Variation of solar cyclicity during the Spoerer Minimum

1. Introduction

[2] The Spoerer Minimum (1415–1534 A.D.) is one of the grand solar minima that occurred prior to the Maunder Minimum (1645–1715 A.D.). This minimum, as well as the Maunder Minimum, is thought to have played an important role in the global cooling of the Earth from the 15th to the 19th century referred to as the “Little Ice Age” [Mann et al., 1999]. The existence of this minimum had been suggested by a 100-year-long increase of the cosmogenic nuclides (Figure 1) [Eddy, 1976], whose production rate is inversely correlated to the intensity of solar activity [Stuiver, 1961]. Scarcity of reports of sunspots and auroral phenomena also indicates the existence of the Spoerer Minimum [Eddy, 1976]. Since continuous observations of sunspots with telescopes have been conducted only since the 17th century, records of sunspots during the Spoerer Minimum are limited to a few seen with naked eyes [Eddy, 1976]. However, it is commonly speculated that sunspots were almost absent during this minimum as well as during the Maunder Minimum (Figure 2) [Stuiver and Quay, 1980].

[3] Since the records of both aurorae and sunspots during the Spoerer Minimum are even more scarce than those for the Maunder Minimum, it is difficult to discuss features of the 11-year solar cycle during this period quantitatively. In that context, this cycle has been investigated by using cosmogenic nuclides. The most widely used cosmogenic nuclides are carbon 14 in tree rings and beryllium 10 in ice cores. Both tree rings and ice cores are stratified structures and thus contain records of the variation of the nuclide abundance.

[4] Cosmogenic nuclides are produced mainly by galactic cosmic rays modulated by the interplanetary magnetic field and the geomagnetic field. Cosmic rays produce several kinds of radioisotope, mainly in the stratosphere, by spallation of nuclei or by neutron capture on atmospheric nuclei. Generally, the production rate of cosmogenic nuclides is less when solar activity is intense, and greater when solar activity is weak. Carbon 14, which is one of the cosmogenic nuclides, is oxidized immediately to form carbon dioxide and circulates within the atmosphere with a mean residence time of a decade [Siegenthaler et al., 1980]. Some of this carbon is absorbed into trees via photosynthesis and is incorporated into tree rings.

[5] The advantage of examining carbon 14 over other cosmogenic nuclides is the reliability of dating and the ease of separating the annual layers since there is a definite boundary between annual rings. On the other hand, the amplitude of carbon 14 variation caused by the 11-year solar cycle is strongly suppressed in the global carbon circulation [Siegenthaler et al., 1980]. The variation of about 20–30% in the flux of cosmic rays in the upper atmosphere will be reflected in the carbon 14 as a variation of only 2–3‰. Thus high-precision measurement is required to investigate the 11-year solar variation using the carbon 14 content of tree rings.

[6] For the period around the Spoerer Minimum, there are several previously obtained time sequences of carbon 14 and beryllium 10. Annual carbon 14 data obtained using trees from the Pacific northwest of the United States are available since 1511 A.D. [Stuiver and Braziunas, 1993; Stuiver et al., 1998]. Annual beryllium 10 data obtained using the Dye3 ice core from Greenland are available from 1424 A.D. [Beer et al., 1990]. In addition, decadal carbon 14 data are available for this period [Stuiver et al., 1998]. Using the record of beryllium 10, it was recently suggested that the Sun did not cease cyclic variations completely during the Maunder Minimum [Beer et al., 1998]. Periodic variation of solar activity during the Maunder Minimum has been also suggested using the carbon 14 record by Stuiver and Braziunas [1993]. Periodicity of carbon 14 content during the Maunder Minimum has been analyzed in our earlier paper using two carbon 14 records [Miyahara et al., 2004]. In the spectrum, we have found an existence of solar cycle with a period of about 14 years.

[7] Recently, more and more attention is being given to the mechanisms and the extent of the influence of solar variability on the Earth's climate. The change of the average temperature in the past has been reconstructed with a scale of hundreds to thousands of years by using oxygen isotopes [Bond et al., 2001]. This also manifested a periodic variation correlated with solar activity. However, in addition to the mechanisms of the solar influence on the climate, the mechanisms of periodic long-term variations such as occurrences of grand solar minima are still unclear. Determining the specific characteristics of the 11-year cycle during the grand minima may provide clues to these mechanisms. Whether the frequency modulation is seen also during the Spoerer Minimum as well as the Maunder Minimum is quite an important question. In this paper, carbon 14 concentration in the tree rings from 1413 to 1554 A.D. with annual time resolution are reported and then the variation of the 11-year solar cycle is discussed.

2. Method

[8] For this study, a 712-year-old Japanese cedar tree (Cryptomeria japonica) was used. The tree was obtained from the Yakushima Island located at the southernmost tip of Kyushu in Japan (30.18° N, 130.30° E). The Yakushima Island is inscribed as a World Heritage site for its extraordinary well-conserved primitive nature and holds a lot of long-lived cedar trees. The radius of the tree we obtained was about 1.4 m, and the average ring width for the period around 1400–1560 A.D. was about 0.5 mm. In order to determine the absolute date of each of the 712 tree rings, we used the bomb effect peak in 1964 A.D. Additionally, dendrochronology has been applied for this tree by using ring width data of several trees from Yakushima Island. In this way, it was revealed that the tree sprouted in 1280 A.D. and cut in 1991 A.D. Each of the annual rings was then carefully separated and chemically treated as described below in order to obtain the constituent that cannot move between rings and turn it into graphite as target material for measurement of carbon 14 concentration using the Accelerator Mass Spectrometer (AMS). We used the HVEE AMS of Nagoya University in Japan [Nakamura et al., 2000], which achieves an accuracy of 3 to 3.5‰.

[9] The separated wood was shredded into small pieces and dried. These were then (1) soaked in a benzene-ethyl alcohol mixture for 6 hours to remove resin, etc., and dried, (2) bleached in a NaClO₂/HCl solution at 80°C for 8 hours to remove the lignin, (3) washed in boiling distilled water for 30 min, and (4) rinsed with boiled distilled water and dried; after that, the resultant cellulose was (5) reacted with CuO in vacuum to obtain CO₂, which was (6) purified with cold traps at the temperatures of −90° and −130°C and (7) converted to graphite on Fe powder by hydrogen reduction according to the method by Kitagawa et al. [1993].

3. Results and Discussion

[10] The measurements have been completed for the period from 1413 to 1554 A.D., including the whole of the Spoerer Minimum. Figure 3 shows the data (dots) and the 3-year running averages (solid line) of the data. We applied the generally used calculation method by Stuiver and Polach [1977] to determine the carbon 14 abundance. The errors of the data were statistically estimated by the counting error of the carbon 14 and the stability of the system, which was checked with several standard samples simultaneously prepared. Depending on the system condition, we repeated measurement a few times to obtain enough accuracy, and thus the errors of some measurement points are smaller than the average accuracy of 3.3‰. In Figure 3, decadal data (as in Figure 1) and the annual data by Stuiver et al. [1998], available from 1511 A.D., are plotted together. Our data are consistent with the decadal data but, on the other hand, indicate some short-term variations of less than several per mil. The annual data from 1511 A.D. onward seem to show some regional difference of carbon 14 content between Japanese and Pacific northwestern area, although short-term variations of the two series are similar (Figure 4).

[11] The carbon 14 data were analyzed by using the S transform [Stockwell et al., 1996] in order to investigate the time variation of the amplitude of the 11-year cycle during this period. The S transform is an extended method of the wavelet transform [Torrence and Compo, 1998]. Similar spectra can be obtained using the wavelet transform, but the S transform is particularly suitable to determining the Fourier period directly. As a preliminary step, two missing points were interpolated by a linear function, and both the long-term trend (>50 years) and the short-term variations (<3 years) were subtracted using a band-pass filter. Figure 5 shows the result of the S transform. Time variations of the amplitude of each periodicity are presented. In the spectrum, the signal for the 11-year cycle can be seen in addition to the high-frequency variations around a period of 3–5 years. It is worthwhile to mention that the signal for the 11-year cycle seems to be more or less continuous except for the period around 1455–1510 A.D., and is strongly modulated only around 1460 A.D. and 1500 A.D. The amplitude of the 11-year variation during 1430–1455 A.D. and during 1510–1540 A.D. is about 2.0‰ and about 1.3‰, respectively, whose significance levels are about 1.9 and 1.2 sigma. We regard the noise level as the standard deviation of the high-frequency variations, which was estimated by subtracting the 3-year running averages from the raw data. The variation of 1.3–2.0‰ of carbon 14 content corresponds to about 13–20% of cosmic ray variation, which is similar to the recent cosmic ray variations. Around 1495–1510 A.D. no prominent signal could be detected. In contrast, the signal for the 22-year cycle, which can be emerged because of solar polarity reversals, seems to remain quite weak with amplitude less than 1‰ through the period investigated.

[12] The spectrum of the wavelet analysis of the beryllium 10 record from the Greenland ice core [Fligge et al., 1999] is available since around 1460 A.D. It shows a remarkable signal of about a 20- to 24-year period from ∼1500 A.D. to ∼1620 A.D. This signal seems to correspond to the 22-year variation due to solar polarity reversal. Although its significance is much larger than that seen in the spectrum of the carbon 14, the signal for the 11-year cycle is not significant in the beryllium 10 record. Only a weak signal of about 11 years is seen around 1520 to 1570 A.D. From 1570 to 1630 A.D., both 11-year and 22-year periods are significantly detected in the beryllim-10 record. While, the significant signal of about 14–20 years around 1500 A.D. is not seen in the spectrum of our carbon 14 record. This signal seems to be due to the significant temporal decrease around 1500 A.D. in the beryllium 10 data. This may be caused by the climatic effect specific to the accumulation system of beryllium 10.

[13] The attenuation level of the galactic cosmic rays in the heliosphere is dependent not only on the solar activity level but also on the solar polarity, which reverses every 11 years. This causes asymmetric structure in alternate 11-year periods in the cosmogenic nuclei abundance to make a 22-year cyclic behavior [Jokipii, 1991]. It has been also suggested that this effect becomes more prominent when the 11-year variations are suppressed associated with the weakening of solar activity. The 22-year cyclic variations are dimly seen in the raw data of carbon 14 from around 1460 to 1530 A.D. (Figure 3) as the suppressions of every other 11-year cycle. It might be suggesting that this is when solar activity was most suppressed. There is a possibility that we can determine the transition of the solar polarity during this period on the basis of the 22-year variations of carbon 14 record.

[14] With the features of the 11-year cyclicity during the Spoerer Minimum, some important issues can be discussed. The period of the Spoerer Minimum was defined by the increase of the decadal carbon 14 abundance, using as a reference sunspot data and carbon 14 abundance during the Maunder Minimum when sunspots were almost absent [Stuiver and Quay, 1980]. However, it is likely that more specific estimation on the minimum period of solar activity can be made by investigating the behavior of the 11-year cycle in the carbon 14 data with annual time resolution, though one must take notice of 2–3 years of time lag between the production of carbon 14 in the upper atmosphere and absorption into trees [Siegenthaler et al., 1980]. Here we assume that the period when solar activity was most suppressed was around 1455–1510 A.D., or a few years earlier.

[15] An interesting comparison can be made with our result and the records of the temperature reconstructed for the northern hemisphere [Mann et al., 1999]. In the time sequence of the reconstructed temperature anomaly (Figure 6), there is a great decrease around 1460 A.D. to 1500 A.D., which is almost the same period when the 11-year cycle in carbon 14 record was most affected during the Spoerer Minimum. By assuming that there is a strong connection between solar activity and the climate, both of the carbon 14 record and the temperature anomaly record suggest that solar activity might have slightly recovered around 1480 A.D. during the Spoerer Minimum.

[16] There are several reports discussing the inverse correlation between intensity and the length of the 11-year cycle of solar activity in the recent centuries [Solanki et al., 2002]. Our earlier study had also demonstrated the lengthening of the 11-year cycle during the Maunder Minimum. The result of the S transform for the Spoerer Minimum, however, did not show a particular tendency toward longer cycles when the 11-year cycle was suppressed. This feature might distinguish between Maunder-type and Spoerer-type minima that occurred in the past [Stuiver and Braziunas, 1989]. Around 1460 A.D., instead, both shorter and longer periods than 11 years have been detected. Further study on the behavior of the 11-year cycle around 1455–1510 A.D., when the modulation of the 11-year cycle was found, may provide clues to this problem.

4. Summary

[17] Measurements of carbon 14 in a Japanese cedar tree for the time around the Spoerer Minimum have been made to investigate the characteristics of the 11-year solar cycle during this period. As a result of the first measurements with annual time resolution, the variation of the 11-year solar cycle has been clarified: The 11-year cycle (1) had existed in most of the time during this prolonged minimum; however, (2) it seems to have been modulated around 1455–1510 A.D. It is worth mentioning that this nearly corresponds to the period when reduced temperatures were reported. The period around 1455–1510 A.D. may be the time that solar activity was most suppressed among the Spoerer Minimum period. There seems to be some discrepancy between the carbon 14 records and the beryllium 10 record around 1500 A.D. This may be due to the temporal strong effect of climate to the accumulation of beryllium 10.