Records of sunspot and aurora during CE 960-1279 in the Chinese chronicle of the Song dynasty

Records of sunspots and aurora observations in pre-telescopic historical documents can provide useful information about solar activity in the past. This is also true for extreme space weather events, as they may have been recorded as large sunspots observed by the naked eye or as low-latitude auroras. In this paper, we present the results of a comprehensive survey of records of sunspots and auroras in the Songshi, a Chinese formal chronicle spanning the tenth to the thirteenth century. This chronicle contains a record of continuous observations with well-formatted reports conducted as a policy of the government. A brief comparison of the frequency of observations of sunspots and auroras and the observations of radioisotopes as an indicator of the solar activity during corresponding periods is provided. This paper is the first step of our project in which we survey and compile the records of sunspots and aurora in historical documents from various locations and languages, ultimately providing it to the science community as online data.


Keywords
Aurora; Sunspot; Solar activity; Chinese chronicle; Historical resource Background Historical solar magnetic activity is of great interest because it provides crucial information on the origin of solar magnetic fields and its variability as well as the relationship of solar activity and the climate of the Earth (Usoskin 2013). Telescopic sunspot observations have been recorded since the era of Galileo about 400 years ago. The sunspot number is considered to be the standard index of the solar activity (Clette et al., 2014), and indeed, it is known as the longest-term scientific data (Owens 2013). For the older solar activity, cosmogenic radionuclides such as 14 C and 10 Be are commonly used as the proxy (e.g., Stuiver & Braziunas 1989;Beer et al. 1990). Eddy (1980) pointed out that pre-telescopic historical records of sunspots, auroras, and total eclipses (shapes of solar corona) could be useful for the reconstruction of the past solar activity. Several authors have published the lists of astronomical events including naked-eye sunspots and auroras in the literature from Japan (Kanda 1933;Matsushita 1956;Nakazawa et al. 2004;Shiokawa et al. 2005), Korea (Lee et al. 2004), Arabic countries (Basurah 2006), Europe (Fritz 1873;Link 1962;Dell'Dall'Olmo 1979;Stothers 1979;Vaquero & Trigo 2005;Vaquero et al. 2010), Russia (Vyssotsky 1949), North America (Broughton 2002), the Tropical Atlantic Ocean (Vázquez & Vaquero 2010), and China (Schove and Ho 1959;Keimatsu 1970Keimatsu -1976Yau and Stephenson 1988;Saito & Ozawa 1992;Yau et al. 1995). A review on the historical records of solar activity is given in the monograph by Vaquero and Vázquez (2009).
In addition to the long-term changes, studies on historical solar activity have been attracting interest in the context of extremely intense solar flares and space weather events. Araki (2014) searched the historically largest geomagnetic sudden commencement (SC) using records of ground-based geomagnetic observations since 1868. The most intense geomagnetic storm on record is believed to be the so-called Carrington event in 1859 (Carrington 1859), whose estimated disturbance storm time (Dst) index is ~1760 nT (Tsurutani et al. 2003). The superfast coronal mass ejection (CME) on July 23, 2012, is thought to have been as intense as the Carrington event, though it missed Earth (Baker et al. 2013, Russell et al. 2013. Another superfast CME hit the Earth in August 1972 (Vaisberg and Zastenker 1976). This event yielded only moderate magnetic storm (Dst ~125 nT), presumably owing to the northward alignment of the interplanetary magnetic field, but it generated intense energetic particles and very rapid change in local geomagnetic fields (Tsurutani et al. 1992), causing significant damages in space and ground-based infrastructure (Lanzerotti 1992). If it were with a southward magnetic field, it could have produced very intense magnetic storm and aurora. Such intense energetic particles and magnetic storms severely damage space and ground infrastructure, resulting in a significant (slowly rotating G-type stars), whose total energies are estimated to be 10 33-35 erg, i.e., 10 to 1000 times more energetic than the most intense solar flares. Shibata et al. (2013) suggested that, theoretically, such super-flares may also occur on the present Sun. Interestingly, Miyake et al. (2012 found that the atmospheric 14 C content measured in tree rings exhibited marked increases during the Common Era ‡ (CE) 774 to 775 and during CE 993 to 994, which indicate sharp increases in cosmic ray fluxes during those periods. The origin of these intense cosmic ray events is not known, but one potential cause is the solar energetic particles (SEPs) caused by extreme solar flares. Although the strongest SEPs in the modern observational records cannot account for these events, the SEPs from super-flares may be intense enough to cause these 14 C increments. Several studies have been published about the CE 774-775 including those by Usoskin et al. (2013), Cliver (2014, and Neuhäuser and Hambaryan (2014). This paper is the first result of a project in which we aim to re-survey and compile the records of observations of sunspots and aurora from as many auroral sources as possible, and then provide the compilations as digital data to the scientific community. As a preliminary effort, in this paper, we present the data for the formal chronicle of the Sòng dynasty, Sòngshǐ (宋史, CE 960-1279), and we compare these results with data from cosmogenic radionuclide studies, such as the one performed by Miyake (2013).

Astronomical observations in the Sòng dynasty
In this study, we used the Chapter of Astronomy (天文志) in Sòngshǐ, mainly because it includes one of the richest sets of astronomical information among the chronicles of the Chinese dynasties. All the capitals in the Sòng era, where the observations were made, were located at latitudes between 30 N and 35 N. This period (CE 960-1279) overlaps with the CE 993-994 event as well as the so-called Medieval Warm Period (MWP: 10 CE-14 CE).
Among historical records worldwide, Chinese chronicles are remarkable for their feasibility as scientific data. Keimatsu (1970Keimatsu ( -1976 emphasizes the predominance of Chinese ‡ "Common Era" is a non-religious alternative of AD, namely "Anno Domini." astronomical records because trained experts on astronomical observation made continuous observations at specified locations and took dated records of astronomical phenomena, often with detailed notes such as motions, shapes, and colors. Thus, records in the Chinese official chronicles are regarded to be more objective than many other historical documents (Keimatsu 1970(Keimatsu -1976 The reason for their consistent observations is based on Chinese culture and the politics of dynasties. Astronomical phenomena were traditionally thought to be signs from the heavens to the emperors reflecting their politics (i.e., stated at Sòngshǐ, Astronomical I, p. 949, translated text available in Appendix 1).
In order to deliver those heavenly messages to emperors, especially those of the Sòng dynasty, two observatories were constructed both in the imperial palace and near the capital city, Kāifēng, until the Jìngkāng Incident in CE 1126 (original text available in Appendices 2, 3, and 4, Yabuuchi 1967). Owing to their political importance, astronomical observations were made even during wars.
There is anecdotal evidence indicating the degree of sophistication and consistency of the astronomical observations in China during this period. During the invasion of the Sòng dynasty by the Jīn from the north, the Jīn captured Kāifēng in CE 1126. They broke all the observational instruments except for the spherical astrolabes, which were brought to Bĕijīng and adjusted by 4° in order to point to Polaris, as the instruments were moved from Kāifēng (35 N) to Bĕijīng (39 N) (the original text is available in Appendices 5-6).
It was in CE 1143 that the Sòng dynasty began reconstructing astrolabes. However, even during the period between CE 1126 and 1143 (without the use of astrolabes), we find records of both auroras and sunspots. This suggests that the Sòng dynasty continued observation even in such a chaotic time and without specially designed instruments (original text available in Appendix 6). Of course, we still have to note that frequency and accuracy of these observations must have been influenced by the political climate and other unknown factors.

Search method
The Sòngshǐ includes descriptions of many kinds of phenomena observed in the sky. Since we are interested in the past solar activities, we searched for descriptions that could be regarded as records of sunspots or auroras. For this purpose, we used a search engine, Scripta Sinica, (http://hanchi.ihp.cinica.edu.tw) provided by Academia Sinica in Taiwan (http://www.sinica.edu.tw) that incorporates all the text data. This allowed us to automatically flag sentences that included keywords such as "black spot (黒子)" and "red vapor (赤氣)," which may refer to sunspots and red auroras, respectively. Once the sentences that include the keywords were picked up, we read corresponding parts of the original text to check whether they actually refer to sunspots or auroras. We also calculated the moon phase to determine the sky conditions for each date of observation.

Sunspot records
Sunspots are described as black spots or black vapors in the sun or in terms such as "the sun was weak and without light" (Saito & Ozawa 1992). In the Astronomical Treatise of Sòngshǐ, these are categorized in the subsection of unusual phenomena in the sun. Since this period was long before the invention of telescopes, the observations were presumably made by the naked eye during sunrises/sunsets or through clouds.
Sunspot records are often accompanied with information about number, shape, and size.
Examples are given as follows: (a) 元豐元年十二月丙午, 日中有黑子如李大, 至丁巳散. The sizes are expressed by comparison with something tangible. In the above example, it was "as large as plums." The other expressions of size included are peach, glass, hen's egg, and duck's egg. Although we usually think of peaches as larger than plums, we do not know the relative and absolute size indicated by these expressions.

Auroral records
Although the ancient Chinese did not know the physical nature of the phenomena, there are numbers of records that can be considered as the observations of auroras. For example, Keimatsu (1969aKeimatsu ( , 1969b and Saito and Ozawa (1992) showed that the word "vapor" is likely to indicate auroras. Keimatsu (1970Keimatsu ( -1976 listed all the luminous phenomena seen at night and also discussed whether they correspond to auroras or not.
Similarly to Keimatsu, we assume that the records of luminous phenomena observed at night are potentially those of auroras. Therefore, we surveyed the words that refer to luminous phenomena, such as vapor, light, and cloud. In the Astronomical Treatise of Sòngshǐ, these are categorized in the subsection of clouds and vapors. From the list of the potential auroral candidates, we manually removed the following two types. The first are those without dates.
Most of these are found in the chapters explaining how fortunetelling is performed, which includes conversations between the emperor and servants or sages. Therefore, they are unlikely to be the direct records of the observations. The second type includes those seen during the daytime. Some of the daytime phenomena are presumably the halo around the sun.
Records of auroral candidates often include information about their color, motion, and direction; length, shape, and number of their bands; and sometimes the location of the observation when it was not made in the capital city. Examples of auroral records are given as follows: (a) 至道二年二月丙子夜, 西方蒼白色氣長短八如彗掃. 稍經天漢, 參錯如交蛇. Sometimes, these descriptions include information about constellations, planets, or the moon accompanying the auroras. Their lengths are given in units of "chǐ"(尺) or "zhàng"(丈).
In the Sòng Era, 1 chǐ was equal to 30.72 cm and 1 zhàng was 10 chǐ (Tonami et al. 2006). At this moment, we do not know how the lengths expressed using these units correspond to what was actually seen in the sky.
The shape and number of bands are described only in some of the recorded events. The shapes are expressed in a figurative way, such as "like serpents" or "like silk textiles."

Moon phase calculation
The moon phase is a trait of the sky that we can calculate accurately for historic dates.

Overall result
The lists of the candidates of sunspot and aurora are shown in Tables 1 and 2, Figure 1 shows the annual number of records. We do not have enough records to see any signature of the 11-year solar cycle, but we can observe some long-term modulations. Overall, the period of CE 1100-1200 looks more active than the remainder. In particular, the sunspot records cluster during the periods of CE 1100-1145 and CE 1185-1205, when many auroral candidates are observed as well. There are also many auroral candidates during CE 1160-1175, but no sunspot candidate was found during this period.

Color of aurora
The In very intense auroras, blue, violet, or pink colors in the low altitude (80-100 km) are sometimes seen that come from the molecular nitrogen lines in the blue and red parts of the spectrum. These lines are mostly excited by energetic electrons injected from the magnetosphere (Chamberlain 1961). Moreover, during large geomagnetic storms, so-called stable auroral red arcs (SAR arcs) become visible in the mid latitude (Rees andRoble 1975, Kozyra et al. 1997). All the auroral candidates presented in this paper are associated with colors.
Among the 193 auroral candidates, 75 events are described as white, 58 as red, and 28 as blue white. The other 32 events include yellow, blue white, gold, etc. As mentioned above, the ancient Chinese use the five colors associated with the Wǔxíng, namely white, red, blue, yellow, and black. There is no green in Wǔxíng. Therefore, we assume that white colors in the Chinese records correspond to the green color of the aurora from the oxygen 557.7-nm line. Of course, there are always possibilities that other phenomena such as clouds and atmospheric optics are included. Keeping this in mind and recalling that the observations are made in the latitude of 30 N-35 N, the fact that there are more white auroras than the red auroras in the Chinese record seems puzzling. In such low-latitude region, usually only the red auroras are seen. This is either because the usual aurora in the higher latitude is seen from the lower latitude so that only the red color in the higher altitude (>200 km) is visible above the horizon, or because the aurora is the SAR arc, in which case the pure red (without green) color may be visible even near the zenith.
The white (green) auroras in the record indicate that the usual auroral oval expanded as low as 30 N-35 N where the capital of the Sòng era was located. Here, we should also note that the geomagnetic latitude of China in the Sòng era could be higher than that in the present (Butler 1992). There are some modern observational records that the auroral oval expanded to such a low latitude during the extremely intense geomagnetic storms (Loomis 1860, Kimball 1960).
Therefore, it is likely that the green auroras were observed in the capital of Sòng when the solar activity was strong. However, that there are more red auroras than white (green) auroras still remains puzzling.

Naked-eye sunspots
During the Sòng era, the sunspot observations are likely to have been made by the naked eye.
Therefore, only very large sunspots could have been detected. Schaefer (1991Schaefer ( , 1993) developed a theoretical model of sunspot visibility, including naked-eye observations. The simplest limit of visibility can be expressed as ~1.22 λ/d (λ and d are the mean wavelength of sunlight and diameter of the optical system) (Vázquez & Vaquero 2010). Assuming a value of 1.5 mm for a diurnal observation, a value of 500 nm for the wavelength of the sunlight, and observers with standard eyesight (20-20 vision), the minimum angle of a naked-eye sunspot will be 70″, and a sunspot larger than that or as large as one thousandth the area of the solar hemisphere can be regarded as naked-eye sunspots.
As a previous study (Heath 1994) suggests, the number of such large sunspots is a good proxy of the overall solar activity. Figure 2 shows the number of large (greater than one thousandth of the area of the hemisphere) sunspots, as well as the sunspot number (the Wolf number or the international sunspot number) in the most recent three-and-a-half solar cycles.
This figure demonstrates that the number of naked-eye sunspots is in fact a good proxy for the sunspot number.
Indeed, the emergence of large (naked-eye) sunspots is likely to be the necessary condition for the occurrence of extremely intense solar flares. Shibata et al. (2013) presented a scaling law that relates the energy of solar (and stellar) flares and the size of sunspots; for example, the maximum energy of a solar flare of a naked-eye sunspot can be around 10 35 erg (10 3 times larger than our scientific observation record).

Comparison with 14 C data
Using the 14 C content in tree rings, Miyake (2013) reported that there is a sharp increase reported in cosmic ray flux during CE 993-994. We did not find any candidates of sunspot nor aurora in these years. However, as can be seen in Fig. 1, we found a cluster of auroral candidates several years after this event. The closest aurora recorded is in 996 and there is a record of a sunspot in CE 1005. In this chronicle, we could not detect records clearly associated with the CE 994 event of rapid 14 C increase to which Miyake (2013) referred. Figure 3 shows 14 C data from Miyake (2013) in dots with error bars and counts of Sòngshǐ auroral records in bars.
According to Fig. 3, in CE 994, neither aurora nor sunspot observation was recorded, but after the 14 C peak of CE 994, the number of observed auroras increased. There are two interpretations for this. The first is that this CE 994 event is not related to a short-time solar activity such as a flare. The second is that the record that involves the CE 994 event is not recorded in the Sòngshǐ for various reasons: weather, lack of observation, lack of importance for Chinese astronomers, etc., but the increase in the solar activity during that period is reflected in the increased auroral records some years later. This kind of difficulty is one of the limitations that our surveys frequently face.
We also note that auroras around CE 993-994 were observed in other parts of the world. These may correspond to the astronomical event in CE 993-994.

Auroras and moon phase
For studies of auroral records from historical articles, concerns on moon phase might be important in estimating the actual events indicated by the records. One reason for our interest in moon phase is that there are some natural optical events possibly expressed as vapor, such as "moon halo" and paraselene (or "moon dog"), which occur rarely with a nearly full moon and ice crystals in cold air. This kind of atmospheric optics would occur more frequently in a polar region, but may possibly occur at lower latitudes given appropriate weather conditions.
If the frequency of the "potentially aurora" records depends on the moon phase, it indicates that there may be significant number of events that are actually atmospheric optics. In order to examine this, we calculate the moon phase (0 for the previous new moon and 1 for the next new moon) and make a histogram of historical records of vapors from Sòngshǐ.

Conclusions
We have surveyed sunspot and aurora-like descriptions in the official chronicle of the Sòng dynasty during CE 960-1279. We found 38 sunspots and 193 auroral candidates during that period, with information of their size, color, etc. The lists are provided in this paper. This is the first step for the project in which we will survey the records of sunspot and aurora in a variety of historical documents and provide the data online for use by the scientific community. We welcome and encourage the use of our data as well as the contributions that provide more data from diverse historical sources.

Authors' contributions
This research was performed with the cooperation of authors as given: HH made philological and historical contributions. HT and ADK made contributions on the interpretation and analysis of the database. HI supervised this study. All authors read and approved the final manuscript.