© The Authors 2025

1 Introduction

Meteoroid streams are formed by meteoroids on similar orbits originating in the same parent body, typically a comet. Young streams are compact and narrow and can be easily linked with their parent bodies on the basis of orbital similarity in most cases. After several thousand years, the streams become dispersed and their mean orbit can differ from the orbit of the parent body, especially if the parent body or a part of the stream is subject to a close encounter with Jupiter or another major planet (Vaubaillon et al. 2019). Meteoroid streams persist even if their parent comet completely disintegrates and remain then the only witness of the comet’s existence.

When crossing Earth’s orbit, meteoroid streams produce meteor showers. Meteor showers are of popular interest and a traditional subject of meteor astronomy. Studies of them provide information about the properties and history of their parent bodies and about the processes in the inner Solar System. Each meteor shower has its own unique history. The published information about meteor showers is compiled by the Meteor Data Center of the International Astronomical Union (IAU MDC) (Hajduková et al. 2023). The center maintains a list of established showers, whose existence has been confirmed (currently 110 showers in total), and a working list containing proposed showers. Regrettably, the parameters of many established showers remain poorly known, since the radiants and orbits published by different authors differ substantially. Although some of these differences may be real, for example because various techniques are sensitive to meteoroids of various sizes or because the showers evolve in time, most of them are due to limitations of data samples and observational uncertainties. A discussion about cross-correlation of meteor showers can be found in Jopek et al. (2024).

Here, we investigate the established shower κ Cygnids, listed in the IAU MDC as number 12 with the acronym KCG. The activity of the shower lasts for most of August and the shower is known to produce bright fireballs (Jenniskens 2006). Many authors have noted that other showers with radiants nearby are active at the same time and the situation is therefore complex (see Sect. 2). Other proposed showers include the August Draconids, which are now listed as the established shower 197 AUD. We used precise, recent data from the European Fireball Network (EN) to study these showers independently of previous analyses.

The overview of past observations of κ Cygnids and related showers is provided in Sect. 2. Our goals and data procedures are explained in Sect. 3. The groups of radiants we identified are defined in Sect. 4 and the properties of these potential showers are described in Sects. 5–7. The cross-identification of our groups with previously reported showers is given in Sect. 8. Section 9 provides the refined parameters of the showers based on our data. Physical properties of the meteoroids are evaluated in Sect. 10. The article is then finished by discussion and conclusions.

2 History of κ Cygnid observations

Recognition of the κ Cygnid meteor shower is connected with visual observations in Great Britain. The first strong manifestation was reported by Denning (1879a,b), who observed an extraordinary display from a radiant located at (RA, Dec) = (291°, +60°) on August 21–23, 1879. He noted that the shower was not new. Meteors were observed from similar radiants in some earlier years too, though probably with a lower frequency than in 1879. Another remarkable display was observed by the same author and other observers in 1893 (Denning 1893). Denning did not connect the two displays. In his catalog of meteor shower radiants (Denning 1899), he treated these two outbursts as separate showers. The 1879 shower was called o Draconids with the radiant at (291°, +60°) and activity on August 21–25, while the 1893 shower was called θ Cygnids with the radiant at (292°, +53°) and activity on August 4–16. Later, he noted the activity of o Draconids on August 15, 1907 Denning (1907).

Another similar shower with remarkable activity and a radiant near α Lyrae (280°, +44°) was reported by several observers on August 8–23, 1914 (Davidson 1914). The θ Cygnids were active at that time as well. The connection between the two showers was discussed, with similar orbital elements being noted. High activity of θ Cygnids, including fireballs with bright terminal flashes, was further reported on August 15–25, 1922 (Cook 1924). In that report, high activity on August 8–12, 1901, was also mentioned, though the original report by Besley (1901) just listed the shower among those observed in August 1901. Cook (1924) quoted Mr. Denning considering the shower to have a period of about 7.5 years, probably the first such note in the literature. Fireball presence was also mentioned in 1929 but the overall activity was not very high (Prentice 1929).

The Harvard photographic program provided the first five instrumentally determined orbits of the members of the shower, which was in the meantime renamed to κ Cygnids (Whipple 1954). The orbits were obtained between August 9–22, 1950, during another year of high activity. The obtained semimajor axes correspond to orbital periods of 6.6–10 years. Letfus (1955) computed the orbits of four of these meteors independently and obtained periods of 5.5–9 years.

Since the work of Whipple (1954), κ Cygnids have been considered as a confirmed shower but no special attention was devoted to them in the following four decades. They were listed in the catalog of meteor showers by Cook (1973) with a geocentric radiant at (286°, +59°) and an orbital period of 5.3 years. They were also detected among faint radar meteors by Sekanina (1973) with an orbital period of only 4.2 years.

A computer search among 3518 published photographic meteor orbits by Lindblad (1995) identified 37 “Cygnid” meteors, which could be split into four separate showers when the search was refined. One of them was κ Cygnids, which contained nine meteors. Five of them were observed in 1950, and one in 1957, 1958, 1978, and 1993 each. The dates span August 12– 22. The mean radiant was at (286°, +55°) and the mean orbital period was 7.2 years, with the range from 5.3 to 10 years. The other three similar showers were called α Lyrids, ζ Draconids, and August Lyrids by Lindblad (1995).

In between, another outburst was observed visually and photographically by the members of the Dutch Meteor Society (DMS) in 1993 (Jenniskens 2006). Jenniskens (2023) mentioned other outbursts observed by the DMS visually in 1978 and 1985.

Jenniskens (1994), compiling visual observations by two groups in 1981–1991, listed κ Cygnids among annual showers with a ~30 day activity peaking at solar longitude λ_⊙ = 146° (~August 19) with a zenith hourly rate (ZHR) of 2.3.

Similar to Lindblad (1995), Jones et al. (2006) performed a computer search of κ Cygnid-related meteors among now more extensive meteor catalogs. They confirmed the four Lindblad’s substreams of what they called a “Kappa Cygnid meteoroid complex” and added another one named γ Draconids. They also noted that there are no gaps between the locations of radiants of adjacent substreams. When the IAU MDC, following Jenniskens (2006), prepared a numbered list of meteor showers with three- letter acronyms in 2007–2009, only κ Cygnids were included (as 12/KCG) from these five showers. The name ζ Draconids was reused for another shower (73/ZDR) proposed by Molau & Rendtel (2009) to be active at the end of July.

Another outburst of bright κ Cygnids was observed in 2007 (Jenniskens & Trigo-Rodriguez 2007). Trigo-Rodriguez et al. (2009) published the radiants and orbits of nine meteors. They also noted low tensile strength of the meteoroids, normal chondritic composition derived from one spectrum, and the absence of faint meteors (magnitude > +4) in the shower. Nevertheless, visual observations of one of us (TW) on eight nights indicated that about 20% of κ Cygnids were of magnitude +4 or fainter.

SonotaCo (2009) published a meteor shower catalog based on 240 000 single-station video meteors observed in 2007–2008. κ Cygnids are listed with the largest radiant area of all showers, having a radius of 10°.

Koseki (2014), using video data from the SonotaCo’s network, pointed out high activity of κ Cygnids in 2007 in comparison with the years 2008–2012. He also compiled literature data and introduced the term “Cygnids–Draconids Complex” (C–D Complex). He divided video radiants within the complex into seven groups, A–G. Group A was in fact active at the end of July and the orbits had a much larger eccentricity than the other groups. It could be identified with a distinct shower, the July γ Draconids (184/GDR). Groups B and C were the ones that exhibited the high activity in 2007. The author was not sure if this activity was the same shower as the activity observed photographically in 1950 (Whipple 1954). He nevertheless expected high activity in 2014. Group G was identified with the ζ Draconids of Lindblad (1995).

The outburst predicted for 2014 really occurred and was observed visually and by radar, video, and photographic techniques (Rendtel & Molau 2015; Moorhead et al. 2015). Rendtel & Molau (2015) found that the enhanced activity was produced by radiants corresponding to κ Cygnids and α Lyrids defined by Lindblad (1995), not by his ζ Draconids. This result is consistent with the analysis of the 2007 outburst by Koseki (2014). Moorhead et al. (2015) presented 21 good-quality photographic orbits from the Czech part of the EN and analyzed them together with video data from two North American networks and radar data from the Canadian Meteor Orbit Radar (CMOR). The orbits were characterized by wide range of semimajor axes with concentrations near the 5:3, 2:1, 9:4, and 3:1 resonances with Jupiter. The 5:3 resonance with a 7.116-year period, corresponding to the apparent shower activity period, was discussed in more detail. Only a relatively small fraction of the dataset fell into this resonance so it seemed that it played only a small role in the outburst. The radar activity in 2014 was found to be five times higher than in 2007 and nearly ten times higher than in all other years between 2002 and 2013. Video observations in 2015 showed the activity at the background level again (Molau et al. 2015).

Jenniskens et al. (2016a) analyzed many meteor showers from the video observations by the Cameras for Allsky Meteor Surveillance (CAMS) project. The data for κ Cygnids cover the years 2011 and 2012, when no outburst occurred. An extended band of radiant concentrations, bent in the RA–declination plot, was observed. The activity moved from the eastern to the western part during August. The authors divided the meteors into four showers, κ Cygnids (12/KCG), August Draconids (197/AUD), August µ Draconids (470/AMD), and ι Draconids (703/IOD), but noted that the division may not be unique. August Draconids were identified from radar data already by Sekanina (1976). August µ Draconids were first defined on the basis of six meteors found by cluster analysis in the SonotaCo and CAMS video databases by Rudawska & Jenniskens (2014). They were also found in the European video database (EDMOND) by Kornoš et al. (2014).

Rendtel & Arlt (2016) reanalyzed the visual observation reports from the years 1975–2015 and found a clear activity enhancement only in 1985 and 2014. Hints of higher activity were present in 1978, 2008, and 2011, while the outbursts observed by other techniques in 1993 and 2007 were not apparent in visual data.

Shiba (2017) analyzed video data of the SonotaCo network from 2007 to 2016. Based on the radiant position, he distinguished only two showers: κ Cygnids (which also included α Lyrids) and August Draconids (which also included August µ Draconids and ι Draconids). Later, Shiba (2022) also included ν Draconids, discussed by Jenniskens et al. (2016b), as part of the August Draconids. The κ Cygnids showed activity eight times more enhanced than in normal years in both 2007 and 2014. The activity in 2013 was found to be two times enhanced and occurred at lower solar longitudes. The orbital periods of κ Cygnids were generally longer than corresponding to the 5:3 resonance but suffered from observational errors.

Koseki (2020), as a preparation for the expected return in 2021, studied the radiants in the κ Cygnids region in three video databases. Besides κ Cygnids themselves, he identified two other showers. The first one was named August ξ Draconids by him (the name was not recognized by the IAU MDC) and was identical to Group F (also designed KCG3) in his previous work (Koseki 2014). The second one was called ζ Draconids following the earlier nomenclature of Lindblad (1995) (distinct from the 73/ZDR) and was identical to August Draconids (197/AUD) of Jenniskens et al. (2016a).

The high activity of kappa Cygnids in 2021 was already reported on August 9 in a telegram by Jenniskens (2021). The outburst was observed visually and by video (Miskotte et al. 2022).

3 Our goal and methods

It is obvious that many open questions about κ Cygnids remain. There are strong suggestions that a meteoroid swarm in a 5:3 resonance in Jupiter is responsible for periodic enhancements of activity. That hypothesis was, however, not directly proven by measurements of orbital periods of the meteoroids. Little is known about the annual component of κ Cygnids and its relation to other showers active in August with similar radiants. There is not even any consensus about the number of these similar showers. In his latest book, Jenniskens (2023), following Shiba (2017, 2022), considers only August Draconids. On the other hand, the IAU MDC¹ keeps more showers. August Draconids (197 AUD) are listed as an established shower, August µ Draconids (470 AMD) and ν Draconids (220 NDR) are marked as nominated to be established, and ι Draconids (703 IOD) are on the working list.

Our goal is to study κ Cygnids and other nearby showers independently using new data from the EN. The EN data were already used for the study of κ Cygnids in 2014 (Moorhead et al. 2015). At that time, the analog Autonomous Fireball Observatory (AFO) was used as the main instrument but the new Digital Autonomous Fireball Observatory (DAFO) was already installed at six stations. As of 2024, DAFOs, which are more sensitive and more efficient, are used as the main instrument at 20 stations. DAFOs take two high-resolution digital images of the whole sky per minute and are equipped with LCD shutters to measure meteor velocities. They also contain radiometers, which provide high-resolution light curves of brighter meteors with up to 5000 samples per second. More details about the network and the data procedures used can be found in Spurný et al. (2017) and Borovička et al. (2022a).

The data were analyzed without previous assumptions about meteor showers, by just selecting the period of activity and broad radiant area. All meteors observed in the month of August in the years 2016–2024 and that have radiants in the region of interest were considered. The region of interest included western Cygnus, southern Draco, Lyra, and eastern Hercules. It is plotted in ecliptic coordinates in Fig. 1. Meteors too short for DAFOs to be included in our catalog (Borovička et al. 2022a) were also considered. Additional measurements were done on images from SDAFOs, the spectral versions of DAFOs, and on videos from the supplementary video cameras (IP cameras). SDAFOs are more sensitive than DAFOs, since they use lenses with longer focal lengths and do not use LCD shutters. They can be therefore used to improve the trajectory and radiant solution. Even more useful are the IP cameras. Because of their higher sensitivity (in comparison with DAFO), they capture meteors earlier and can be used to improve the solution of initial velocity. IP cameras started to be deployed in 2016. Nowadays, there are batteries of them that together cover the entire sky at two stations and each other station has at least one IP camera with a typical field of view of 100° × 55°.

The limiting sensitivity of the network depends on the position of the meteor. Figure 2 shows that meteors with a maximum magnitude of −2 or even fainter could be analyzed if they were close to one of the stations (i.e., visible near zenith). Bright meteors (fireballs) could also be observed if they were further away. The brightest recorded fireballs reached a magnitude of −12, typically in a flare.

The total number of analyzed meteors is 179. In the following section, we try to identify groups of meteors belonging to the same meteor shower on the basis of the radiant position and elements of the heliocentric orbit.

Fig. 1 Positions of geocentric radiants of the analyzed meteors in Suncentered ecliptical coordinates in polar projection. The shaded area is the investigated region of interest. The radiants of meteors that were assigned to one of five groups are plotted with colored symbols. Other radiants are plotted as crosses.

Fig. 2 Meteor maximum absolute magnitude plotted against the distance of the meteor from the nearest observing station.

4 Defining meteor groups

The radiant plot in Fig. 1 shows that radiants are not distributed randomly within the region of interest. We tentatively defined five meteor groups. Although the radiant position was the primary criterion, some boundary cases were decided according to additional criteria described below.

Considering the radiants, Group 1 (blue in Fig. 1) is relatively well defined, although the radiant area is very elongated, spanning more than 20^o in ecliptic latitude. Groups 2, 3, and 4 also correspond to obvious radiant concentrations but are not well separated each from other. Group 5 is a rather spurious radiant concentration and contains only six members. Meteors marked as Group 0 do not belong to any of the five groups and are probably sporadic.

The next criteria for group assignment are the day of appearance, expressed as the solar longitude, and the semimajor axis or, alternatively, the orbital period. Figure 3 shows the period as a function of the solar longitude. No concentration in the solar longitude can be seen. Nevertheless, Group 3 was more active in the first half of August and Group 4 was active only in the second half of August. Since we limited our analysis to meteors observed in August, we have possibly missed part of Group 4, which may be active at the beginning of September.

Group 1 is clearly separated from other groups in terms of the orbital period. We defined the core of each group by a range of orbital periods. For Group 1, it was 6.5–8.2 yr, for Group 2, 4.5–6 yr, and for Groups 3–5, 4.8–6 yr. Meteors falling outside these ranges are marked as Groups 1A to 5A and are plotted by open symbols in all graphs. They may be either normal members of the groups with an inaccurately determined period, scattered members with a modified period, or random interlopers that do not belong to the group. We cannot distinguish between these possibilities.

An interesting pattern occurs when plotting the perihelion distance and inclination (Fig. 4). There is a clear correlation between these two quantities for Group 1. As the perihelion distance increases from 0.94 to 1 AU, the inclination also increases from 28° to 44°. This plot can be used to identify possible members of Group 1. Groups 2–5 together cover a similar range of perihelion distances but the inclination for a given perihelion distance is always lower. Group 4 overlaps with Groups 2 and 3 in this plot.

An alternative plot is between the argument of perihelion and the inclination (Fig. 5). All groups can be separated. This plot reflects to some extent the radiant distribution in Fig. 1.

Fig. 3 Orbital periods as a function of solar longitude. Formal one- sigma errors of periods are indicated. Only periods within 3–10 years are shown. Seven meteors had periods outside of this range; the full range was 1.3–15 years. Solar longitudes correspond to the month of August. Long-dash colored lines show the limits of periods for the cores of individual groups. Short-dash black lines mark the 5:3 and 7:3 resonances with Jupiter.

Fig. 4 Relation between perihelion distance and inclination. Groups are coded as in Fig. 1 and marked by colored ellipses.

Fig. 5 Relation between argument of perihelion and inclination. Groups are coded as in Fig. 1.

5 Understanding the geometry of the orbits

The orbits of all groups have perihelia close to Earth’s orbit (0.94–1.015 AU) and moderate inclinations of 20–45°. The meteoroids encounter the Earth in the descending node when crossing the ecliptic from top to bottom. The projections of representative orbits from Groups 1–4 into the ecliptic plane are shown in Fig. 6. Group 5 is similar to Group 3 in this projection.

The aphelia of Group 1 lie well behind the orbit of Jupiter. The ascending node is close to Jupiter’s orbit. The crossing with Earth’s orbit occurs before reaching perihelia. The perihelion points lie below the ecliptic.

All other groups have aphelia inside of the Jupiter’s orbit but relatively close to it. Group 2 has ascending nodes at aphelia and descending nodes at perihelia. The perihelia lie very close to Earth’s orbit. Since the meteoroid radial speed (toward the Sun) is zero during the Earth encounter, and the longitudinal speed (along Earth’s orbit) is close to Earth’s orbital speed, the geocentric radiants are very close to the north pole of the ecliptic.

Groups 3 and 4 are nearly symmetrical. The perihelia lie just inside of Earth’s orbit in both cases and the inclinations are nearly the same. Group 3 has ascending nodes before aphelia and descending nodes before perihelia. The perihelion points lie below the ecliptic. Group 4 has ascending nodes after aphelia and descending nodes after perihelia. The perihelion points lie above the ecliptic.

To explain the correlation between the perihelion distance, q, and inclination, i, in Group 1, three orbits with different q are plotted in Fig. 7 in a side view. The perihelion points lie inside of the Earth’s orbit below the ecliptic. To intersect Earth’s orbit and thus be observable, the orbits with larger q must have larger i than those with lower perihelion distances. The observed q–i correlation thus seems to be a selection effect from the condition of the Earth intersection. Orbits with other combinations of q and i probably exist in the stream but cannot be observed on Earth.

There is a smooth transition between Groups 2, 3, and 4. All groups have similar semimajor axes and are probably parts of a single meteoroid stream. Group 2 has slightly higher inclinations than the other two groups (Fig. 4) but this may again be a geometric condition of the Earth intersection. Group 5 is less compact and not well represented in our sample but may also be part of the same stream. Its semimajor axis is similar to Group 3 and it is close to Group 3 in Figs. 4 and 5.

6 Orbital periods and resonances

Determining precise semimajor axes and orbital periods of cometary meteoroids from meteor observation is a challenge. Small errors in the measurement of the entry speed result in large errors of semimajor axes. Since fragile cometary meteors usually have short atmospheric trajectories, precise speed determination is difficult. Previous studies failed to confirm the suspected orbital period of κ Cygnids of ~7 years, which was based on the periodicity of the activity. Our data are more precise but there are still appreciable uncertainties for the orbital periods of individual meteoroids.

In Fig. 8, a histogram of orbital periods of meteoroids in Groups 1 to 4 is given. No period cutoff was applied, so all meteoroids falling into the group on the basis of the radiant position are included in the plot. Group 5 was omitted because of the small number of meteoroids.

As previously, we can see a clear separation between Group 1 and other groups. There is a nice distribution of periods of Group 1, with the peak corresponding to the 5:3 resonance with Jupiter (7.12 yr). The average periods for all groups are given in Table 1. Only the cores of the groups were taken into account here. When including outliers, the mean orbits are almost the same, but the errors are much larger (7.18 ± 0.52 yr for Group 1; for other groups, the errors are between 0.6 and 1 year).

The periods of Groups 2–4 (and also Group 5) are the same within the uncertainties. They may fall into one or more higher- order resonances with Jupiter, namely 7:3 (5.08 yr), 9:4 (5.27 yr), or 11:5 (5.39 yr), but it is not certain if the orbits are indeed resonant. There is no evidence for a peak at the 2:1 resonance (5.93 yr) and also the 5:2 resonance (4.75 yr) is probably absent.

7 Periodic activity

The numbers of meteors of each group observed in individual years are shown in Fig. 9. To be sure that only true members of the showers were considered, the outliers were excluded. It is immediately obvious that the activity of Group 1 occurred almost exclusively in 2021. From 42 meteors classified as the core of this group, 37 appeared in 2021. Two were observed in 2024 and three in other years. All these three lie rather close to the boundary of Group 1 in Fig. 4, so they were not typical members of the group. From six outliers in this group with orbital periods outside the selected range (not shown in Fig. 9), five appeared in 2021, suggesting that they were related to the group.

The prevalence of the year 2021 in Group 1 is consistent with the 7 year periodicity of κ Cygnids. The previous years of high activity were 2014 and 2007 (see Sect. 2). We plotted the q-i graph for the 2014 EN data (Moorhead et al. 2015), 2007 Spanish data (Trigo-Rodriguez et al. 2009), and 1950 Harvard data (Whipple 1954) in Fig. 10. Of 21 fireballs observed in 2014, 18 nicely fit into the Group 1 region. Group 1 was therefore responsible for the enhanced activity in 2014. The fit is not so nice for the 2007 data, probably due to the lower quality of these data. Still, most data points lie within or close to the Group 1 region and we can conclude that this group was also responsible for the 2007 activity. We have not compared the semimajor axes, since they are less reliable in older works. The five 1950 orbits fall into Group 1 well. The year 1950 does not align with the 7-year periodicity but the period is in fact slightly larger than 7 years (see Sect. 11.1).

None of the other groups show such concentration as Group 1 (Fig. 9). There is a hint of a fluctuation with a minimum in 2019– 2020, but the statistics is poor. Moreover, the overall number of detected meteors in August 2019 was lower than in other years due to worse observing conditions. Year-to-year variations in the activity of Groups 2–4 can therefore be neither confirmed nor ruled out at this stage.

Fig. 6 Projections of typical meteoroid orbits from Groups 1 and 2 (left) and 3 and 4 (right) into the ecliptic plane. The parts of orbits above ecliptic are shown by thick line. The orbits of planets Mercury to Jupiter are drawn in black. The Earth’s orbit is thicker. Vernal equinox is to the right. All planets and meteoroids move counterclockwise. Note the different positions of nodes relative to aphelion in Groups 2–4.

Fig. 7 Three representative orbits of Group 1 meteoroids in a threedimensional plot with Earth’s orbit. The positions of perihelion points are depicted as circles. The parts of orbits below the ecliptic are shown by a thin line.

Fig. 8 Histogram of orbital periods of meteoroids of Groups 1 to 4, including the outliers. Positions of selected mean motion resonances with Jupiter are also indicated.

Fig. 9 Histogram of year of appearances of meteors belonging to the cores of Groups 1 to 4. The four core members of Group 5 are not shown. They appeared in four different years (2020–2022, 2024). The solid blue line denotes the total number of meteors observed in August each year. The corresponding scale is on the right.

Fig. 10 Relation between perihelion distance and inclination for the 2014 κ Cygnid data from the EN, 2007 data from the Spanish Meteor Network, and 1950 data from Harvard cameras. Groups are marked as in Fig. 4. One 2014 fireball with q = 0.916 AU is not plotted.

8 Identification with previously reported showers

In this section, we identify our Groups 1–5 with meteor showers reported previously in the literature under various names. We primarily use the radiant positions and velocities. Because of very large sizes of radiant areas and low precession rate near the ecliptic pole, we ignore the differences between different equinoxes and take the mean shower radiants directly as published in original papers.

The radiant positions and velocities are compared graphically in Fig. 11. We can see that the observed meteor radiants and the reported showers largely overlap. The velocities are also consistent in most cases. The radiant derived for κ Cygnids by most authors overlaps with the center of Group 1. Only the radar- determined radiant of Sekanina (1973) is off. The radiant area of Group 1 is very extended in declination and also encompasses o Draconids of Denning (1879a) to the north and α Lyrids of Lindblad (1995) to the south. Groups B to D of Koseki (2014) are also parts of our Group 1. There is a clear correlation between declination and velocity. It is investigated in the next section.

The transition between Groups 2, 3, and 4 is smooth not only in terms of the radiant position but also in terms of the velocity. Similarly to Group 1, radiant positions and velocities are correlated. In contrast, while the radiants of Groups 1 and 3 are also close, there is a jump in velocity between them. This is more evidence that Groups 2–4 are branches of the same stream and Group 1 is a separate stream.

The radiant of κ Cygnids reported by Jenniskens et al. (2016a) falls within Group 3. This is not surprising because the authors used data from 2011–2012 when Group 1 was probably not active (or only marginally active). Also, the κ Cygnid radiant of Porubčan & Gavajdová (1994) falls within Group 3. They probably included members of both Groups 1 and 3 in their computer search. Group 3 further includes August Lyrids of Lindblad (1995), Koseki’s Group E, and the γ Draconids of Jones et al. (2006). Koseki’s Group F lies at the boundary of Groups 2 and 3. Group 2 contains August Draconids (197/AUD), the ζ Draconids of Lindblad (1995), and Koseki’s Group G. Also, the late activity of κ Cygnids observed in 2012 by Molau et al. (2012) was in fact Group 2. Finally, the August µ Draconids (470/AMD) reported by Jenniskens et al. (2016a) fall within our definition of Group 2. The radiants and velocities of the same shower found by Rudawska & Jenniskens (2014) and Kornoš et al. (2014) lie outside our groups but close to both Groups 2 and 4. There is no direct overlap of any radiant with Group 4. The ι Draconids (703/IOD) lie farther in that direction.

Group 5 corresponds well with κ Lyrids (464/KLY). This shower, reported by Rudawska & Jenniskens (2014), Kornoš et al. (2014), and Jenniskens & Nénon (2016), has maximum activity in July and was therefore not completely covered by our analysis. It will therefore not be discussed in more detail here. Similarly, the µ Lyrids (413/MUL) and ν Draconids (220/NDR) (= ξ Draconids of Sekanina 1976) have similar types of orbits as our groups but are mostly active in July and September, respectively, and cannot be confirmed here. However, it is clear that there are other showers of the proposed C-D Complex active in August besides the κ Cygnids, but they cannot be distinguished on the basis of radiants and velocities. It was therefore legitimate that Jenniskens (2023) kept only κ Cygnids and August Draconids and removed all other showers mentioned here.

9 Refined parameters of the showers

9.1 κ Cygnids

Based on the comparison with previous work, Group 1 was identified with the κ Cygnid meteor shower (Sect. 8). Here, we present refined parameters of that shower based on our data. The usual meteor shower parameters are the mean radiant, radiant daily motion, geocentric velocity, and heliocentric orbit. However, it is not possible to give such a simple set of parameters for κ Cygnids. On the one hand, there is a nice dependency of the right ascension on the solar longitude (Fig. 12). Two of the three most deviating meteors were observed outside the year 2021 (in 2016 and 2018). On the other hand, there is no such dependency for declination (Fig. 13). While the mean declination increases by about 15° in 20 days, there is also a random scatter of about ±12° at any given time. To describe the deviation of declination from the mean value, we introduced a variable, p ∈ <−1,1>. The value p = 0 corresponds to the mean value, while the values p = ±1 are for the extreme negative and positive deviations. Depending on the solar longitude, λ_⊙ ∈ <130°, 154°>, and p, the positions of the κ Cygnid radiant can be then written as $${\alpha }_{\text{g}}={286}^{\circ }.5+0.45\left({\lambda }_{\odot }-{142}^{\circ }\right)$$(1) $${\delta }_{\text{g}}=+{51}^{\circ }.4+0.73\left({\lambda }_{\odot }-{142}^{\circ }\right)+{12}^{\circ }p,$$(2)

where α_g and δ_g are the geocentric right ascension and declination, respectively. The solar longitude 142° (~August 15) corresponds to the middle (and the maximum) of the activity.

The variable p can describe the variations not only in declination but also in other parameters. There is a pronounced correlation between declination and velocity (Fig. 14). Perihelion distance, inclination, and argument of perihelion are also correlated with declination. We first fit these parameters as a function of solar longitude to obtain the expected mean values at a given time. Then, the deviations of actual parameters from the expected means were plotted for individual meteors as a function of p computed from Eq. (2). Linear dependencies were found in all cases. The maximal deviations for p = ±1 were then found. Only the core meteors observed in 2021 and 2024 were used.

The resulting equations for the velocity and orbit of the κ Cygnid shower are: $$v_{\text{g}}=23.36+0.18\left({\lambda }_{\odot }-{142}^{\circ }\right)+3.4p$$(3) $$a=3.74\pm 0.12$$(4) $$q=0.9739+0.00119\left({\lambda }_{\odot }-{142}^{\circ }\right)+0.031p$$(5) $$i={35}^{\circ }.55+0.41\left({\lambda }_{\odot }-{142}^{\circ }\right)+7^{\circ }.7p$$(6) $$\omega ={204}^{\circ }.1-0.46\left({\lambda }_{\odot }-{142}^{\circ }\right)-{10}^{\circ }.4p$$(7) $$\text{Ω}={\lambda }_{\odot },$$(8)

where υ_g is the geocentric velocity in kilometers per second, a is the semimajor axis, q is the perihelion distance (both in astronomical units), i is the inclination, ω is the argument of perihelion, and Ω is the longitude of the ascending node (all in equinox J2000.0). The semimajor axis is the same for all meteors. No trend was found. The given value is the average of 39 meteors. The 5:3 resonance is at 3.70 AU, well within the error bar. The value of p for a given meteor can be computed from any of the equations that contain p. The result should be similar (within about 0.2 in most cases). A nearly normal (Gaussian) distribution of p values around zero was found (Fig. 15). In exceptional cases, the absolute value of p can exceed unity.

Five representative radiants and orbits for selected values of p are given in Table A.1. As it is customary, the orbits are given for the maximum of the activity. They differ just by p. While the dependence on λ_⊙ expresses the changes along the Earth’s orbit, the dependence on p describes the dispersion in perpendicular direction. We note that different orbits than the ones given here may exist in the meteoroid stream. The shower data contain only orbits that intersect Earth’s orbit.

Table A.1 also contains detailed data on showers from the literature, which could be identified with κ Cygnids (Sect. 8). We note that authors who performed a computer search in the orbital section of the IAU MDC (Porubčan & Gavajdová 1994; Lindblad 1995; Jopek et al. 2003; Jones et al. 2006) partly used the same data, including those of Whipple (1954).

Fig. 11 Geocentric radiants of meteors belonging to the cores of Groups 1–5 (color squares) plotted on the background sky map. Geocentric velocity is color-coded. The approximate limits of the groups are plotted as black ellipses or other shapes. Also plotted are radiants of various showers from the literature (color circles) with velocities (if known) coded with the same colors. The showers are labeled according to the names used by the authors and the paper numbers as given in Tables A.1–A.4 (number 19, not listed in the Tables, belongs to Jenniskens et al. 2016b). Only paper numbers are given for κ Cygnids. The other shower names are as follows: tCy (θ Cygnids) oDr (o Draconids), zDr (ζ Draconids), gDr (γ Draconids), xDr (ξ Draconids), ADr or AUD (August Draconids), AMD (August µ Draconids), IOD (ι Draconids), NDR (ν Draconids), aLy (α Lyrids), ALy (August Lyrids), MUL (µ Lyrids), KLY (κ Lyrids), and GB to GG (Groups B to G of Cygnid-Draconid Complex).

Fig. 12 Geocentric right ascension of the radiant as a function of solar longitude. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1 and 3 (solid lines). Mean values are plotted for Groups 2 and 4 (dashed lines).

Fig. 13 Geocentric declination of the radiant as a function of solar longitude. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1 and 3 (solid lines). For Group 1, limits of ±12° in declination from the fit are also drawn (dotted lines). Mean values are plotted for Groups 2 and 4 (dashed lines).

Fig. 14 Correlation between geocentric declination and velocity. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1–4 (solid lines).

Fig. 15 Histogram of values of p variable (computed from Eq. (2)) for 39 members of Group 1 (κ Cygnids).

9.2 August Draconids

Based on the similarity of the orbits, in particular the semimajor axis, and the absence of clear boundaries in radiant position and velocity, Groups 2–4 can be considered as a single meteor shower. Following Shiba (2017) and Jenniskens (2023), we call the shower August Draconids (197/AUD). Nevertheless, the groups can be still recognized according to some characteristics. They can be considered as three branches of the August Draconids. Some showers, for example Taurids, have a northern and southern branch. Northern Taurids have the radiant north of ecliptic and meet the Earth at the descending node. The opposite is true for Southern Taurids. Both branches meet the Earth on the way to perihelion. The perihelion point of Northern Taurids lies below the ecliptic plane (to the south) and the perihelion point of Southern Taurids lies above the ecliptic.

In the case of August Draconids, all radiants lie well north of the ecliptic, in fact close to the pole of the ecliptic. All branches meet the Earth at the descending node. As is shown in Sect. 5, Group 3 meets the Earth before reaching perihelion and the perihelion lies below the ecliptic. Group 4 meets the Earth after reaching perihelion and the perihelion lies above the ecliptic. Group 2 meets the Earth near perihelion, which lies close to the ecliptic. According to the location of the perihelion, we call Group 3 the lower branch, Group 4 the upper branch, and Group 2 the middle branch.

The distinction between branches can be done either according to the radiant position (Fig. 11) or according to the argument of perihelion. The middle branch as defined by us has ω between 173° and 192° (Figs. 5 and 16).

9.2.1 Lower branch

The radiants of the lower branch of August Draconids (Group 3) are tightly clustered in right ascension (Fig. 12) but scattered in declination (Fig. 13). In both cases, an increase with solar longitude is observed. The situation is similar to that for the κ Cygnids. In fact, the geometry of the orbits is also similar. The Earth encounter occurs before the perihelion and the perihelion point lies to the south of the ecliptic. Similarly to the κ Cyg- nids, the scatter in declination can be described by a variable, p, reaching the values (approximately) from −1 to +1 with a normal distribution around zero. The scatter in geocentric velocity, perihelion distance, inclination, and argument of perihelion can be also expressed by the parameter p. We have found the following relations: $${\alpha }_{\text{g}}={276}^{°}1+0.20\left({\lambda }_{\odot }-{140}^{°}\right)$$(9) $${\delta }_{\text{g}}=+{50}^{°}.6+0.60\left({\lambda }_{\odot }-{140}^{°}\right)+8^{°}p$$(10) $$v_{\text{g}}=20.12+0.10\left({\lambda }_{\odot }-{140}^{°}\right)+2p$$(11) $$a=3.08\pm 0.09$$(12) $$q=0.9926+0.00090\left({\lambda }_{\odot }-{140}^{°}\right)+0.012p$$(13) $$i={30}^{°}.60+0.26\left({\lambda }_{\odot }-{140}^{°}\right)+4^{°}.7p$$(14) $$\omega ={198}^{°}.0-0.49\left({\lambda }_{\odot }-{140}^{°}\right)-5^{°}.3p$$(15) $$\text{Ω}={\lambda }_{\odot }.$$(16)

The members of this branch were observed in the interval of solar longitudes, λ_⊙, from 130° to 148°.

The spread of the radiants and orbits as a function of p is lower than in κ Cygnids. We therefore only give three representative sets in Table A.2. The showers from the literature that can be identified with this branch are also listed there.

Fig. 16 Relation between argument of perihelion and perihelion distance for three branches of August Draconids.

9.2.2 Middle branch

The radiant of the middle branch of August Draconids (Group 2) does not show any clear dependence on solar longitude. The spread in both the right ascension and the declination is larger than any possible functional dependence (Figs. 12 and 13). In fact, the radiants fill an irregular area in the vicinity of the north ecliptic pole (Fig. 11). There is no concentration and no Gaussian distribution. Instead, the radiants cover the area more or less evenly. There is a clear dependence of the geocentric velocity, and also some orbital elements, on the position of the radiant.

To describe the dependencies quantitatively, we defined the mean radiant position, α₀ , δ₀ , and the deviation from the mean position, ∆α, ∆δ. With α₀ = 268°.4 and δ₀ = 62°.0, the actual geocentric radiant is: $${\alpha }_{\text{g}}={268}^{°}.4+\text{Δ}\alpha$$(17) $${\delta }_{\text{g}}=+{62}^{°}.0+\text{Δ}\delta ,$$(18)

from which ∆α, ∆δ can be computed for any observed meteor. The observed range was ∆α ∈ 〈−15°, +7°〉 and ∆δ ∈ 〈−5°, +5°〉. The observed range of solar longitudes was λ_⊙ ∈ 〈138°, 158°〉.

The expected orbital elements for radiants and solar longitude within these ranges are: $$v_{\text{g}}=22.32+\text{Δ}\alpha /10+\text{Δ}\delta /2.6$$(19) $$a=2.98\pm 0.12$$(20) $$q=1.009+5\times {10}^{-4}\text{Δ}\delta$$(21) $$i={35}^{°}.67+\text{Δ}\alpha /5.2+0.75\text{Δ}\delta$$(22) $$\omega ={183}^{°}9+0.64\text{Δ}\alpha -0.7\text{Δ}\delta$$(23) $$\text{Ω}={\lambda }_{\odot },$$(24)

where ∆α and ∆δ are given in degrees.

Four representative sets of radiants and orbits are given in Table A.3. The showers from the literature that can be identified with this branch are also listed. Not listed are ξ Draconids (Sekanina 1976, listed under 220/NDR in the IAU MDC) and ν Draconids (220/NDR, Jenniskens et al. 2016b). These proposed showers have similar radiants and orbits but their period of activity is at a later time (and the longitude of the ascending node is therefore larger) than is the case for the middle branch of August Draconids.

9.2.3 Upper branch

The number of observed meteors in the upper branch (Group 4) was lower than in other two branches and in κ Cygnids. Since no dependency of radiant position on solar longitude is apparent (Figs. 12, 13), we used the same approach as for the middle branch to describe the extend of orbital elements. The following equations were found: $${\alpha }_{\text{g}}={242}^{\circ }.9+\text{Δ}\alpha$$(25) $${\delta }_{\text{g}}=+{60}^{\circ }.3+\text{Δ}\delta$$(26) $$v_{\text{g}}=20.40+\text{Δ}\delta /3$$(27) $$a=3.06\pm 0.08$$(28) $$q=1.0012+7.5\times {10}^{-4}\text{Δ}\alpha$$(29) $$i={31}^{\circ }.3+0.75\text{Δ}\delta$$(30) $$\omega ={168}^{\circ }.0+0.55\text{Δ}\alpha$$(31) $$\text{Ω}={\lambda }_{\odot },$$(32)

with observed limits of ∆α ∈ 〈−6°, +6°〉, ∆δ ∈ 〈−5°, +2°〉, and λ_⊙ ∈ 〈148°, 160°〉.

Three representative sets of radiants and orbits are given in Table A.4. The only shower from the literature that could be identified with the upper branch is ι Draconids. They do not correspond exactly with the branch parameters but lie in the prolongation toward lower ω.

10 Physical classification

The basic physical classification of meteoroids can be done using the PE criterion of Ceplecha & McCrosky (1976) or using the Pressure-factor (Pf) recently defined by Borovička et al. (2022b). The PE criterion is based on the fireball terminal height for a given mass and velocity, while Pf considers the maximal dynamic pressure. Both approaches give the same picture in the present case.

Figure 17 shows Pf as a function of the meteoroid mass for all studied meteors. A part of the meteoroids is strong with Pf ~ 1. That value corresponds to material similar to stony meteorites. Most of these meteoroids are sporadic. Some of them are suspected shower members based on their radiants but not confirmed according to their orbital period (plotted with open symbols). Only three meteors are classified as shower members on the basis of the orbital elements: two as August Draconids and one as a κ Cygnid. All of them are small, with masses of ≲10 grams. Considering the large width of the streams, we cannot exclude that these meteoroids are also interlopers from the sporadic background. Alternatively, they could represent a minority strong fraction of the shower material. In any case, their exclusion from the analysis would not affect the results of the previous sections.

A large majority of shower meteors have Pf < 0.5. There is a clear trend of decreasing strength with increasing mass. This effect was previously observed in Taurids (Spurný et al. 2017). It has been found that large bodies are composed almost entirely of weak material, which contains stronger inclusions. Small meteoroids represent these inclusions (Borovička & Spurný 2020). It seems that κ Cygnids and August Draconids have a similarly hierarchical structure as the Taurids. Such a structure is probably usual for cometary material on Jupiter-family orbits. At the same time, it is evident that κ Cygnids are on average significantly more fragile than August Draconids. That fact confirms that they are different showers with different origins. κ Cygnids also contain a larger fraction of more massive meteoroids. Masses of up to 1 kg were observed. There is no difference in the physical properties of individual branches of August Draconids. A more detailed study of physical properties, including fragmentation modeling such as was done for Taurids (Borovička & Spurný 2020), is left for future work.

Fig. 17 Pressure factor, Pf , describing meteoroid mechanical strength as a function of mass for all studied meteors. Orbital groups as defined in Figs. 1 and 3 are marked by different symbols. The boundaries between Pf-classes as defined by Borovička et al. (2022b) are also shown.

11 Discussion

Meteor shower activity in August with radiants in the Cygnus- Draco region has been studied for almost 150 years. The radiant situation seemed to be very complex. More recent work suggested that only two showers are in action, κ Cygnids and August Draconids (Shiba 2017; Jenniskens 2023). Our work confirms this conclusion. Both showers nevertheless have a complex structure and cannot be described by a single set of radiant coordinates and orbital elements. We have provided equations to compute the plausible range of orbital elements and listed sets of typical elements within those ranges. The ranges cover elements that can currently be registered on Earth. We cannot exclude that different combinations of orbital elements, which do not lead to an intersection with Earth, exist within the streams.

11.1 Periodicity

One important topic discussed in the past is the 7-year periodicity of κ Cygnids and their presence in a 5:3 resonant swarm with Jupiter. We were able to confirm that hypothesis by direct measurements of meteoroid orbital periods. They really do cluster around the expected resonant value of 7.117 years. The measurements are still not precise enough to show that the periods sit exactly at this value but considering long-term periodicity it is almost certain. Since the period is larger than 7 years, the activity is not expected to repeat identically after 7 years. Nine cycles take 64.05 years, so the situation will repeat closely after 64 years. The evolution in between depends on how are the meteoroids distributed in mean anomaly.

Table 2 shows the expected years of activity. Our data show that there was high activity in 2021 but no noticeable activity in the neighboring years 2020 and 2022, when the stream was sampled at ±50° in mean anomaly from the maximum in 2021. The extent of the swarm is therefore lower than 100°. The next maximum is expected in 2028 and then every 7 years until 2049. However, the maximum will shift forward in time and is expected to be stronger in 2057 than in 2056. In fact, there are two possibilities. If the swarm is very narrow in mean anomaly, the maximum can fall in between 2049 and 2050 and then in between 2056 and 2057. No activity may be observed in any of these years. But if the swarm extent is larger, activity can occur in all these years at a similar level, lower than in 2021.

To decide between these two possibilities, we can look into the past. The years when a κ Cygnids outburst was reported in the literature (see Sect. 2) are marked in bold in Table 2. In fact, all well-documented outbursts occurred in years listed in Table 2. These years include 1985 and 1993, which are equivalent to 2049 and 2057, respectively, when the 64 year period is considered. Since the activity was detected in these transition years (8 years apart), the swarm cannot be very narrow.

In some years, no enhanced activity was reported, though it could be expected. In the more distant past, that fact can be ascribed to a lack of observations. Most suspicious is the year 2000. McBeath (2001) reported low visual activity of κ Cygnids in 2000 but noted that the expected peak was lost to moonlight (the full Moon occurred on August 15, 2000). It is interesting that Miskotte (2020) mentioned an outburst in 1999, but without any detail. McBeath (2000) reported that though κ Cygnids were observed throughout August 1999, the level of activity was unimpressive. Checking the archive photographs from the EN, we did not find any high κ Cygnid fireball activity in either 1999 or 2000.

Trying to find more data, we checked the archive² of the International Meteor Organization (IMO) Video Meteor Network (Molau & Barentsen 2014). The archive contains meteor shower assignments based on single station video observations. To obtain information about the activity level of κ Cygnids, the ratio of the number of observed κ Cygnids to the number of observed sporadic meteors was computed in four five-day intervals between August 6–25 in 1996–2010 (Fig. 18). The data from 1996 to 1998 are sparse, since the number of cameras was low. We must also keep in mind that meteors classified as κ Cygnids very probably also include August Draconids, at least the lower branch.

The high activity in 2007 is clearly visible in Fig. 18. Higher activity than normal was observed also in 1999, 2000, and 2006. The activity in 2000, and thus in all expected years from Table 2 since 1978, can therefore be confirmed. Moreover, the activity was also observed in the years 1999 (probably at a similar level as in 2000), 2006 (lower than in 2007), and, as Shiba (2017) reported, in 2013 (much lower than in 2014). This fact is not surprising and suggests that the width of the resonant swarm approaches ~90º in mean anomaly. We can therefore expect the activity to also occur in the years preceding the ones listed in the upper part of Table 2 and in the years following the ones listed in the lower part. The years in the middle row, such as 2021, should show high activity with no activity in neighboring years. In this view, it is somewhat surprising that no activity was reported in 1986 and in 1992. But we must realize that the high activity is never very high in terms of ZHR and represents just several meteors seen visually per hour. What is remarkable is the presence of fireballs and that the activity lasts for at least several nights.

The situation may be more complex owing to the fact that the κ Cygnid stream is extended in longitude of the node and in inclination. According to Shiba (2017), the 2013 activity occurred at low solar longitudes (in early August) and contained orbits with high inclinations. Our analysis shows that the inclination generally increased with solar longitude in 2021 (see Eq. (6)). Different parts of the swarm can therefore have different distributions in solar longitude and inclination. The 2006 activity also occurred in early August, while the 1999 activity seems to predominate at later times (Fig. 18). More data from different years are needed to study these details.

August Draconids do not show such a pronounced periodicity as κ Cygnids. Nevertheless, a minor variation from year to year cannot be excluded.

Fig. 18 Ratio of meteors classified as κ Cygnids to sporadic meteors as observed by the IMO Video Meteor Network in 1996–2010. The upper four panels show the data for five-day intervals. The lower panel shows the summary data for the whole period studied, August 6–26. The statistical uncertainties are indicated.

11.2 Parent body

Several authors tried to find the parent body of κ Cygnids. Jones et al. (2006) noted that the parent body should not only have a similar orbit but should also exhibit similar orbital evolution as the stream meteoroids when integrated backward. They found that their substreams, August Lyrids, γ Draconids, and ζ Dra- conids, which are all part of August Draconids (197/AUD) in our view, show sinusoidal variations in some orbital elements (namely eccentricity, perihelion distance, and inclination) with a period on the order of 2000 years. In contrast, κ Cygnids and α Lyrids, which are also part of κ Cygnids (12/KCG) in our view, showed flat behavior, in some cases with quick oscillations. Jones et al. (2006) proposed asteroids 2001 MG1 and 2004 LA12 as possible parent bodies of κ Cygnids.

Jenniskens & Vaubaillon (2008) found that the newly discovered asteroid 2008 ED69 shows similar sinusoidal variations in orbital evolution, as was described by Jones et al. (2006), and proposed 2008 ED69 as the parent body of κ Cygnids. However, that behavior corresponds to August Draconids. Indeed, Jenniskens (2023) lists 2008 ED69 as the probable parent body of August Draconids.

Moorhead et al. (2015) simulated the transport of meteoroids from several asteroids and again found 2001 MG1 to be the best candidate, though the match of simulated radiants and velocities with κ Cygnids was far from perfect. Jenniskens (2023) proposes that asteroid 2021 HK12 is the source of κ Cygnids without giving further details.

The orbital elements of the proposed parent bodies are given in Table 3. All of them have a semimajor axis significantly smaller than κ Cygnids. We searched the small body database³ for bodies that have a semimajor axis, a, perihelion distance, q, and inclination, i, similar to κ Cygnids. Four objects were found within the limits 3.5 < a < 3.9 AU, 0.9 < q < 1.1 AU, and 25 < i < 50º. They are listed in the lower part of Table 3. Three of them are asteroids; the fourth is comet 21P/Giacobini-Zinner, the parent comet of the October Draconid meteor shower. Little is known about the three asteroids. 2015 OA22 has a low albedo (1.3%), suggesting a primitive composition. The albedos of the remaining two asteroids are not known. In any case, the angular elements (ω, Ω) of all four objects are markedly different from κ Cygnids. It is nevertheless worth mentioning that except for 2001 XQ, the longitude of perihelia (π = Ω + ω) of the other three objects are not much different from κ Cygnids.

Physical properties of κ Cygnids are distinctly cometary. They belong to the weakest known meteoroids, which is also true for October Draconids. Comet 21P was rejected as a possible parent body of κ Cygnids by Jones et al. (2006) because of the difference in the longitude of the node. More recently, Neslušan & Tomko (2023) studied the past orbital evolution of 21P and simulated the associated meteoroid stream. They noted rapid orbital evolution. The comet’s orbit can only be considered to be reliably known during the last 1500 years. The ejected meteoroids are expected to be dispersed into the sporadic background within 1000 years. We can speculate that a part of the stream was trapped in the 5:3 resonance several thousands years ago, avoided further dispersion, and forms the κ Cygnids today, but it will be difficult to prove such a scenario. More work is also needed evaluate the possible relation of asteroids from Table 3, in particular 2010 HY22 and 2015 OA22, to κ Cygnids.

12 Conclusions

We have proposed a clarification of the situation with meteor showers that are active in August and that have radiants in the Cygnus-Draco area. There are two separate showers referred to as κ Cygnids (12/KCG) and August Draconids (197/AUD). Both showers are extended systems active (at a low level) for most of the month and have large radiant areas and a wide range of entry velocities. Their orbits cannot be described by a single set of orbital elements.

The more spectacular are κ Cygnids, which are rich in bright, often bursting, fireballs produced by extremely fragile cometary meteoroids with masses of up to at least one kilogram. Noticeable activity, however, does not occur every year. Most κ Cygnids activity is produced by an isolated swarm of meteoroids locked in a 5:3 main-motion resonance with Jupiter. The orbital period is 7.12 years and we estimate the extent of the swarm to be at most 90º in mean anomaly (±45° from the center). The shower can therefore be observed in a single year or, at most, during two consecutive years (but at a lower level) within each 7-year period. Only occasional κ Cygnids can appear in other years.

The activity mostly occurs at solar longitudes of 130–152°. The radiant extends over almost 30° in declination. Some authors have considered more showers to be active within that area but we have shown that it is a single shower. The extent in declination reflects the extent in the inclinations of the orbits, which is 28–44°. A general increase in mean inclination with solar longitude was observed in 2021, but meteoroids with a wide range of inclinations (±8°) could be encountered at any given time. The inclination is correlated not only with the declination but also with the geocentric velocity, perihelion distance, and argument of perihelion. We defined a variable p, reaching values between −1 and +1, to describe the correlated deviations of these parameters from their mean values. The reason for the correlation is probably the condition that the orbit must intersect Earth’s orbit.

The August Draconids are even more complex. The whole stream can be divided into three branches. The middle branch has perihelia and aphelia near the nodes. The perihelia lie close to the Earth’s orbit and the radiants are close to the northern pole of the ecliptic. The lower branch has perihelia south of the ecliptic plane and radiants to the south of the ecliptic pole. The upper branch has perihelia north of the ecliptic plane and radiants to the west of the ecliptic pole. The activity period of the lower branch is similar to that of κ Cygnids. Because the radiants are not far apart, these two showers have sometimes been confused. Nevertheless, August Draconids have lower velocities and shorter semimajor axes. The activity of the middle branch is shifted to later times and the upper branch is active even later. Nevertheless, the activity periods of the branches partly overlap. It is possible that the system is wider and the activity starts as soon as July and ends in September. It is a matter of future work to check if other proposed showers such as κ Lyrids (464/KLY), which are active in July, or ν Draconids (220/NDR), which are active in September, are part of the August Draconid complex.

August Draconids seem to be active every year, though some variations in intensity cannot be excluded. The orbits are close to the 7:3, 9:4, and 11:5 resonances with Jupiter with periods between 5 and 5.5 years. It is not clear, however, how many meteoroids are indeed in the resonances. The meteoroids are more compact than κ Cygnids though still cometary. The stream also does not contain so many large (>50 g) meteoroids as κ Cygnids. The candidate parent body is asteroid 2008 ED69, though the match is not assured in our opinion. The parent body of κ Cyg- nids is even more elusive. Taking into account its chaotic orbit, we do not exclude comet 21P/Giacobini-Zinner as the parent body, though the current orbit is different.

In summary, the precise data from the EN enabled us to describe the orbital structure of κ Cygnids and August Dra- conids. The directions of future research can be detailed modeling of atmospheric fragmentation to reveal the material properties, on the one hand, and orbital integration to search for parent bodies, on the other hand.

Data availability

The trajectories, radiants, orbits, magnitudes, and physical classifications of all 179 fireballs used for the present analysis are available at the Centre de Données astronomiques de Strasbourg (CDS) via anonymous ftp to cdsarc.cds.unistra.fr (130.79.128.5) or via https://cdsarc.cds.unistra.fr/viz-bin/cat/J/A+A/695/A83. The format is almost identical to our previous catalog of 824 fireballs (Borovička et al. 2022a). Eighteen fireballs are present in both catalogs. Instead of shower assignment, we give here the group number as in Fig. 1, G1–G5 for the core members and G1A–G5A for possible members (deviating in orbital period) of the groups. G1 is identical with κ Cygnids, G2–G4 are three branches of August Draconids. Sporadic fireballs with the radiants in the region of interest are included as well.

Acknowledgements

We thank all technicians and operators of the EN for their work. Student Anna Marie Vančurová helped us with the measurement of some fireballs and the initial analysis. We acknowledge the use of the IAU Meteor Data Center Shower Database. We are grateful to an anonymous referee for carefully checking the manuscript and providing detailed comments. This study was supported by grant no. 24-10143S from Czech Science Foundation. The operation of Slovak DAFOs was supported by Slovak Grant Agency for Science VEGA (grant no. 2/0059/22).

Appendix A Additional tables

References

All Tables

All Figures

	Fig. 1 Positions of geocentric radiants of the analyzed meteors in Suncentered ecliptical coordinates in polar projection. The shaded area is the investigated region of interest. The radiants of meteors that were assigned to one of five groups are plotted with colored symbols. Other radiants are plotted as crosses.
In the text

	Fig. 2 Meteor maximum absolute magnitude plotted against the distance of the meteor from the nearest observing station.
In the text

Fig. 3 Orbital periods as a function of solar longitude. Formal one- sigma errors of periods are indicated. Only periods within 3–10 years are shown. Seven meteors had periods outside of this range; the full range was 1.3–15 years. Solar longitudes correspond to the month of August. Long-dash colored lines show the limits of periods for the cores of individual groups. Short-dash black lines mark the 5:3 and 7:3 resonances with Jupiter.

In the text

	Fig. 4 Relation between perihelion distance and inclination. Groups are coded as in Fig. 1 and marked by colored ellipses.
In the text

	Fig. 5 Relation between argument of perihelion and inclination. Groups are coded as in Fig. 1.
In the text

Fig. 6 Projections of typical meteoroid orbits from Groups 1 and 2 (left) and 3 and 4 (right) into the ecliptic plane. The parts of orbits above ecliptic are shown by thick line. The orbits of planets Mercury to Jupiter are drawn in black. The Earth’s orbit is thicker. Vernal equinox is to the right. All planets and meteoroids move counterclockwise. Note the different positions of nodes relative to aphelion in Groups 2–4.

In the text

	Fig. 7 Three representative orbits of Group 1 meteoroids in a threedimensional plot with Earth’s orbit. The positions of perihelion points are depicted as circles. The parts of orbits below the ecliptic are shown by a thin line.
In the text

	Fig. 8 Histogram of orbital periods of meteoroids of Groups 1 to 4, including the outliers. Positions of selected mean motion resonances with Jupiter are also indicated.
In the text

	Fig. 9 Histogram of year of appearances of meteors belonging to the cores of Groups 1 to 4. The four core members of Group 5 are not shown. They appeared in four different years (2020–2022, 2024). The solid blue line denotes the total number of meteors observed in August each year. The corresponding scale is on the right.
In the text

	Fig. 10 Relation between perihelion distance and inclination for the 2014 κ Cygnid data from the EN, 2007 data from the Spanish Meteor Network, and 1950 data from Harvard cameras. Groups are marked as in Fig. 4. One 2014 fireball with q = 0.916 AU is not plotted.
In the text

Fig. 11 Geocentric radiants of meteors belonging to the cores of Groups 1–5 (color squares) plotted on the background sky map. Geocentric velocity is color-coded. The approximate limits of the groups are plotted as black ellipses or other shapes. Also plotted are radiants of various showers from the literature (color circles) with velocities (if known) coded with the same colors. The showers are labeled according to the names used by the authors and the paper numbers as given in Tables A.1–A.4 (number 19, not listed in the Tables, belongs to Jenniskens et al. 2016b). Only paper numbers are given for κ Cygnids. The other shower names are as follows: tCy (θ Cygnids) oDr (o Draconids), zDr (ζ Draconids), gDr (γ Draconids), xDr (ξ Draconids), ADr or AUD (August Draconids), AMD (August µ Draconids), IOD (ι Draconids), NDR (ν Draconids), aLy (α Lyrids), ALy (August Lyrids), MUL (µ Lyrids), KLY (κ Lyrids), and GB to GG (Groups B to G of Cygnid-Draconid Complex).

In the text

	Fig. 12 Geocentric right ascension of the radiant as a function of solar longitude. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1 and 3 (solid lines). Mean values are plotted for Groups 2 and 4 (dashed lines).
In the text

	Fig. 13 Geocentric declination of the radiant as a function of solar longitude. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1 and 3 (solid lines). For Group 1, limits of ±12° in declination from the fit are also drawn (dotted lines). Mean values are plotted for Groups 2 and 4 (dashed lines).
In the text

	Fig. 14 Correlation between geocentric declination and velocity. Groups are coded as in Fig. 1. Linear fits are drawn for the core members of Groups 1–4 (solid lines).
In the text

	Fig. 15 Histogram of values of p variable (computed from Eq. (2)) for 39 members of Group 1 (κ Cygnids).
In the text

	Fig. 16 Relation between argument of perihelion and perihelion distance for three branches of August Draconids.
In the text

	Fig. 17 Pressure factor, Pf , describing meteoroid mechanical strength as a function of mass for all studied meteors. Orbital groups as defined in Figs. 1 and 3 are marked by different symbols. The boundaries between Pf-classes as defined by Borovička et al. (2022b) are also shown.
In the text

	Fig. 18 Ratio of meteors classified as κ Cygnids to sporadic meteors as observed by the IMO Video Meteor Network in 1996–2010. The upper four panels show the data for five-day intervals. The lower panel shows the summary data for the whole period studied, August 6–26. The statistical uncertainties are indicated.
In the text