of 4554% versus 563% for the next biggest event.

Zoom In Zoom Out Reset image size

Figure 2 shows energy spectra obtained by Tylka & Dietrich (2009) for the nine GLEs from 1956 to the present with >1 GV proton (pr) fluence ⩾10⁶ pr cm⁻². The spectral form is a Band et al. (1993) function: a double-power-law form in energy that smoothly rolls one power-law into another. A proton energy of ∼430 MeV (1 GV) is the approximate minimum energy required for a primary proton to produce ¹⁰Be and ¹⁴C at Earth (Beer et al. 2012). The >1 GV fluence of the 1956 event (4.9 × 10⁷ pr cm⁻²; solid black line in Figure 2) is ∼6 times that of the next largest event (1960 November 12; 8.0 × 10⁶ pr cm⁻²) for which neutron monitor observations are available. The calculated ¹⁴C production of 2.9 × 10⁶ atoms cm⁻² (Usoskin & Kovaltsov 2012) for the 1956 GLE and the revised production of 1.3 × 10⁸ atoms cm⁻² obtained by Usoskin et al. (2013) for 775 a.d., implies that a ∼45 times multiple of the 1956 SPE spectrum (dashed black line in Figure 2) is required for the 775 event.

Zoom In Zoom Out Reset image size

2.1.2. Low-energy Branch

Because the 1956 SPE occurred before satellite observations of solar protons, the low-energy branch (<430 MeV) of the spectrum is uncertain. As noted by Reedy (1977), the primary source of low-energy data points for events prior to 1963 is Webber et al. (1963). The list therein of SPE fluences during cycle 19 (1954–1964)—based on an amalgamation of ground-based ionospheric data and rocket and balloon observations—is largely used by all subsequent researchers. Of the biggest events there is a significant difference only for 1960 November 12 event for which Feynman et al. (1990) and Shea & Smart (1990) list a >30 MeV fluence (F₃₀; >240 MV) value of 9.0 × 10⁹ pr cm⁻² (see Masley & Goedeke 1963), well above the 1.3 × 10⁹ pr cm⁻² value from Webber et al.

Usoskin et al. (2013) used their 45 times multiple of the 1956 February 23 spectrum and the Webber et al. (1963) value of 1.0 × 10⁹ pr cm⁻² at >30 MeV for this event to infer F₃₀ = 4.5 × 10¹⁰ pr cm⁻² for 775.

Reedy (1977) analyzed the concentration of the cosmogenic nuclides ²²Na and ⁵⁵Fe (both with a half-lives of ∼2.5 yr) in lunar rocks and determined that the Webber et al. (1963) F₃₀ estimate of 9.4 × 10⁹ pr cm⁻² (adjusted upward from 8.9 × 10¹⁰ pr cm⁻² by inclusion of small events by others) for cycle 19 was too low by a factor of ∼5. Subsequent additions of events (and a few changed F₃₀ values) common to the lists of Feynman et al. (1990) and Shea & Smart (1990) increase the Webber et al. F₃₀ value to 1.0 × 10¹⁰ pr cm⁻². The most recent lunar-rock-based estimate of F₃₀ for cycle 19 is 2.3 × 10¹⁰ pr cm⁻² (Reedy 2012). Comparison of a list of riometer-based >10 MeV SPE peak fluxes for cycle 19 from Bailey (1964) with those for common events in Webber et al. indicates a similar result; the median value of the ratios of the >10 MeV peak fluxes from Bailey to those of Webber et al. (1963) is ∼2.4 (cf. Reedy 1977).

Recently, Webber et al. (2007) revised the Webber et al. (1963) estimates of the three largest F₃₀ events of cycle 19, after considering a number of studies that had been published in the interim. The F₃₀ fluences for the three events were adjusted upward by factors of ∼1.3–3.0 as follows: 1956 February 23, 1.0 ⇒ 1.8 × 10⁹ pr cm⁻², 1959 July 10–17, 3.2 ⇒ 4.0 × 10⁹ pr cm⁻², and 1960 November 12–15, 2.0 ⇒ 6.0 × 10⁹ pr cm⁻².

Of primary interest here is the revised F₃₀ value for 1956 February 23. It increases the F₃₀ estimate for the 775 event from ∼4.5 × 10¹⁰ to ∼8 × 10¹⁰ pr cm⁻². We regard this as a conservative estimate because the 1.8 multiplier used by Webber et al. falls below both the ∼2.0 ratio of the Bailey (1964) to Webber et al. (1963) peak >10 MeV fluxes for this event and the 2.3 ratio of the solar cycle 19 F₃₀ value from Reedy (2012) to that of Webber et al. (1963). As can be seen in Figure 3, the red point corresponding to the revised F₃₀ value of ∼8 × 10¹⁰ pr cm⁻² for 775 falls outside the 90% confidence interval (Gehrels 1986) of the occurrence probability distribution function (OPDF) recently constructed by Kovaltsov & Usoskin (2014) for >30 MeV SPEs. This OPDF is based on direct satellite measurements of SPEs during the space age, measurements of sharp increases of ¹⁴C and ¹⁰Be concentration in tree rings and ice cores during the last 11,400 yr, and cosmogenic nuclides in lunar rocks during the last half-million years.

Zoom In Zoom Out Reset image size

Figure 4 shows the sunspot drawing from Sacramento Peak Observatory for 1956 February 17, when Greenwich active region 17351 was near disk center (N22W02) with a corrected (for foreshortening) spot area of 1393 millionths of a solar hemisphere (msh). It was part of a noteworthy string of six active regions spanning ∼60º of heliolatitude at ∼N20 with a combined area of 3809 msh. The GLE-associated flare from region 17351 on February 23 (N23W80) had its Hα maximum (class 3) at 03:42 UT. The sunspot number (SSN) for February 23 was 177. The 1956 SPE occurred during solar cycle 19, the largest cycle (peak 13 month smoothed SSN of 201.3) since sunspot recording began in 1610. Cycle 19 occurred during the strongest ∼50 yr interval of solar activity (∼1945–1995) in the last ∼2000 yr (Usoskin et al. 2007).

Zoom In Zoom Out Reset image size

In contrast to the 1956 SPE, the cosmogenic nuclide event of 775 occurred during a period of inferred low solar activity. Figure 5, adapted from Usoskin et al. (2007), contains a smoothed decadal record of solar activity based on ¹⁴C concentration in tree rings for the last 6000 yr. The interval, indicated by a black oval, in which the 775 event occurred has a peak ¹⁴C-based-SSN of only about 25 compared to ∼70 for the epoch in which the 1956 SPE occurred (red oval). While recent re-analysis of the SSN (Cliver et al. 2013) indicates that the reconstructed SSN in this figure is understated by ∼50% before ∼1885, the relatively quiet characterization of the period ca. 775 still applies. It has been suggested (McCracken et al. 2004) that large SPEs might occur preferentially during periods of low solar activity but the nitrate-in-ice-cores proxy for SPEs on which this picture was based has recently been invalidated by Wolff et al. (2012).

Usoskin et al. (2013) noted that the 775 event was accompanied by a cluster of low-latitude aurora reported from Shanxi Province, China, in 770 (twice), 773, and 775 based on the compilations of Keimatsu (1973) and Yau et al. (1995). Of these four events, Yau et al. (1995), who expunged many unreliable records of aurora given in the Keimatsu catalog, only list the two 770 events. From a recent re-examination of Keimatsu (1973), F. R. Stephenson (2013, private communication) notes that for the two 770 events, there is no mention that the phenomenon occurred at night, making it doubtful whether aurora are being referred to. He suggests, however, that a report of "more than 10 bands of white vapor" on January 12 776 may be auroral in nature. In either case, the evidence for a grouping of strong auroral activity ca. 775 is weakened.

During the satellite era, the largest F₃₀ events have been produced by sequences of eruptive flares originating near solar central meridian (Smart et al. 2006; Cliver & Dietrich 2013). Prominent examples of such sequences occurred in 1972 August (5.0 × 10⁹ pr cm⁻²; Shea & Smart 1990) and 1989 October (4.3 × 10⁹). Both of these episodes exhibited a strong shock spike in their F₃₀ time–intensity profile. Table 1 gives various F₃₀ values for similar composite events in cycle 19: 1959 July and 1960 November. In addition to the initial Webber et al. (1963) F₃₀ value, there are values that were: (1) obtained by scaling up the Webber et al. values by the ratios of the peak >10 MeV proton fluxes reported by Bailey (1964) and Webber et al. (1963), (2) scaled up from the Webber et al. (1963) values by the factor of 2.3 based on the analysis of cosmogenic nuclides in lunar rocks (Reedy 2012), and (3) determined by Webber et al. (2007). From (1)–(3) we take average values of ∼7 × 10⁹ pr cm⁻² and ∼5 × 10⁹ pr cm⁻² for 1959 July and 1960 November, respectively. Since the time resolution of tree ring records is 1 yr, we considered the strong activity from 1989 August–October, including five GLEs, as a single event to obtain a worst case scenario in modern times. The F₃₀ for the 3 month interval from 1989 August 12 to November 10 was 7.2 × 10⁹ pr cm⁻². Thus a F₃₀ = ∼8 × 10¹⁰ pr cm⁻² event for 775 a.d., based on a 45 times multiple of the updated (Webber et al. 2007) F₃₀ value for the 1956 event, would be ∼10 times larger than any event of the modern era.

Cliver & Dietrich (2013) obtained relationships, with broad scatter, between (1) F₃₀ and 1–8 Å soft X-ray (SXR) fluence and (2) flare bolometric energy and SXR fluence, that give a best guess estimate of total energy of ∼9 × 10³³ erg for a 775 flare versus ∼2 × 10³³ erg for the Carrington event. This estimate is based on the F₃₀ value of ∼8 × 10¹⁰ pr cm⁻² for the 775 event and the flare energy-partition work of Emslie et al. (2012) who found that, on average, mass ejection kinetic energy = ∼3 times flare bolometric energy. From Figure 5 in Veronig et al. (2002) which relates SXR fluence to intensity, we infer a flare classification of ∼X230 (versus ∼X45 for the 1859 event). Corresponding values for the 2003 November 4 flare are ∼10³³ erg and ∼X35 (Emslie et al. 2012; Cliver & Dietrich 2013). Emslie et al. (2012) estimated a free magnetic energy of 2.9 × 10³³ erg for NOAA region 486 (2610 msh on 2003 October 29). Scaling linearly by sunspot area (assuming an energy conversion efficiency of ∼0.5; Emslie et al. 2012) implies an active region for the 775 event with area of ∼16,000 msh, ∼2.5 times the size of the largest region yet recorded (6132 msh on 1947 April 8).