Taurid Stream #628: A Reservoir of Large Cometary Impactors

In 2015, the Australian DFN covered 1.5 million km² of sky viewing area, established around 30°S latitude (Howie et al. 2017a). Each DFN observatory comprises a high-resolution still imaging system: a 36 MP digital camera (Nikon D800, D800E, or D810), associated with a Samyang 8 mm f/3.5 fish-eye lens, taking 25 s exposures at 6400 ISO. In 2015 all observatories operated with these settings. The field of view of the cameras is all-sky, except for a crop of 10° on the horizons of long sides of the sensor (usually North and South). The pixel size is 119 s of arc in the center, decreasing toward the edges (87 s of arc at 5° elevation). The cameras are sensitive to stellar magnitude 0.5 for meteors (7.5 for stars) and reliably detect meteors that are brighter than apparent magnitude −1.5 for >0.9 s.

Meteor events are automatically detected in the images by the software procedures described by Towner et al. (2020). Astrometric measurements are performed in the same manner as described by Devillepoix et al. (2018), resulting in measurements precise down to 1–2 minutes of arc. The triangulation of meteor trajectories is performed using a weighted straight line least-squares approach, similar to the one described in Borovička (1990). In order to get an appropriate entry velocity for the meteoroid, an extended Kalman smoother is applied to the positional data, throughout the visible bright flight (Sansom et al. 2015). This method also yields statistical uncertainties that encompass both model errors and measurement errors. These results are crucial for initializing orbit determination, as orbital parameters, like the semimajor axis and eccentricity, are very sensitive to the errors in initial velocity. The heliocentric orbits of the meteoroids are determined using a backward integration from the start of the visible bright flight. The meteoroid is back-tracked through the upper layers of the atmosphere and out of the sphere of influence of Earth to a distance of one Hill sphere (Jansen-Sturgeon et al. 2019). Uncertainties on the orbital parameters are computed using a Monte Carlo method based on the uncertainties of the first velocity vector observed.

The DFN data reduction pipeline uses aperture photometry on the fireball track to calculate brightness. Doing photometry on the reference stars used for astrometry yields the instrumental zero-point of each camera, accounting for extinction and vignetting. The fireball brightness is converted into magnitudes by accounting for the different exposure times: the effective exposure time for stars is typically 11.2 s (25 s exposure modulated by the liquid crystal shutter), and 0.06 s or 0.02 s for a fireball shutter break (see Howie et al. 2017b for details on the action of the liquid crystal shutter). Apparent magnitude is converted to absolute (for constant distance of 100 km) after triangulation, using the observation range. The main limitation on this technique is the saturation of the sensor, which typically happens when the fireball exceeds apparent magnitude −6. Blooming of the trail enables brightness measurements out to about −10 mag.

The main use of photometric measurement in the present study is to calculate meteoroid strength and to get a zero-order mass estimate. As detailed by Brown et al. (2016), the peak brightness instant of a fireball is a good indicator of catastrophic fragmentation, and therefore a reasonable proxy for calculating a bulk tensile strength for the entering body. This method is more robust to instrumental bias than the PE criterion introduced by Ceplecha & McCrosky (1976) and has the advantage of being inferred directly from observable parameters (no mass calculation involved). We therefore use the following relation from Bronshten (1981) to calculate tensile strength S: S = ρ_atm v², where ρ_atm is the density of the atmosphere estimated using the NRLMSISE-00 atmospheric model (Picone et al. 2002). Parameter v is the velocity at that instant calculated by the Kalman smoother described by Sansom et al. (2015). The main limitation of the method comes from the uncertainty on the instant of peak brightness, dominated by the sampling rate (10 Hz), which translates into 2 km of altitude for the average Taurid, or a ≃1.3 factor in strength.

Thanks to the continental scale of the network, operational and weather biases are mitigated by the large collecting area and observation time. However, a consequence of this is that precisely determining the surveying area probed by the instrument is difficult. Calculating probing area as a function of time may be done accurately and relatively easily when a small number of narrow angle optics are used, such as described in Blaauw et al. (2016). But even at a basic level, this kind of work with all-sky cameras spaced on a continent-scale network is more tricky, and debiasing the DFN data set to get precise fluxes will be the subject of a future paper.

The DFN observatories, combined with the data reduction methods described above, have led already to multiple meteorite recoveries: Creston (Jenniskens et al. 2019), Murrili (Sansom et al. 2020), and Dingle Dell (Devillepoix et al. 2018), as well as three more recent (not yet published) recoveries. These successes in precisely pinpointing the location of meteorites are good indications that the data reduction process is free of major systematic issues.

The main goal of CAMS is to map the presence of meteor showers of +4 to −5 mag meteors throughout the year. In 2015 November, CAMS networks had been established in California, Arizona, Florida, the BeNeLux, and New Zealand. Most CAMS networks are on the northern hemisphere, but they experienced a relatively small number of cloudy days that year. CAMS methods are described in detail in Jenniskens et al. (2011). In brief, CAMS utilizes a network of analog low-light video cameras, mostly Watec Wat902H2 Ultimate cameras with 30° × 20° field of view each and +5.4 stellar limiting magnitude. Customized software detects the meteors, calibrates the background star field to obtain astrometric positions, and then combines such data from two or more stations to triangulate the meteor trajectory. CAMS yields more than 100,000 meteoroid orbits per year and has proven to be a very efficient tool for studying meteor showers and linking them to possible parent objects (Jenniskens et al. 2016a, 2016b, 2016c).

The high detection rate of meteors not assigned to showers by Jenniskens et al. (2016c), with geocentric entry speed <35 km s⁻¹ from the antihelion source at the same time as the Taurid showers, provides a baseline of sporadic meteor shower activity that can be used to calculate the effective observing time due to weather. The CAMS flux data were debiasing by assuming a constant sporadic flux during the s-Taurid activity period. The main CAMS networks are situated at a latitude of ϕ = +37°, where the Southern Taurid radiant is up almost all night in early November and the mean altitude of the Southern Taurid radiant is hR ≃ 40°. This results in a correction factor of $1/\sin ({hR})=1.35$ to get the equivalent zenithal hourly rate. For scaling the distribution we use the flux density of Grün et al. (1985), f_G = 6.85 × 10⁻⁸ > 1 g meteoroids m⁻² yr⁻¹, and the correction factor between interplanetary and top of the atmosphere S = 0.67 of Moorhead et al. (2019). From this, we can calculate the influx of s-Taurid on Earth as

$$\begin{eqnarray*}&&\begin{array}{l}N(\#628\gt 1\,{\rm{g}}\,{\rm{impacts}}\,{\rm{on}}\,{\rm{Earth}})=\displaystyle \frac{n}{m}\ast \,\displaystyle \frac{1}{\sin ({hR})}\\ \quad \ast \,{f}_{G}\ast S\ast 2\ast {\rm{\Delta }}{\lambda }_{\odot }\ast {{\rm{X-section}}}_{{\rm{Earth}}}.\end{array}\end{eqnarray*}$$

In this formulation Δλ_⊙ is the exposure time to the shower as observed by CAMS, and the factor of 2 is there to compensate that a surface at the top of the atmosphere is effectively twice the collecting area of the model randomly tumbling plate of Moorhead et al. (2019). In relation to a meteoroid stream, Earth's exposure to the stream can be effectively represented as a cross-section area defined by its radius.

The reported flux values are limited to a 1 g threshold mass using the observed magnitude distribution. The count of all sporadic meteors with the 25–30 km s⁻¹ entry speed of Taurids was assumed to be exponential in shape of this magnitude interval, from which a detection probability function was derived by fitting an exponential slope to the bright end of the magnitude distribution and then dividing observed counts by the fit-predicted count. The fraction of completeness for magnitudes −1 and up was P(m) = 1.00, 0.80 ± 0.02, 0.37 ± 0.01, 0.093 ± 0.003, 0.014 ± 0.002, 0.001 1 ± 0.0002, and ∼5e−6. This probability function was then applied to the detected count of shower meteors to derive the magnitude size distribution of different Taurid Complex component showers.

Zoom In Zoom Out Reset image size

The s-Taurid shower stands out well from other Southern Taurid complex meteors by their geocentric speed. Figure 2 plots the geocentric speed and time (solar longitude) of all meteors associated with the Southern Taurid complex in both the CAMS (top) and DFN (bottom) data sets. Vertical white bands are due to cloudy weather with less-than-complete coverage. The data are split into two groups: the outburst years of 2012 and 2015 (Table 2), and the no-outburst years of 2010, 2011, 2013, 2014, and 2016. The 2012 encounter with the s-Taurids is Δ_M = 35° from the center of the resonance according to the model of Table 1. Hence, the weak detection of the s-Taurids in 2012 by CAMS implies an extension of this component until at least mean anomaly 35°, in agreement.

Zoom In Zoom Out Reset image size

The outburst years show a component that produces a narrow range of geocentric entry speed at any given solar longitude, with a strong change in the speed as a function of time. This component is only weakly present in nonoutburst years (Figure 2). This component was earlier identified as shower #628, the s-Taurids (IAU code STS). The period of activity for this component is λ_⊙ ∈ [213, 234]°.

The presence of this STS component is also evident in the 2015 DFN data (Figure 2), despite a lower number of orbits, as the STS stream largely dominates the Southern Taurid activity at fireball sizes. The change in velocity with solar longitude translates into a strong increase of perihelion distance with increasing solar longitude and a decreasing eccentricity. The semimajor axis and inclination of the orbits remain nearly constant, as does the longitude of perihelion.

Figure 3 shows the debiased STS rates for CAMS, along with that of the remaining STA and NTA streams. The rates are normalized to that of all sporadic meteors with speeds <35 km s⁻¹. This ensures that the total sporadic count reflects the observing conditions during that part of the night when the antihelion source is best observed. The sporadic apex and toroidal sources have been removed from the count. The 2015 STS count was compared to the sporadic meteor rate in 2015 only. The multiyear debiased distribution produced better-defined shower activity profiles than early results in Jenniskens et al. (2016b). The shower components identified in Jenniskens et al. (2016c) are still present. The STA and NTA shower profiles are different, an indication that the nodal line of individual meteoroid orbits did not fully rotate, as earlier pointed out.

Zoom In Zoom Out Reset image size

Figure 4 shows the distribution of peak magnitudes in 0.5 mag intervals for CAMS-detected NTA, STA, and STS meteors. The STS population is significantly more skewed toward brighter meteors. The cumulative mass–frequency distribution as observed by CAMS (Figure 8) can be expressed as N(>m) = am^−b with a = 1.42 × 10⁵ and b = 0.94 when expressed in grams. This is representative of the total influx of s-Taurids during the 2015 encounter.

Zoom In Zoom Out Reset image size

The annual Southern and Northern Taurids have differential magnitude indices of χ ≃ 3.0 (Figure 3); this is close to the typical value for JFC material (χ = 3.29 ± 0.09, as determined by Jenniskens et al. 2016c). On the other hand, s-Taurids have a much shallower distribution with χ = ∼2.0 (assuming a sporadic χ = 3.4). This gives a differential mass index for the observable stream of s = 1.75 ($s=1+2.5\mathrm{log}(\chi )$).

The observed size distribution confirms that the s-Taurids are relatively rich in large meteoroids. Indeed, a study by Soja et al. (2011) on radar meteors observed by the Canadian Meteor Orbit Radar in 2005 (typical observed mass of 10⁻⁷ kg, which roughly corresponds to optical magnitude +7) failed to identify the 7:2 resonance from regular Southern Taurids. They discuss that this is partly due to the poor constraints the radar observations put on the velocities (and therefore the semimajor axes), so it is not possible to distinguish STSs from STAs dynamically. Therefore, unless the STS outburst is strong enough to significantly skew the overall Southern Taurids rates, it is not detectable. Soja et al. (2011) do not provide an upper constraint on the STS/STA activity, but even without hard numbers this analysis confirms the trend shown in Figure 4: the STA branch dominates the STS branch at the low-mass end (${{Mv}}_{\max }\gt -4$).

Beyond CAMS data, toward fireball sizes, flux density data for this stream become more scarce. The DFN and EN fireball networks do not yet have time–area debiasing information to calculate flux densities. The masses reported by Spurný et al. (2017) give an idea of the slope of the distribution at magnitudes below −9, where the EN sample appears to be complete (Figure 8). Although it is apparently shallower than the slope calculated by CAMS, the numbers are so small that the extrapolated flux agrees within 2 error bars (2σ). Even at these bright magnitudes there may still be an effect of the brightness-dependent variation in effective covering/reporting: the brighter the bolide, the further it can get detected and studied.

Figure 5 shows the altitude of peak brightness as a function of peak meteor magnitude. The result shows that larger meteoroids penetrate deeper into Earth's atmosphere before reaching peak brightness, as expected. Among CAMS-detected visual meteors, that is a fairly continuous trend. However, at the larger fireball sizes, meteoroid penetration enters a strength-dominated regime. The transition into the strength-dominated regime occurs at about −7 mag. In the strength-dominated regime, the altitude of peak brightness is independent of mass, although there is a weak trend that the largest meteoroids are weaker than the smaller meteoroids near this transition. All s-Taurids experience their peak brightness above 60 km altitude (Figure 5) and do not survive below 50 km altitude. The deepest penetrating fireball had a minimum height of 54.5 km, while the average end height of the DFN fireballs was 67.2 km (see Table A1). This is consistent with the results of Spurný et al. (2017).

Zoom In Zoom Out Reset image size

Figure 6 is an ecliptic orbit plot of all Southern Taurids observed by the DFN in 2015. The figure shows that most observed Taurid fireballs clustered in a tight stream with constant longitude of perihelion (all ellipses pointed in the same direction). These s-Taurids are shown in blue. This stream is highly stratified: they form a series of more or less concentric ellipses for fireballs detected at different solar longitude along Earth's orbit. Unlike most meteoroid streams, this stream appears to be narrower at aphelion than at perihelion. The dispersion of perihelion distances (q) of the blue orbits is 8.9% (one standard deviation), while the aphelion distance (Q) is dispersed by only 2.4%, and the semimajor axis (a) by 2.1%. a and Q are tightly correlated; a and q are not. The stream approaches the orbit of Jupiter near aphelion, suggesting that the secular perturbations responsible for the observed dispersion are strongest near aphelion, not near perihelion.

Zoom In Zoom Out Reset image size

Figure 7 plots the semimajor axis of the 2015 Taurid fireballs measured in the DFN network as a function of solar longitude (time in the year). The DFN-derived orbits during solar longitudes 217.5 and 227.5 show a clear concentration of semimajor axis values around the mean semimajor axis a = 2.234 ± 0.007 au with a standard dispersion of 0.041 au. The mean calculated error in the semimajor axis values is 0.034 ± 0.014 au, in good agreement with the observed dispersion if all these orbits have exactly the same semimajor axis of a = 2.2563 au corresponding to the 7:2 MMR with Jupiter (dashed line). This result confirms earlier conclusions from EN fireball observations reported by Spurný et al. (2017) that the meteoroids appear to be trapped in this resonance, and it demonstrates the accuracy of the semimajor axis calculations. Most of the observed dispersion in semimajor axis is due to measurement error. However, there is a small systematic error of −0.022 ± 0.007 au.

Zoom In Zoom Out Reset image size

The fireballs shown in pink are outside the main dispersion. Most have slightly smaller semimajor axis, while three orbits just reach the orbit of Jupiter. A total of 52 of the observed trajectories have a semimajor axis within 2σ from the resonant value; 21 do not. Surprisingly, the latter scatter around the time of the outburst over 2165–2317 solar longitude (CAMS has 212°–235° range; see below), while the ones that are within 2σ from the resonant value scatter over the whole observing interval of 207.7–255.9 solar longitude. Hence, whether or not the trajectories scatter around the resonant value is not a mark of the s-Taurids; it is a mark of the Southern Taurids as a whole.

Some of the pink orbits may represent measurement errors in the initial velocity determination, or they are a more perturbed population of meteoroids that possibly already experienced some changes in semimajor axis due to encounters with the terrestrial planets. If so, they are likely part of the background Southern Taurids.

Spurný et al. (2017) first pointed out that the stream contains meter-sized asteroids. Here, we will attempt to quantify this. The US government (USG) satellite sensors detect meter-scale impactors in Earth's atmosphere over the entire planet as a collecting area, or if a meteoroid stream is involved approaching from a given direction, that collection area is the projected Earth surface to that stream. These data are reported online on the NASA JPL fireball website. ⁴ Even with this large of a collection area, the chance of detecting impacts is small. Figure 8 shows the cumulative mass–frequency distribution from CAMS and DFN data. The USG-detected bolides are about −18 mag and brighter (Tagliaferri 1994). That leaves a large gap without debiased data.

Zoom In Zoom Out Reset image size

Extrapolating the mass–frequency distribution established in Section 3.2, we predict 0.37 meteoroids > 940 kg impacting Earth during the 2015 s-Taurid episode. Indeed, the USG satellites detected no unusual number of fireballs in late October and early 2015 November. One bolide was detected on 2015 October 31, one on 2015 November 2, and one on 2015 November 13. No velocity components are reported, so we do not know the radiant or speed of these bolides.

However, there are data from multiple years of observation now. If any Taurids are among these impactors, we expect their penetration depth to be relatively shallow based on the size dependence shown in Figure 5. The observed centimeter-to-decimeter-sized meteoroids by CAMS, DFN, and EN show no significant decrease in peak height as sizes get larger.

The USG data only include the altitude of peak brightness. We started by filtering the USG data set by height of maximum brightness >60 km as a first pass to identify weak cometary impacts, as STS observed by the DFN breakup >66 km (Figure 5). We note that the stated peak brightness altitude from the sensors is generally reliable to about ±4 km, as shown by Brown et al. (2016), and that these altitudes are reported for most events from the beginning of 2005 onward. Detections are made at night, but also in daytime. As mentioned by Devillepoix et al. (2019), the typical energy report limit is 0.1 kT TNT; therefore, we exclude event 2011 August 4 07:25:57 (0.098 kT reported yield) from our analysis for detection significance issues. We are left with 10 significant events that fit the height criterion (Table 3). However, there are no velocity vectors reported for these events, so there is no direct dynamic link between any of these and the Taurid complex.

Nevertheless, 3 out of these 10 very weak meteoroids fall in the STS activity period, and even more remarkable, they happen in 2005 and 2015, two years during which strong STS activity has been reported, and are predicted by the model of Asher & Izumi (1998) (Table 1). In 2005, two events occurred in short succession. All three events suggest that the largest fragments in this stream are at solar longitude [213, 234]°. Given this, combined with data in Figure 3, these USG bolides occurred during the STS activity period. We confirm that these three events happened while the Southern Taurid radiant was above the local horizon.

Can these bolides signify a detection of the s-Taurids? Considering the very low number of events observed, we need to build a statistical test to assess the significance of this apparent rate increase during a swarm episode. Let us test the hypothesis H₁, "An airburst from weak material (main explosion >60 km) is more likely to happen during an STS activity period," against the null hypothesis H₀, "No increase in the rate of impacts from weak bodies can be observed during an STS activity episode." We define an STS swarm episode as a period that happens on a year predicted by the model of Asher & Izumi (1998), within the interval where the USG sensors have consistently observed airburst heights (2005, 2008, 2012, 2015), and within the activity period observed by CAMS (solar longitude ∈ [213, 234]°). We use the rateratio.test R package, ⁵ which implements the methods described in Fay (2010) to carry out the statistical test. At 95% confidence, the background weak meter-scale impact rate is [0.001, 0.005], compared to $[0.007,0.1]\,{\mathrm{Earth}}^{-1}{\lambda }_{\odot }^{-1}$ when λ_⊙ ∈ [213, 234]°, which corresponds to a weak impactor influx increase of [2.5, 55]× (see Table 4 for full test test data and results). The small p-value = 0.004 shows strong evidence against the null hypothesis (at 95% confidence).

Table 4. Poisson Test on the Significance of the Impact Rate Increase of Meter-sized Weak Material (Peak Brightness >60 km) Hitting Earth Observed by the USGS during STS Outburst Episodes (λ_⊙ ∈ [213, 234]°)

Population	Surveyed Years	Observed Events	λ⊙ Integrated (deg)	Rate (${\mathrm{Earth}}^{-1}{\lambda }_{\odot }^{-1}$)
Weak impactor population	[2005–2016]	10	3960	[0.001–0.005]
Probable STS	2005, 2008, 2012, 2015	3	84	[0.007–0.1]

Note. Ranges given are at 2σ confidence. The influx increase factor during an STS outburst is [2.1, 46]×.

Download table as:  ASCIITypeset image

Although we cannot definitely link any individual events with the Taurids, the apparent rate increase in meter-scale weak impactors during the STS outburst activity is statistically significant, and we can say that during an STS outburst episode Earth is more likely to get impacted by a meter-scale STS than a sporadic meteoroid of the same size. Over an s-Taurid episode, the number of >0.1 kT TNT of s-Taurid impactor is [0.15, 2.1] (Figure 8).

Adding this detection rate to the overall picture shows that the meteoroid size distribution does not change down to meter-size scale (Figure 8). This result implies that the #628 STS stream contains some of the largest meteoroids known to any cometary meteor shower.

Finally, it is possible that the USG satellite detections do not represent all large bolides in Earth's atmosphere. The European Network detected one superbolide during the 2015 outburst called EN311015_180520 (Spurný et al. 2017). The observation of this 1300 kg bolide over a superbolide coverage area of roughly 0.1% Earth is statistically unlikely but not impossible (Figure 8). The EN311015_180520 superbolide (0.17 kT TNT total kinetic energy) should have been well within the detection range of the USG sensors, but this bolide was not reported. A single mismatch is not indicative of a particular issue, as the USG sensors are known to have had as low as 70%–80% Earth coverage in the past (Brown et al. 2002). The USG sensors' s-Taurid flux could have been underestimated by a factor of 2 considering that both 2008 and 2012 were taken into account for the USG clear-sky calculations (Section 3.5), as these two years were quite distant from the resonance center. It could mean that the meter-scale s-Taurid is a factor of 2 larger than what we calculated here. That would still be consistent with a constant mass distribution index from centimeter to meter scale (Figure 8).

The low magnitude distribution index (χ = 2.0) and small differential mass index (s = 1.75) are atypical of most JFC showers. Such values point to a collisionally relaxed population, with equal combined cross-section area in each magnitude bin (s = 1.67), or a collisional cascade where each meteoroid is broken by a mass just big enough to do so (s = 1.83). This points to a very gentle collision process such as would be experienced during a cometary breakup of relatively strong material with intrinsic low-s size distribution. High-velocity collisions in the interstellar medium or other aging processes like grain charging and thermal cycling did not significantly affect the size distribution index of this population of meteoroids. That means that the stream is relatively young and was born from relatively strong cometary material. Weak cometary material breaks into very steep distributions toward small particles (Jenniskens 2006).

The unusually large number of shower fireballs detected by DFN and EN points to the presence of large meteoroids in this stream. Continuation of the particle size distribution to larger sizes implies a population of meter-sized objects in this stream that appears to have been detected in USG satellite observations. The presence of such large bodies in the stream is consistent with the material being relatively strong for cometary material in general. This stream is currently a major contributor to the overall population of large weak meteoroids. If the USG statistics are representative of larger cometary impact hazards, the s-Taurids dominate the flux when they are active, increasing the impact risk by [2.5, 55] × (2σ), and overall they represent a significant fraction of all large cometary impactors, even though their activity period is only a few weeks every couple of years!

Our measured DFN fireball mean semimajor axis of 2.234 ± 0.007 au is 0.019 au higher than the semimajor axis of 2P/Encke but 0.022 au lower than the a = 2.2563 au corresponding to the 7:2 MMR with Jupiter. It is also 0.031 au lower than the current semimajor axis of 2015 TX24. Natural oscillations of semimajor axis are about 0.03 au in this part of the asteroid belt (Nesvorný & Morbidelli 1998), so the motion of all objects will be affected by the MMR. If the breakup happened in the resonance with small relative ejection velocities, it is likely that the resonance prevented the dispersion of the dust by avoiding Jupiter's presence at aphelion when the dust was there.

Spurný et al. (2017) argued that the s-Taurids are the product of fragmentation event (of 2004 TG10?) at about 3.6 au distance from the Sun, where the fireball orbits have lowest dispersion, with relatively high ejection speeds of 1.5 km s⁻¹. The observed asteroids 2005 TF50, 2015 TX24, and 2005 UR (and 2004 TG10) were created in that event and now have evolved along the rotation of the nodal line, now located 2000, 2300, and 2400 yr in rotation behind 2004 TG10. Material not trapped in the 7:2 MMR has since been lost.

That high ejection speed of 1.5 km s⁻¹ is contrary to the gentle breakup conditions implied by the particle size distribution we measured in this paper. It would also suggest that much of the material was ejected into orbits outside of the MMR and material would still be dispersing rather quickly along mean anomaly over time. Instead, we see that dust is confined in a narrow range of mean anomaly. The best handle on age comes from the work by Whipple & El-Din Hamid (1952) and Olech (2016), who found that individual fireball orbits originated from a common orbit about 1500 yr ago. However, if the MMR is involved, this age may be an upper limit only.

On closer inspection, we find that the change of argument of perihelion with node does not align all three bodies (Figure 9). As noticed before by Spurný et al. (2017), the longitude of perihelion is not quite constant in the s-Taurids stream. Instead, the trend of argument of perihelion points directly at asteroid 2015 TX24, which is only 35 further in node than the range of DFN fireballs observed. A few EN fireballs cover the node of 2015 TX24. 2015 TX24 is also close in position to the nodal dependence seen with perihelion distance, eccentricity, and inclination, but those parameters can scatter more easily and that difference may merely reflect the change in orbital elements needed to intersect Earth's orbit.

Zoom In Zoom Out Reset image size

Asteroids 2005 UR and 2005 TF50 have similar longitude of perihelion to 2015 TX24 but do not fall along the trend line seen among the meteoroids. We conclude that these larger asteroids are not simply fragments of this stream as proposed by Spurný et al. (2017).

A likely scenario is that we are seeing recent activity from a breakup involving surviving asteroid 2015 TX24. 2015 TX24 now has a semimajor axis of 2.2647 au (epoch 2020 December 17.0 TDB). That puts it within the normal semimajor axis oscillation of ±0.03 au from the resonance. Solar radiation pressure will slightly increase the semimajor axis of the DFN meteoroids, but if the ejection conditions were slightly lowering the semimajor axis, this could have been compensated. It is possible that the change of longitude of perihelion away from the node of 2015 TX24 is due to the influence of the 7:2 MMR. Presumably, the further the meteoroid node now is from 2015 TX24, the stronger the influence of the resonance. The precision of the semimajor axis is not good enough to verify that. If so, that could mean that the breakup happened rather recently, perhaps as recently as a few centuries ago.

This supports a model for the Taurid Complex showers that involves an ongoing fragmentation cascade of comet 2P/Encke siblings that were created following a breakup some 20,000 yr ago (Jenniskens 2006). In this scenario, 2015 TX24 broke with 2005 UR and 2005 TF50 from a larger precursor body about 1500 yr ago, and in the past few centuries a further breakup of 2015 TX24 created the fragments observed as the s-Taurids today. Asteroid 2003 WP21 does not belong to this group and was created earlier.

It is more difficult to understand how we could be seeing activity from comet 2P/Encke. In that case, the resonance must have more dramatically changed the nodal line. Comet 2P/Encke would be expected to cause meteoroid activity centered on λ_⊙ = 2246, using method "H" of Hasegawa (1990), implemented by Neslusan et al. (1998). However, Encke now has nodes close to perihelion near Mercury and aphelion in the asteroid belt. Comet Encke's orbit has changed its longitude of perihelion significantly over the past two centuries. In 1769, the node was at 15746, Pi = 15919. In 2020, the node is at 15455, Pi = 16111. That rate of change is about the same as seen in the meteor stream. Around AD 1700, the longitude of perihelion of Encke was the same as that of the core of the s-Taurids. If we are seeing meteoroids ejected from 2P/Encke in the century around AD 1700, there must have been a subsequent dramatic change of the nodal line.

It is not clear whether or not the periodic nature of the s-Taurids is due to the 7:2 MMR with Jupiter, other than preventing close encounters with Jupiter. The mean semimajor axis of the meteoroids appears to be related to that of its parent body. It is perhaps possible that the dust was released with very low relative ejection speeds and has remained concentrated in the range of mean anomaly owing to its young age. In particular, it is interesting that Spurný et al. (2017) noted that the 1995 detected Taurid fireballs did not quite have the orbital elements of the 2015 Taurids. They had larger semimajor axes and smaller perihelia that did not change so much with solar longitude. This could point to the presence of streams being more dispersed but narrow, which occasionally wander in Earth's path.

Brown et al. (2013) identified the Taurid showers as a potential source of macroscopic meteorite dropping events if a large enough meteoroid enters Earth's atmosphere. We have seen that the STS branch contains large members; do members of that population have a chance of surviving entry and falling as a meteorite? Large s-Taurids behave like weak matter (Figure 5). They experience catastrophic disruption at very high altitudes (>66 km; see Table A1). Their weak nature is not compensated by size, as Spurný et al. (2017) noted that the larger s-Taurids tend to be the weaker ones. The deepest penetrating STS observed by the DFN (DN151114_04) is not visible below 52 km. According to the criteria of Brown et al. (2013), which states that a height of 35 km and velocity of 10 km s⁻¹ are approximate terminal dynamical criteria for a given event to have a chance of producing a meteorite fall, this is unlikely to produce a recoverable meteorite on the ground. The two very bright STSs described by Olech (2016) and Spurný et al. (2017) also terminate at high altitudes of 57.86 and 60.20 km.

Is this weakness a feature of all Southern Taurids? The deepest penetrating Southern Taurid (MORP #715) described in the MORP data set (Halliday et al. 1996) only penetrates to 54.8 km. As outlined by Brown et al. (2013), one of the EN fireballs in 1995 penetrated as deep as 30 km. Although this fireball was tentatively linked to the Taurid Complex, no definite association with either branch of the Taurids was reported, and the final velocity was not reported either. More generally, to our knowledge there is no report in the literature of a Southern Taurid that comes close to the terminal parameter of Brown et al. (2013).

On the other hand, we have examples of Northern Taurids that are able to penetrate much lower than the 50 km ceiling that Southern Taurids seem to hit. For example, on 2016 October 9, the DFN observed a Northern Taurid penetrating as deep as 36.4 km, slowing down to 9.7 km s⁻¹; the terminal parameters for this NTA are much closer to the cutoff criteria of Brown et al. (2013), but still greater than required for a meteorite to survive. Are we seeing inhomogeneities here in the original composition of 2P/Encke? Or are the stronger materials representative of older meteoroids that survived the harsh conditions in the interplanetary medium?