Volume 44, Issue 16 p. 8621-8628
Research Letter
Free Access

Cosmic ray event in 994 C.E. recorded in radiocarbon from Danish oak

A. Fogtmann-Schulz

Corresponding Author

A. Fogtmann-Schulz

Department of Geoscience, Aarhus University, Aarhus, Denmark

Correspondence to: A. Fogtmann-Schulz,

[email protected]

Search for more papers by this author
S. M. Østbø

S. M. Østbø

Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
S. G. B. Nielsen

S. G. B. Nielsen

Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
J. Olsen

J. Olsen

Department of Physics and Astronomy, Aarhus University, Aarhus, Denmark

Search for more papers by this author
C. Karoff

C. Karoff

Department of Geoscience, Aarhus University, Aarhus, Denmark

Stellar Astrophysics Centre, Department of Physics and Astronomy, Aarhus University, Aarhus, Denmark

Search for more papers by this author
M. F. Knudsen

M. F. Knudsen

Department of Geoscience, Aarhus University, Aarhus, Denmark

Search for more papers by this author
First published: 14 August 2017
Citations: 31

Abstract

We present measurements of radiocarbon in annual tree rings from the time period 980–1006 Common Era (C.E.), hereby covering the cosmic ray event in 994 C.E. The new radiocarbon record from Danish oak is based on both earlywood and latewood fractions of the tree rings, which makes it possible to study seasonal variations in 14C production. The measurements show a rapid increase of ∼10‰ from 993 to 994 C.E. in latewood, followed by a modest decline and relatively high values over the ensuing ∼10 years. This rapid increase occurs from 994 to 995 C.E. in earlywood, suggesting that the cosmic ray event most likely occurred during the period between April and June 994 C.E. Our new record from Danish oak shows strong agreement with existing Δ14C records from Japan, thus supporting the hypothesis that the 994 C.E. cosmic ray event was uniform throughout the Northern Hemisphere and therefore can be used as an astrochronological tie point to anchor floating chronologies of ancient history.

Key Points

  • New high-resolution radiocarbon record from Danish oak shows a 10‰ increase in Δ14C across cosmic ray event in 994 C.E.
  • The cosmic ray event in 994 C.E. likely took place between April and June in 994 C.E.
  • Atmospheric radiocarbon changes across the 994 C.E. cosmic ray event were uniform in the Northern Hemisphere

1 Introduction

Calibrated radiocarbon-based chronologies rarely provide estimates with a precision better than 50 years, due to ambiguities associated with the radiocarbon calibration curve (IntCal13) [Reimer et al., 2013]. This may lead to so-called floating chronologies, i.e., chronologies with high internal precision that are not firmly fixed on an absolute time scale. A prominent example is the Egyptian chronology, which is characterized by small relative uncertainties and large uncertainties associated with the absolute timing of events that exceed 50 years [Kitchen, 1991]. One approach to fix such floating chronologies on an absolute time scale is through the identification of distinct cosmic ray events. Miyake et al. [2012, 2013, 2014a] recently discovered two rapid increases in the 14C/12C ratio in Japanese cedar and hinoki cypress tree rings that took place in 775 C.E. and 994 C.E. These events are referred to as cosmic ray events or so-called Miyake events [Dee and Pope, 2016], and they are characterized by an increase in atmospheric Δ14C of 10‰ or more within a year. This abruptness represents a distinct feature that characterize such cosmic ray events, and the identification of such an event may therefore help anchor floating chronologies [Dee and Pope, 2016]. Wacker [2014] thus used the 775 C.E. event to absolutely date a wooden beam in the Holy Cross chapel in Val Müstair, Switzerland, hereby highlighting the great potential of using these cosmic ray events to obtain very precise dates in archeology.

The cosmic ray event in 775 C.E. has since been identified in other high-resolution radiocarbon records from trees, such as German oak [Usoskin et al., 2013], Siberian larch, North American bristlecone pine [Jull et al., 2014], New Zealand kauri [Güttler et al., 2015], and Polish oak [Rakowski et al., 2015], as well as corals [Liu et al., 2014]. The magnitude and timing of the 775 C.E. event is reasonably consistent among the different tree ring radiocarbon records from around the globe, indicating that the 775 C.E. event left a globally synchronous imprint on the atmospheric radiocarbon concentrations. The 775 C.E. event has also been observed as large positive anomalies in 10Be and 36Cl in ice cores from Greenland and Antarctica, attesting to the global imprint on the cosmogenic nuclide production of this event [Miyake et al., 2014b; Sigl et al., 2015; Mekhaldi et al., 2015]. So far, the cosmic ray event in 994 C.E. has only been identified in two radiocarbon records from Japan, but 10Be and 36Cl anomalies in ice core records from Greenland show that this event impacted the cosmogenic nuclide production in the entire Northern Hemisphere. The peak amplitudes in 10Be and 14C differ somewhat around 994 C.E. but nevertheless agree within the margin of the errors [Mekhaldi et al., 2015]. As the age models of the ice cores are subject to accumulating uncertainties associated with the annual layer counting, the time scales of the ice core records were adjusted in order to align the 10Be and 36Cl peaks with the 14C anomalies in tree rings around 775 and 994 C.E. [Sigl et al., 2015]. Despite the uncertainties associated with the absolute time scale of the ice cores, the cosmogenic nuclide records from the ice cores are instrumental for our understanding of the cosmic ray events, as they provide key insights concerning their origin.

The origin of the rapid increases in atmospheric radiocarbon concentration associated with the cosmic ray events has been widely discussed, primarily based on the event in 775 C.E. Due to the magnitude and, more importantly, rapidity of the increase in Δ14C (12‰ in 775 C.E. and 10‰ in 994 C.E.), it is not directly associated with the Schwabe solar cycle, which only produces atmospheric radiocarbon changes of 3–4‰ over an average of 11 years [Stuiver and Braziunas, 1993; Baroni et al., 2011]. Miyake et al. [2012] originally proposed two explanations: a supernova explosion or a solar proton event (SPE). Subsequently, Allen [2012] argued in favor of a supernova explosion, but since no remnant of such an explosion has been observed, it is not considered a very likely source of the cosmic ray events [Miyake et al., 2012]. Liu et al. [2014] suggested a cometary impact on the Earth's atmosphere as the explanation of the first cosmic ray event, but this is considered unlikely due to the size of the comet required to produce the amount of excess radiocarbon and an erroneous interpretation of a historical text describing a comet in 773 C.E. [Overholt and Melott, 2013; Usoskin and Kovaltsov, 2014; Chapman et al., 2014]. Other proposed explanations include a cometary impact on the Sun [Eichler and Mordecai, 2012] and a gamma ray burst (GRB) [Hambaryan and Neuhäuser, 2013; Pavlov et al., 2013], but such sources are now considered less likely, in part because the 775 C.E. event was globally uniform with respect to amplitude and timing, which most likely would not be the case if it was caused by a GRB. Also, GRBs would not produce the 10Be anomalies observed in ice cores from Greenland and Antarctica, and they are consequently an unlikely source of the cosmic ray events [Hambaryan and Neuhäuser, 2013; Pavlov et al., 2013]. Notably, early studies invoking sources outside the solar system were based on incorrect estimates of the strength of the 775 C.E. event [Usoskin et al., 2013]. More recent studies indicate that the cosmogenic nuclide records documenting the 775 C.E. cosmic ray event are consistent with a solar origin, suggesting that the event was caused by a very large SPE or a series of SPEs [Melott and Thomas, 2012; Usoskin et al., 2013; Thomas et al., 2013; Cliver et al., 2014; Güttler et al., 2015; Mekhaldi et al., 2015]. This interpretation is supported by historical chronicles reporting enhanced auroral sightings around this period [Usoskin et al., 2013]. Similar to the cosmic ray event in 775 C.E., the 994 C.E. event was most likely also caused by a very large SPE [Mekhaldi et al., 2015], but additional records are needed to increase our understanding of this event. Here we present a new subannual radiocarbon record based on earlywood and latewood samples from the Danish oak dendrochronology. The new record covers the period 980–1006 C.E., thereby bracketing the cosmic ray event in 994 C.E.

2 Tree Physiology

Most of the trees used to study the cosmic ray events are gymnosperms (also called softwoods or conifers), such as cedar, cypress, larch, pine, and kauri. Oak, on the other hand, used by Usoskin et al. [2013] and Rakowski et al. [2015], is a ring-porous angiosperm (also called hardwood or deciduous tree) [Ladefoged, 1952; Speer, 2010]. The structure of the wood differs between these two tree types, especially in the way the trees transport water from the roots to the leaves. The cell types forming gymnosperms are generally simpler than the cell types in angiosperms. Gymnosperms and ring-porous angiosperms all create annual rings if they grow in a seasonally changing climate. The annual rings consist of earlywood (EW) and latewood (LW), which are produced early and late in the growing season, respectively [Speer, 2010]. Studies of modern Δ14C values in earlywood and latewood fractions from gymnosperms, in particular, cedar [Yamada et al., 2008; Rakowski et al., 2010] and pine trees [Sheng et al., 2016], suggest that both fractions are consistent with the average atmospheric Δ14C value of the warm summer months (May–September) during the year of formation.

Ring-porous angiosperms differ from gymnosperms by the presence of large vessels that enable them to transport water up the tree more efficiently. Ring-porous angiosperms thus have a clear division between EW and LW, where the EW cells can be identified as a row of vessels that are formed before the leaves come out and take up carbon dioxide from the atmosphere. Because these vessels are formed early in the growth season, when photosynthesis have not yet been produced, the tree uses stored reserves from the previous growth season for the formation of earlywood [Speer, 2010]. Latewood is formed during the growth season using carbon absorbed from the atmosphere in the year of formation [Speer, 2010]. This difference between EW and LW in angiosperms is consistent with a study of Swedish oak [Olsson and Possnert, 1992] and can be interpreted as a memory effect, where the Δ14C value of the EW fraction reflects a combination of the Δ14C level from the previous year and the Δ14C level of the early-spring atmosphere during the year of formation [Olsson and Possnert, 1992]. One advantage of the ring-porous angiosperms is that they will form a ring every year until they die, although the LW section of the ring may be very narrow if growing conditions are unfavorable. Gymnosperms, on the other hand, may have locally absent rings or false rings, which potentially lead to erroneous age determinations [Ladefoged, 1952; Speer, 2010]. This was the case with the Japanese cedar tree used for the first identification of the 994 C.E. event [Miyake et al., 2013].

3 Samples and Methods

In this study, we use a piece of oak (referred to as Moj) from Mojbøl in Southern Jutland, Denmark, that was dendrochronologically dated by the Danish National Museum [Eriksen, 2004]. This piece of wood covers the time span 980–1006 C.E. The tree rings were first cut and separated into annual samples of LW and EW whenever possible. Because the carbon isotope ratio in the LW fraction only reflects atmospheric ratios during the growth season in the year of formation, LW represents the preferred part of the ring used for this study. However, LW and EW could not be separated for the following years: 989, 990, 999, 1000, 1001, 1005, and 1006 C.E., due to narrow rings and particularly narrow LW sections. For these years, the sample material therefore consists of the whole ring with no distinction between LW and EW. Separate EW and LW samples were obtained for the remaining years, and these EW and LW fractions were measured separately for the years 991–998 C.E. For the remaining years, only LW fractions were measured.

Cellulose represents the most stable part of the wood, since carbon in the cellulose fraction does not cross ring boundaries, whereas other carbon fractions such as lignin and hemicelluloses may cross ring boundaries or be exchanged with the atmosphere after formation [Wilson and Grinsted, 1977; Leavitt and Danzer, 1993; Gaudinski et al., 2005; Loader et al., 1995; Santos et al., 2001; Nemec et al., 2010]. α-cellulose is a chemically extractable component of the wood, which mostly consists of cellulose with only a minor fraction of hemicelluloses [Green, 1963]. α-cellulose is therefore the preferred component of wood for high-precision analysis of radiocarbon. We extracted α-cellulose from the samples by following the method of Southon and Magana [2010] (with a few adjustments) that follows a revised edition of the Jayme-Wise method [Green, 1963]. The procedure is to first remove lignins by using acidified sodium chlorite (1M, NaClO2); we used hydrochloric acid (1M, HCl) for this acidification and bleached the samples for at total of 6 h at 70°C. Afterward, hemicelluloses are dissolved for 1 h in 17% sodium hydroxide (NaOH) at room temperature, which leaves α-cellulose. In order to remove any modern atmospheric CO2 component added to the samples during the alkaline extraction process, the samples were furthermore treated with 1 M HCl for 30 min. at 70°C. After the pretreatment, the samples were combusted to CO2 and subsequently reduced to graphite. The radiocarbon measurements were carried out using a 1 MV Tandetron Accelerator Mass Spectrometry (AMS) system from High Voltage Engineering at the Aarhus AMS Centre (AARAMS), Denmark. The accelerator and the graphitization protocols are described in Olsen et al.[2016]. Radiocarbon ages are reported as conventional 14C dates in 14C yr BP (see supporting information), based on measured 14C/12C ratios corrected for isotopic fractionation using online 13C/12C by normalization to a standard δ13C value of −25‰ Vienna Pee Dee Belemnite (VPDB) [Stuiver and Polach, 1977]. Age-corrected Δ14C values are calculated according to Stuiver and Polach [1977], where it is denoted as Δ.

4 Results

The measured radiocarbon (Δ14C) in the tree rings from Mojbøl is shown in Figure 1, with a distinction between LW, EW, and whole ring data (LW and EW combined from the same year). Because LW is produced during summer months, all LW years are displaced by +0.5 year. A sharp increase of 10.5 ± 3.4‰ in Δ14C is clearly seen from C.E. 993 to 994 in the LW data. Since oak does not have false or missing rings, our results confirm that the cosmic ray event occurred from 993 to 994 C.E. and that it was of hemispheric extent. In the EW data (Figure 1), the rapid increase occurs 1 year later, i.e., between 994 and 995 C.E., but the magnitude of 10.2 ± 3.5‰ is fully consistent with the increase observed in the LW data. The new Δ14C data from Danish oak generally show a declining trend before the cosmic ray event in 994 C.E., although with considerable variability that is slightly higher than the variability in the period following the event (Figure 1). After the increase in 994 C.E., the Δ14C values indicate a modest decline over the ensuing 4–5 years, as expected due to exchange of radiocarbon with the biosphere and oceans. After this initial decline, the Δ14C values remain relatively constant at a level around 12‰ over the ensuing 6–7 years.

Details are in the caption following the image
Δ14C values (in ‰) for the Moj wood piece (LW, EW, and whole ring fractions) from Southern Jutland, Denmark. The LW data from the Danish record are displaced by half a year because LW is produced during summer months. The IntCal13 curve is also shown for the relevant time period [Reimer et al., 2013].

The EW sample from 991 C.E. shows an anomalously high Δ14C value, and we consider it an outlier for the reason discussed below. This sample was by far the smallest of all samples, containing only 0.41 mg of carbon, whereas the other samples contain around 1 mg of carbon. The size of the background samples used to correct all the AMS measurements was also around 1 mg carbon. If the sample size is too small, it may result in radiocarbon ages that are too young, which corresponds to Δ14C values that are too high [Brown and Southon, 1997; Santos et al., 2007]. The anomalously high value observed in the EW sample from 991 C.E. is likely related to the size of the sample, and we therefore omit this sample from further analysis. A comparison between the LW and EW samples for the 7 years where both samples were measured (excluding year 991 C.E.) shows no significant differences between samples from the same year, based on a reduced χ2 test, although with one exception: year 994 C.E. A similar test comparing the LW samples with EW samples from the following year reveals no significant difference for any of the 7 years. Overall, this indicates that the EW and LW fractions generally agree with respect to their 14C/12C ratio and that the year-to-year changes are relatively small. Changes in Δ14C associated with the cosmic ray event in 994 C.E. stand out, as the radiocarbon contents of the LW and EW samples are statistically different for this year.

5 Discussion

The subannual resolution of the new record from Mojbøl in Southern Jutland, Denmark, may help further constrain the timing of the cosmic ray event around 994 C.E. The event thus occurred after the formation of LW in 993 C.E. and after the formation of EW in 994 C.E., but before the formation of LW in 994 C.E. As the growth season for oak trees in Denmark generally lasts from May to September [Schweingruber, 1988; Ladefoged, 1952], it suggests that the event occurred between October 993 C.E. and September 994 C.E. However, the fact that the Δ14C value does not increase in the years following the increase in 994 C.E. suggests that the event did not occur during the late summer or early autumn of 994 C.E., a notion that is supported by the fact that the mass transport across the tropopause in the Northern Hemisphere has a maximum in May/June and a distinct minimum in the early autumn [Appenzeller et al., 1996]. This further constrains the possible time interval to October 993 C.E. to June 994 C.E., as it appears unlikely that the extra 14C produced in the stratosphere after June is incorporated into the LW fraction of the same year without resulting in a further increase the following year. This estimate, however, remains a relatively conservative estimate, assuming that all carbon in the EW fraction from 994 C.E. derived from the growth season in 993 C.E. This is probably not correct [Olsson and Possnert, 1992], as the EW fraction from 994 C.E. most likely took up a considerable amount of carbon from the atmosphere during the early spring of 994 C.E., which further constrains the timing of the event to the late spring of 994 C.E., most likely the period April–June 994 C.E., since there is no trace of the event in the EW data from 994 C.E. The improved timing of the event highlight the potential value of the 994 C.E. cosmic ray event as an astrochronological tie point that can be used to anchor floating chronologies.

The new radiocarbon record from Denmark is generally in good agreement with the existing radiocarbon records from Japan (Figure 2) [Miyake et al., 2013, 2014b]. The timing and magnitude of the cosmic ray event is consistent in all three data sets (Miyake et al. [2013]: 9.2 ± 2.6, Miyake et al. [2014b]: 11.3 ± 2.5, combined record based on Danish oak: 10.5 ± 3.5 (LW data)), confirming that the change in atmospheric radiocarbon concentration was highly uniform in the Northern Hemisphere. Both records from Japan additionally indicate a declining trend in the years preceding the 994 C.E. event, although the trend is rather weak in the cedar tree record [Miyake et al., 2013], and a modest decline Δ14C over the 4 years ensuing the event. The longer of the two Japanese records [Miyake et al., 2013] continues the decreasing trend, albeit with a tendency to flatten out in the period 998–1005 C.E., which resembles the flat trend observed in the new record from Denmark. The most significant decline in Δ14C values from Japan thus occurs around 1007 C.E., i.e., after the plateau observed in the Danish record. It is unclear why the Δ14C values remain at this relatively high plateau in the period 998–1005 C.E., but the same tendency, although weaker, can be observed in the 10Be flux record from the North Greenland Ice Core Project (NGRP) ice core [Mekhaldi et al., 2015]. Similar to the new radiocarbon record from Denmark, this 10Be flux record also shows an anomalously high peak around 1006 C.E. Data from the North Greenland Eemian Ice Drilling (NEEM) ice core are more ambiguous, as the 10Be flux record shows no indications of this plateau [Mekhaldi et al., 2015], whereas the 10Be concentrations indicate a slight increase around 1000 C.E. after adjustment of the time scale [Sigl et al., 2015].

Details are in the caption following the image
Δ14C values (in ‰) for the Moj wood piece (LW, EW, and whole ring fractions) from Southern Jutland, Denmark, are displayed along with the existing high-resolution Δ14C data from a cedar tree and a hinoki tree from Japan [Miyake et al., 2013, 2014a]. Moj LW and Moj whole ring years are displaced by +0.5 years. The IntCal13 curve is also shown for the relevant time period [Reimer et al., 2013].

Similar to the Japanese Δ14C records, the LW record from Denmark reflects changes in atmospheric radiocarbon during the summer months. The Danish LW record shows a remarkable agreement with the cedar tree record of Miyake et al. [2013], albeit with small, but highly consistent, negative offsets with an average of −2.1 ± 1.0‰ (Figure 3). The average offset between these records is smaller than the offset between the Danish record and the hinoki tree record (−5.1 ± 1.1‰) [Miyake et al., 2014a] as well as the offset between the two Japanese records (−2.8 ± 0.9‰). These systematic offsets could reflect laboratory offsets, or they could be related to the geographical origin of the records. The latter hypothesis is supported by the fact that the Japanese samples originate from locations (Iida City and Yaku Island) in close proximity to the Pacific Ocean, where upwelling of old, radiocarbon-depleted deep water takes place. This may result in lower radiocarbon values compared to records that derive from locations in the Northern Hemisphere at a further distance far from the Pacific Ocean, such as the record based on Danish oak in this study. This explanation is also consistent with a recent study by Nakamura et al. [2013] who generally find older radiocarbon ages for Japanese samples compared to IntCal09. However, the location relative to the ocean cannot explain the difference between the two Japanese records, as the hinoki record derives from the inland city of Iida and therefore is expected to give higher Δ14C values than the cedar tree record from Yaku Island, which is closer to the ocean [Miyake et al., 2014a].

Details are in the caption following the image
(a) Δ14C values for the LW fractions from Danish oak (this study) as well as for the cedar tree and hinoki tree data from Japan [Miyake et al., 2013, 2014a] and Moj LW years are displaced with +0.5 years. (b) Differences between the Δ14C values of the Danish record and the cedar tree record [Miyake et al., 2013] from Japan. (c) Differences between the Δ14C values of the Danish record and the hinoki tree record [Miyake et al., 2014a] from Japan. The lines represent the weighted averages of the differences where the records have overlapping data. The weighted average in Figure 3b is −2.1 ± 1.0‰, whereas the weighted average in Figure 3c is −5.1 ± 1.1‰.

The new Δ14C record from Danish oak, along with the Δ14C records from Japan, reveal that the changes in Δ14C were more dynamic than those suggested by the IntCal13 curve (Figures 1 and 2). For instance, the annual radiocarbon measurements presented in this study indicate a declining trend in Δ14C values leading up to the 994 C.E. event, a trend that is also present in the high-resolution records from Japan, although it is less evident in the cedar tree record from Yaku Island [Miyake et al., 2013]. This declining trend is not visible in the IntCal13 curve, which generally display modest changes prior to the onset of the Oort Solar Minimum that peaks around 1055 C.E. (Figure 4). The declining trend in the high-resolution 14C records prior to the 994 C.E. event could be related to the carbon cycle or, alternatively, an increase in solar activity associated with changes over the Schwabe 11 year cycle. This could imply that the 994 C.E. event occurred around the maximum, or right after the maximum, of the 11 year solar cycle. This would be consistent with observations showing that most solar flares happen around the maximum of solar cycles and that the number of flares per active region is larger shortly after the cycle maximum [Hudson et al., 2014]. However, the high-resolution 14C records documenting the cosmic ray event in 775 C.E. do not display a clear declining trend prior to the event, but rather a mix of different behaviors. Consequently, it is currently not possible to conclude that these high-energy cosmic ray events occur around the maximum of solar cycles.

Details are in the caption following the image
The new Δ14C record from Mojbøl in Southern Jutland, Denmark, which is characterized by a subannual resolution across the 994 C.E. cosmic ray event. Also shown are the date behind the IntCal calibration curve [Reimer et al., 2013].

6 Conclusion

We present a new high-resolution radiocarbon record from Danish oak covering the cosmic ray event in 994 C.E. The new record suggests that a sharp increase of 10‰ in atmospheric radiocarbon occurred between the late autumn of 993 C.E. and the early summer of 994 C.E., most likely in the period April–June 994 C.E. Close agreement with radiocarbon records from Japan confirms that the cosmic ray event in 994 C.E. was of hemispheric extent and that the Δ14C signal defining the event was highly uniform throughout the Northern Hemisphere. This highlights the potential value of the cosmic ray event in 994 C.E. as an astrochronological tie point.

Acknowledgments

We would like to acknowledge financial support from the Villum Foundation (VKR023114 and VKR010116). Funding for the Stellar Astrophysics Centre is provided by the Danish National Research Foundation (grant agreement: DNRF106). We would like to thank Niels Bonde and Claudia Baittinger from the Danish National Museum for their help and advice. We would also like to thank the personnel of AARAMS for their great help. The 14C ages and the Δ14C values found in this study are provided in Table S1 in the supporting information.