Introduction

Extreme space weather events, such as the events of August–September 1859 (Carrington 1859; Kimball 1960; Green and Boardsen 2006; Farrona et al. 2011; Hayakawa et al. 2016, 2018b, 2019b; González-Esparza and Cuevas-Cardona 2018), February 1872 (Meldrun 1872; Silverman 2008; Hayakawa et al. 2018a), and May 1921 (Silverman and Cliver 2001; Love et al. 2019a, b), cause extremely intense magnetic storms. Among other effects, one of the most interesting visual phenomena is the occurrence of very intense and bright aurorae. Aurorae during extreme events are not only observed at high latitudes, but at low latitudes (22–23°) as well (Kimball 1960; Silverman 1995; Green and Boardsen 2006; Humble 2006; Silverman 2008; Cárdenas et al. 2016; Hayakawa et al. 2016; González-Esparza and Cuevas-Cardona 2018; Hayakawa et al. 2018b, 2019b).

As such, mid- and low-latitude aurorae have formed footprints of major magnetic storms and hence major geo-effective solar eruptions (Vallance Jones 1992; Shiokawa et al. 2005; Silverman 2006; Willis et al. 2006, 2009; Schlegel and Schlegel 2011). Therefore, such auroral reports have been one of the key pathways to understand solar–terrestrial interactions in the past in terms of their long-term variability and cyclicity (Silverman 1992; Usoskin et al. 2013, 2015; Lockwood and Barnard, 2015; Lockwood et al. 2016; Vázquez et al. 2016; Domínguez-Castro et al. 2016).

However, auroral phenomena have been rarely seen at low latitudes during moderate and even quiet magnetic conditions. Such events are known as sporadic aurorae (Silverman 2003). Botley (1963) introduced this term to the scientific community, citing previous descriptive usages of the same word by Abbe (1895). She introduced nine cases of low-latitude aurorae in Europe and in the Middle East observed in the twelfth and nineteenth centuries. Botley (1963) was the first to clearly define sporadic aurorae as “comprise such instances as a single ray in a sky otherwise seemingly clear of auroral light, or isolated patches well to the equatorial side of a great display”. Botley (1963) also noted references to reports of two low-latitude aurora occurrences without the occurrence of high-latitude aurorae (Fritz 1881; Eddie 1894).

It took another four decades for the next paper on sporadic aurora to be published. Silverman (2003) provided a survey of considerable sporadic aurora observations in low-latitude regions of the United States during a time span of over half a century, and highlighted the occurrence of sporadic aurorae in the context of mid- to low-latitude aurorae during moderate-to-low magnetic activity. That paper was later followed by other papers with reports on sporadic aurora sightings from Iberia and the Canary Islands (Vaquero et al. 2007; Vázquez and Vaquero 2010), East Asia (Willis et al. 2007), Mexico (Vaquero et al. 2013), and the Philippines (Hayakawa et al. 2018c). Interestingly, Shiokawa et al. (2005) reported three cases of instrumental observations of mid-latitude aurora in Hokkaido (Japan) under fairly moderate magnetic activity as well.

Silverman (2003) speculated that sporadic aurorae may be caused by localized and ephemeral magnetospheric energy input into the low-latitude ionosphere, but he does not clearly suggest any physical mechanisms that may explain this phenomenon. In fact, considering the known correlation between intensity of magnetic disturbance and equatorward boundary of auroral ovals (Yokoyama et al. 1998), Silverman (2003)’s comprehensive survey was striking and casted an open question on its physical mechanism. Hayakawa et al. (2018c) suggested that at least part of sporadic aurorae might have been caused by the impact of inclined interplanetary shocks (see also Oliveira et al. 2018; Oliveira and Samsonov 2018) that strike the magnetosphere in the pre-dusk sector. However, despite all these efforts, a comprehensive understanding of the causes of sporadic aurorae still remains an open question in space weather research.

The main goal of this article is to show an aurora observation report published in a Rio de Janeiro’s newspaper on 17 February 1875, hitherto unknown to the scientific community. Based on the event descriptions, the expertise of the observer, and the sporadic aurora characteristics presented in this introduction, as well as the magnetic latitude location of Rio de Janeiro and the low magnetic activity on the days before the observation, we will show that the event was most likely a sporadic aurora. The paper is structured as follows. “The observational site and the observer” section brings brief descriptions of the observational site and the observer. “The report and its interpretation” section introduces the report along with its interpretation based on current aurora knowledge. Finally, the paper is concluded in “Conclusion and a final remark” along with a final remark.

The observational site and the observer

The imperial observatory

Brazil was a Portuguese colony during the period 1500 to 1822. Due to military and commercial sanctions imposed by Napoleon to Lisbon in the beginning of the nineteenth century, the throne of the Portuguese Empire exiled from Lisbon to Rio de Janeiro in 1808 (Fausto 1994). Later, John VI of Portugal returned back to Lisbon and left his son Peter I as the ruler of the Kingdom of Brazil. Then, on 7 September 1822, Peter I proclaimed Brazil’s independence of Portugal, and became the first emperor of Brazil (Fausto 1994). In 1827, 7 years before his death, Peter I founded the Imperial Observatory (Morize 1987), today known as the National Observatory (Observatório Nacional), still located in Rio de Janeiro. After Peter I’s death, his son Peter II became the second and last emperor of Brazil, when it became a Republic on 15 November 1889 (Fausto 1994). Peter II was a monarch very interested in science who supported many contemporary scientists, and used the auspices of the Imperial Observatory for astronomical observations and scientific discussions (Benevides 1979).

Emmanuel Liais

The likely sporadic aurora reported here was observed from the facilities of the Imperial Observatory by the Frenchman Emanuel Liais (1828–1900) on 15 February 1875. Liais was the director of the Imperial Observatory in 1875, having been directly appointed by Peter II, after leaving the position as the director-adjunct of the Paris Observatory in France (Morize 1987). The observer was a professional nineteenth century scientist. While at the Imperial Observatory, Liais conducted research on astronomy with emphasis on planetary motion and comets, discovering one himself in 1860 (Liais 1860), he also published a popular book on astronomy (Liais 1865). Liais had considerable experience and expertise with optical physics and instrumentation. He published on the 1858 total solar eclipse observation from Brazil, being among the first to photograph the solar corona (Liais 1861; Aubin 2016), and apparently had a good understanding of atmospheric effects with respect to their altitude occurrences (Liais 1859; Muniz Barreto 1997). Liais also published on aurora observations from his home town Cherburg, France, on the Halloween day of 1853 (Liais 1853). More surprisingly, Liais even suggested methods to measure auroral altitudes, showing that aurorae occur far higher than meteorological phenomena (Liais 1851), as is well known today (e.g., Roach et al. 1960). According to Muniz Barreto (1997), these findings would have contributed to classify auroral phenomena as magnetic phenomena as opposed to meteorological phenomena if they had been published in a scientific journal with higher audience.

The report and its interpretation

Presentation of the report

Emmanuel Liais observed the aurora event on 15 February 1875 from Rio de Janeiro, Brazil. At that time, the Imperial Observatory was hosted by the Morro do Castelo (Castle Hill), an old church whose geographic coordinates are 22.75° S, 43.10° W. Liais took notes of his observations and wrote a report to the local Jornal do Commercio (1875, Commerce Newspaper). This report was found in the data base of the National Digital Library of the National Library of Brazil (http://bndigital.bn.gov.br), hereafter BNDigital). We transcribed the full text of early modern Portuguese with its original spelling and grammar style in Appendix A: A1 and translated it into English in Appendix A: A2.

The observer noted the occurrence of the aurora by 19:45 local mean time (LMT), or ~ 16:45 GMT (Greenwich Mean Time). He realized the presence of white light in the sky that was aligned with the terrestrial magnetic field. Later, he noted that the light rays, with variable intensities, moved from west to east and took some reddish color at the bottom and greenish color at the top. This display lasted for approximately 40 min.

While we certainly need to be careful for possible misinterpretation of atmospheric optics as aurorae (e.g., Usoskin et al. 2017; Stephenson et al. 2019), there are three more descriptions that strongly suggest that the phenomenon observed by Liais was a sporadic aurora. First, he mentions that some clouds later passed below the auroral rays. Since Liais knew that the aurora is formed above meteorological phenomena occurring in the troposphere (Liais 1859; Muniz Barreto 1997) based on his previous experience with aurora observations (Liais 1853, 1859, 1865), here it seems that he was very convinced the lights he observed were indeed auroral lights rather than meteorological phenomena associated with clouds. Secondly, its reported direction and coloration seem to rule out this kind of possible contamination. This phenomenon appeared from the direction of the magnetic needle inclination and generated white stripes in a meridian direction. This is consistent and typical with auroral ray structure, extending along the magnetic field line (e.g., Chamberlain 1961). This phenomenon shows reddish color and faint greenish color well after sunset (18:40 LMT), whereas the nighttime atmospheric optics caused by the Moon is too faint to obtain its color detected by human eyes (Minnaert 1993).

Even more decisively, Liais saw this phenomenon with his spectroscope and confirmed “the certain evidence of proper lights”. Spectroscopic observations frequently give us incontrovertible evidence to distinguish aurora from other atmospheric optics, as auroral spectra show emission lines, whereas spectra of solar reflected lights show absorption lines (see Fig. 1; e.g., Capron 1879, 1883; see also Love 2018; Stephenson et al. 2019). Therefore, we can incontrovertibly reject possible contamination of atmospheric optics or clouds with strange color.

Fig. 1
figure 1

From top to bottom, line spectra of oxygen, nitrogen, sulfur, and solar (reflected) light, with wavelengths in the horizontal axes. Whereas the auroral lights by oxygen and nitrogen show bright emission lines, the sunlight and its reflection (solar reflected light) show dark absorption lines

Full size image

Strangely, Liais interpreted the observed spectra as those of sulfur. This contradicts the modern understanding of auroral spectra as the present day understanding of the aurora shows that there are emissions mainly from oxygen and nitrogen (e.g., Gault et al. 1981; Chamberlain 1961). However, it is not Liais’ originality to associate auroral spectra with sulfur. Back in mid eighteenth century, van Musschenbroek (1762) suggested aurorae were partially caused by burning sulfur. Liais performed his observations with a spectroscope only 6 years after the earliest spectroscopic observations of the aurora (Ångström 1869). Even in the 1870s, Capron (1879) acknowledged this early hypothesis of “sulfurous vapors issuing from the earth” as a cause of the aurora, while Capron himself did not seem to agree with this supposed cause. Moreover, this misinterpretation may be justified by the similarity of spectra of sulfur emissions with spectra of mixed emissions of oxygen and nitrogen, as shown in Fig. 1. Since spectrum lines are considered a “fingerprint” of a source or object, this is a very important observation to distinguish aurorae form other optical phenomena. Therefore, most likely Liais was influenced by such early scientific discussions and misinterpreted the compound spectra of excited nitrogen and oxygen emissions as the emission spectrum of sulfur, as is understandable from the comparison of their spectra in Fig. 1.

Unlike aurorae, atmospheric optics or clouds cannot shine by themselves. The light source for clouds is sunshine or solar reflected light, including moonlight (e.g., Capron 1883; Love 2018; Stephenson et al. 2019). Therefore, as opposed to auroral emissions, the spectra of such atmospheric optics must inevitably involve dark absorption lines, typical with the sunlight (see Fig. 1; e.g., Capron 1879). As Liais saw this phenomenon with a spectroscope and associated it with aurora, we cannot associate this phenomenon with atmospheric optics, originated from the sunshine or solar reflected lights.

Modern interpretation of the report

There are only a few magnetic field observations that were regularly recorded around the world during the nineteenth century. The only magnetic indices that can be used for that period are the ak index (Nevanlinna 2004) and the aa index (Mayaud 1972). Unfortunately, there is no ak index for that date, but there is aa index for that date. The aa index is a 3-h time resolution magnetic index derived from two magnetic observatories in England and Australia that are nearly antipodal to each other (Mayaud 1972). The Aa index is then derived from the aa index by taking its daily averages. More detail of these indices can be found in the literature (Rostoker 1972; Mayaud 1980). The aa and Aa indices are provided by the British Geological Survey website.

Magnetic latitudes are computed by the geomagnetic field GUFM1 model (Jackson et al. 2000) from 1600 to 1990. This model is complimentary to the International Geomagnetic Reference Field (IGRF) model (Thèbault et al. 2015). IGRF can compute magnetic fields from 1900 onwards, but GUFM1 can compute magnetic fields as far back as 1590 due to the compilation of a massive data base obtained from observational logs compiled on ships at sea and ports around the world (Jackson et al. 2000; Jonkers et al. 2003).

The solid orange line in Fig. 2 shows the time evolution of Rio de Janeiro’s magnetic latitude (MLAT) from 1600 to 1990. The model shows that MLAT increased from − 18.4° in 1600 to its maximum value (the closest value to the magnetic equator) slightly above − 12° around 1816 when it started to decrease again. The highlighted light purple area (discussed later) corresponds to the 1781–1788 interval between aurorae observed from Rio de Janeiro. The dashed green vertical line marks the Carrington event occurrence (1859), while the dashed brown vertical line indicates the event reported in this letter (1875).

Fig. 2
figure 2

Time evolution of Rio de Janeiro’s magnetic latitude computed with the geomagnetic field model GUFM1 (Jackson et al. 2000). The light purple bar shows the period corresponding to Sanches Dorta’s magnetic and aurora observations in Rio de Janeiro during the period 1781–1788 (Vaquero and Trigo 2005, 2006). The green and brown vertical lines mark the Carrington event (e.g., Green and Boardsen 2006; Hayakawa et al. 2018b, 2019b) and the eventual sporadic aurora observation here reported

Full size image

Figure 3 shows the aa index for the interval 1–24 February 1875. The solid orange line indicates 3-h aa index (in nT), while the shaded green line indicates the daily averaged Aa index. The dashed blue line corresponds to the beginning of Liais’ observations (19:45 LMT or 1645 UT) reported to the Jornal do Commercio (1875). The plot documents that the A(a)a indices showed some weak/mild activity 2–4 days prior to the aurora observation, with maximum Aa around 27 nT. This magnetic activity is consistent with sunspot number observations recorded a few days before, with very low values and one day with the observation number of 60 (Clette et al. 2014; Clette and Lefèvre 2016). The low magnetic activity conditions during that sporadic aurora event are consistent with the description suggested by Silverman (2003).

Fig. 3
figure 3

British Geological Survey 3-h aa index (solid orange line) and the 24-h Aa index (solid green line/shaded green area) for the interval 1–24 February 1875. The dashed blue line indicates the onset of the aurora observation (15 February 1875 at 19:45 LMT/16:45 GMT) reported by Emmanuel Liais to the Jornal do Commercio (1875)

Full size image

Additionally, the results of this study may also explain the reason why great aurora displays observed from Brazil have not been found/reported in the contemporary records for the Carrington event yet. As seen in Fig. 2, Rio de Janeiro’s MLAT by 1859 was very low, around − 12.2°. While we surveyed auroral reports in Brazilian newspapers during the Carrington event in the BNDigital database, we found only references to great aurora displays and even telegraph system failures in North America and Europe, with nothing being reported as having been observed from Brazil.

Since the equatorward boundary of the auroral oval is reconstructed ~ 28.5–30.4° MLAT (Hayakawa et al. 2018b) assuming aurora height ~ 400 km (Roach et al. 1960; Ebihara et al. 2017), the expected elevation of auroral visibility would be at best 3° above the horizon and hence it was quite difficult to observe auroral displays in the sky of Rio de Janeiro (and other Brazilian locations) during the Carrington event. However, it would be worth surveying potential auroral reports in Argentine, Chile, and Uruguay, countries that are located in regions of higher MLATs in South America. Another significant event, the magnetic storm of 4 February 1872, also triggered great aurora displays at low latitudes (Silverman 2008; Hayakawa et al. 2018a). Silverman (2008) reported on possible aurora sightings in latitudes as low as 10° or even 3°, while he casted a caveat on their reliability. The author reported aurora sightings on the French Reunion Island, in the Indian Ocean (21.12° S, 55.54° E). We found mention to these aurora sightings during that storm in Brazilian newspapers while searching the BNDigital database, but none occurring from Rio de Janeiro or anywhere else in Brazil.

Another possibility is to interpret Liais’ optical observations as equatorial plasma bubbles (EPBs), which are structures with depleted plasma density usually formed after sunset in the bottomside ionosphere and move from west to east (Mendillo and Tyler 1983; Kelley 2009; Liu et al. 2017). Since plasma bubbles are faint structures that can be hardly seen by the naked eye (e.g., Wiens et al. 2006), Liais’ event most likely cannot be classified as an EPB event.

It should be mentioned that this is not the first report of an aurora observation performed from Rio de Janeiro, despite its proximity to the magnetic equator. In fact, Vaquero and Trigo (2005, 2006) and Carrasco et al. (2017) presented a series of magnetic observations and aurora sightings conducted by the Portuguese astronomer Sanches Dorta in the eighteenth century. According to the authors, the observations conducted by Sanches Dorta, during the period 1781 to 1788, must very likely have occurred during times of elevated magnetic activity. However, according to Fig. 2, the MLATs of Rio de Janeiro during these events were around − 12.5° (highlighted purple area). If the events reported by Vaquero and Trigo (2005, 2006) and Carrasco et al. (2017) were in fact caused by great magnetic storms, their visibility would have reached MLATs closer to the magnetic equator in comparison to the low-latitude Carrington aurorae previously reported (Green and Boardsen 2006; Hayakawa et al. 2018b, 2019a, b).

Conclusion and a final remark

Sporadic aurorae occur at low-latitude areas (Abbe 1895; Botley 1963) during moderate-to-low magnetic or even quiet conditions (Silverman 2003). However, despite being impressive, this space weather phenomenon is not very well known by the community. In addition, this phenomenon does not happen very often, and there are only a few publications reporting on sporadic aurora sightings (Abbe 1895; Boyer 1898; Botley 1963; Silverman 2003; Vaquero et al. 2007, 2013; Vázquez and Vaquero 2010; Willis et al. 2007; Hayakawa et al. 2018c).

In this letter, we presented for the first time a report on a possible sporadic aurora observation performed from Rio de Janeiro, Brazil, on 15 February 1875. This is the first sporadic aurora report in South America, and the second one in the southern hemisphere (the first observation was reported by Eddie 1894). Additionally, this is the second sporadic aurora observed near the magnetic equator. The original report was authored by Emmanuel Liais, then director of the Imperial Observatory of Rio de Janeiro, and published in the Jornal do Commercio (1875) of the same city. Given the scientific expertise, the contents of scientific descriptions and the experience of the observer, particularly with respect to the use of a spectroscope, Liais’ report may be considered credible and possible misinterpretation of the observed phenomenon, such as caused by atmospheric optics (Usoskin et al. 2017; Hayakawa et al. 2018c), may be discarded. The aurora description presented by Liais is consistent with sporadic aurorae (Abbe 1895; Botley 1963; Silverman 2003). In addition, the very low magnetic latitude of Rio de Janeiro and the weak/mild magnetic activity during the observations are consistent with a previous sporadic aurora observation near the magnetic equator (Hayakawa et al. 2018c).

Furthermore, in addition to the sporadic aurora causes presented in the introductory section, we speculate that sporadic aurorae may also be caused by the flow of solar wind phase fronts with some inclination in the equatorial plane toward the dusk flank. As suggested by Cameron et al. (2019), such flows of solar wind phase fronts during times of low magnetic activity or quiet conditions would increase magnetic activity over time due to shear and viscosity effects, and the sudden release of this energy may cause sporadic aurorae. More observations and possibly numerical simulations are needed in order to test these hypotheses and advance the knowledge of sporadic aurora triggering.