1 Introduction

The Draconid meteor shower is a minor periodic shower produced by meteoroids released from the Jupiter family comet, 21P/Giacobini–Zinner. The shower is active between 6 and 10th October, with a peak occurrence on 8th October. Both the Draconid meteoroid stream and its parent comet are of peculiar nature with an orbital period of 6.6 years and perihelion distance of 1.03 AU. It has been noticed that the comet is the most carbon depleted (A’Hearn et al. 1995), and the meteors are slow moving with relative velocity of 23 km/s. The shower sometimes produces brief but spectacular meteor storms. Such storms were occurred twice during last century, in 1933 and then thirteen years later in 1946 (Jacchia et al. 1950), during which more than 10,000 meteors per hour were recorded. In recent years, dust trail theories (e.g. Maslov 2011; Vaubaillon et al. 2011) predicted a small outburst activity on October 8, 2011 around 20 h UT. The observers around the globe became active and various observations confirmed the predicated outburst (e.g. Koten et al. 2012; Kero et al. 2012; Ye et al. 2013).

The “light curve”—variation of luminous intensity of meteor against time has been used since the 1950s to understand the physical structure and chemical composition of meteoroid, which is an indicative of ablation processes during meteoroid flights in the atmosphere. On the basis of photographic light curves obtained during the 1946 storm, Jacchia et al. (1950) first recognized peculiar properties of the Draconids and concluded that these meteoroids were composed of “softer” material, more easily melted or vaporized than that of other meteors. Further, the trail lengths and durations were found to be three times shorter than that of other shower meteors and showed much larger deceleration. This study also found that Draconids appeared and disappeared at greater heights than sporadic meteors of similar speeds. Further evidence for peculiar properties of the Draconids became familiar after photographic study using Super-Schmidt cameras from two stations and then concluded that fragmentation was a common phenomenon among all photographic meteors (Jacchia 1955).

The light curve of radar meteor is defined as the pulse-integrated Signal to Noise Ratio (SNR) as a function of time, which is analogous to the light curve of optical meteors. Observational evidence suggest that the meteoroids that generate radio meteor trail undergo fragmentation. Many studies have been carried out to show the influence of fragmentation on meteor light curves and different models (dustball, grain and others) were proposed. Numerous authors have used the light curve shapes as evidence of meteoroid fragmentation. A comprehensive review may be found in Ceplecha et al. (1998). Fragmentation of radar meteors has been investigated by Elford and Campbell (2001) and Elford (2001). Elford (2001) has observed monostatic pulsating meteor head echoes with the Adelaide 54 MHz MST radar, and interpreted these events as signatures of meteoroid fragmentation, which produced an interference pattern in the echoes. Elford and Campbell (2001) estimated fragment separation distance and differential velocity by considering the increasing pulsation rate to be caused by interference from two scattering centers. They also noted fragmentation followed by down the beam Range Spread Trail Echo (RSTE) events.

In recent years, variability in the meteor head echo structure has received much attention. Mathews et al. (2008) examined the aspects of radar meteor studies to provide insight into the meteor head echo scattering mechanism and meteoroid fragmentation and they concluded that the fragmentation was the cause of the complex light curves. Kero et al. (2008) report three types of light curves—a smooth light curve event which simply outlines the antenna beam pattern and a “beat pattern” light curve event, shown to be consistent with two slowly separating particles, while yet more complex third type, found to be consistent with three or more fragments and/or continuous fragmentation. Briczinski et al. (2009) and Sugar et al. (2010) discuss the radar meteor head echo “terminal events” that are substantial through the meteor lifetime and then suddenly disappears below the noise level. Mathews et al. (2010) reported many previously unaccounted features in the radar meteor return that are consistent with meteoroid fragmentation including “terminal flare” events. Malhotra and Mathews (2011) presented the first ever statistical study of meteoroid mass loss or disintegration mechanisms.

However, so far, less attention has been paid to radar studies of the Draconid meteoroid fragmentation, except few previous observations (eg.,Greenhow and Neufeld 1957; Novikov et al. 1994). Besides this, a few photographic studies have been published earlier (e.g., Jacchia et al. 1950; Jacchia 1955) and video studies recently (e.g., Borovička et al. 2007). In this context, the present study focus on the 2011 Draconids activity using radar observations and meteor fragmentation and light curve shapes were discussed by taking present meteor ablation models into consideration.

2 Radar System and Meteor Detection

Observations were carried out at the Gadanki (13.5\(^\circ\)N, 79.2\(^\circ\)E) MST radar on 8/9 October 2011, scheduled from 12h30m to 00h30m UT, to cover the expected Draconid meteor outburst. With 53 MHz frequency, Gadanki MST radar is a highly sensitive pulse-coded phase coherent Doppler radar with a peak power aperture product of \(3 \times 10^{10}\,\mathrm{Wm}^2\). The antenna system consists of 1,024 three element Yagi–Uda antennas with inter-element spacing of \(0.7\lambda\), arranged in a \(32 \times 32\) matrix over an area of \(130\,\mathrm{m}\times 130\,\mathrm{m}\). The transmitter total peak power is about 2.5 MW which is achieved by 32 transmitters whose output power ranges from 15 to 120 kW, each feeding a sub array of 32 Yagis. It generates a radiation pattern with a main beam of width \(3^\circ\), gain of 36 dB and a side lobe level of \(-\)20 dB. In principle, the main beam can be positioned at different look angles in NS and EW plane with a maximum of \(20^\circ\) off zenith angle. Further system description and technical details of the Gadanki MST radar can be found in Rao et al. (1995). Though, the present radar facility was originally designed for lower and middle atmospheric dynamics, this narrow beam high power radar is also ideal for meteor studies (Reddi and Nair 1998).

For the present study, uncoded pulse of 4 \(\upmu\)s pulse width has been used with an inter-pulse period (IPP) of 1,000 \(\upmu\)s. For optimum detection of meteors, we used four beams of MST radar, viz. Zenith-X, two oblique beams inclined at an angle of \(20^\circ\) from zenith (i.e, East-20 & West-20) and another beam North-13 inclined at \(13^\circ\) from zenith. With a range resolution of 1.2 km starting at 79.95 km, 34 range bins were sampled in sequential beam orientations. The time series data as a sequence of In-phase (I) and Quadrature (Q) values from four selectable antenna beam orientations with header information can be stored in data acquisition computer. Four successive I and Q—samples for each range bin were coherently averaged, making the effective sampling interval of 0.004 s and each data frame had continuous data for 2.048 s. When an echo in one or more range bins exceeded an average floor noise of a specific threshold (1.35 V), for at least three successive transmitter pulses, the frame is identified to contain backscattered echo from the meteor trail and is subsequently archived for further analysis.

The raw data sample collected from the system was processed by converting into Range Time Intensity (RTI) plots, and these plots were manually examined by off-line display. The heads and trails were identified through manual inspection of the SNR, wherein most of the cases, there exist a clear gap between the head echo and trail echo with a reduced SNR. In our data set, nearly 20 % of all the head echoes produced trail reflections. As expected, a strong head echo is more likely to produce trail echo. When a head echo cannot be easily distinguished from a trail echo by just looking at the RTI plot, we used power plots in which the received power at the height where the echo lasted the longest has been plotted to isolate the head from the trail. The Gadanki MST radar detected head echoes, RSTE reflections, and other low-latitude ionospheric echoes, especially in North-13 beam. However, in present study we focus only on head echoes with down the beam RSTE reflections in multiple range gates along with beat pattern echoes showing fragmentation.

3 Meteor Analysis and Fragmentation

The results consists of approximately 740 meteor echoes detected with the Gadanki MST radar on 8/9 October, 2011, which includes head echoes, RSTE reflections in multiple range gates and beat pattern echoes. However, as mentioned earlier, in this study, we focus only on head echoes with down the beam RSTE reflections in multiple range gates along with beat pattern echoes that showing fragmentation. This sample statistical study leads to conclude that at the 53 MHz narrow beam Gadanki MST radar of low latitude, only a quarter of high SNR meteor events exhibited clear signature of fragmentation. We claim that the rest 75% of events are too small to show fragmentation signature.

Figure 1a, b displays RTI plots for a typical meteor event that represent many of the 2011 Draconids. From the figure, it can be noticed that the head echo followed by a long duration trail reflection is extending over more than five range gates. This kind of echoes observed if the meteor trajectory is perpendicular to the radar beam and are believed to be overdense echoes, i.e., the plasma frequency of the meteor trail exceeding the probing radar frequency. It can be noticed that the head echo is prominent over a large number of range gates and trail echoes are observed to form comparatively at lower altitude.

Some of the echoes exhibit a temporal gap between the head and the trail and some do not. Previously, to explain this, two different mechanisms have been proposed, known as the “glint theory” and the “blob theory” (McKinley 1961, §8 p. 219–224 for an overview). The glint theory assumes that the ionization is produced smoothly and continuously along the trail. But, after examining the faint photographic meteor light curves, Jacchia (1955) commented that the early part of the light curve frequently displays appreciable irregularities, a rapid rise of light followed by a steady decline, which is a reversal of classical trend. As time goes on, wind shear in the upper atmosphere distort the trail more and more. The gap may be due to either the distortion of the trail by wind shear (glint theory) or uneven ionization of the trail (blob theory). The large range extension of the trail echoes can be explained by agreed with the basic postulate of both glint and blob theories that the scattering medium needs to be overdense. However, in our observation the delay between the head and trail echo is very small, and hence the applicability of the glint or blob theory may not be taken for granted. And it is possible that the blob theory together with the meteoroid fragmentation concept may be more appropriate. In such case, fragmented meteor form numerous ionization centers which then expand and part of the surface of the expanded plasma eventually becomes specular reflectors (Zhou et al. 2001).

Fig. 1
figure 1

The RTI plot and light curve of a typical 2011 Draconid meteors detected at a 23:36:29 in W-20 and b 20:24:25 in \(Z_{x}\) respectively

Full size image

In our data set fragmenting meteors are easily distinguishable from normal meteors due to the fact that meteor fragmentation may produce echoes at points on the trail far from the point of perpendicular reflection and these echoes are characterized as multiple echoes at different ranges. The RTI plots of Fig. 1a, b are also useful to examine any indication of fragmentation in the radar meteor echoes. Figure 1a shows a fragmentation features that appears around 0.45 s after the first echo from the meteor, at a height of about 110 km, and persists nearly 1.25 s of the recorded signal. Similarly, Fig. 1b show fragmentation at around 0.40 s at a height of 100 km, and which persist for nearly 1.6 s before it gets buried in the noise. Sudden drastic decrease in echo strength at fragmentation is assumed to be due to a smaller meteoroid that generates less scattering plasma (Mathews 2004).

Fig. 2
figure 2

The RTI plot and light curve of a typical 2011 Draconid meteor echoes detected at a 21:20:42 in W-20 and b 21:43:54 in W-20 respectively. These kind of echoes were confined mainly within one or two range gates over an extended period of \(\ge\)1.5 s

Full size image

To explain meteoroid fragmentation, in blob theory, the echo power as a function of time should appear to be more or less smooth. However, in our case, the observed echoes are not smooth. This can be better seen from the light curves or SNR plots, i.e., power variation as a function of time at selected altitude. In both the light curves 1a, b, the echo at 0.4 s is head echo, while the trail echo begins at about 0.5 s. The large temporal variation of the echo power may imply the existence of multiple overdense blobs arriving at different times. Fragmentation, which is known to occur very frequently for large meteors at least can provide multiple scattering centers (Meisel et al. 1995). In our observation, the light curves of the Draconid meteors show large scatter, as seen in Fig. 1a, b.

Another type of the Draconid meteor trail reflections that detected with the Gadanki radar are characterized by an extended detection period, generally greater than 1.5 s, in only one or two range gates. Such trail reflections may or may not be accompanied by a head echo such as shown in RTI plots Fig. 2a, b. These kind of echoes mainly confined to one or two range gates over an extended period of \(\ge\)1.5 s, and classified as type 1 echoes by Zhou et al. (1998), who detected by using the Arecibo 46.8 MHz VHF radar. Since there is an extended time period with essentially no temporal change in the echoing range, when a meteoroid travels perpendicular to the radar line of sight, then it is believed that the echo can stay within one or two range gates. In this viewing geometry, a head echo cannot be easily distinguished from a trail echo by just looking at the RTI plot. To identify whether an echo is a head or trail echo, light curve, plotted at the height range where the trail reflection remained for longest period is of great help.

The light curves in Fig. 2a, b exhibit a well-defined signal strength before becoming invisible in the background noise. The smooth decay is indicative of a classical underdense echo, where decrease in echo strength is due to the increase in trail diameter resulting from ambipolar diffusion (Zhou et al. 1998). It is evident from the light curves that all fragmenting echoes are not of the same. The echoes shown in Fig. 2a, b are more complicated and that cannot be explained by simple classical underdense scattering theory alone. It is possible that both overdense scattering and head echo scattering are involved. When the trail is overdense, the received echo power does not change very much as long as the trail remains overdense (McKinley 1961). The echoes shown in Fig. 2a, b are likely overdense echoes, the longer duration with lower power of the echoes may indicate that these echoes are observed farther off the beam center. From Fig. 2a, the echo can be largely divided into three stages: (1) from 0.5 to 0.8 s—head echo; (2) from 0.8 to 1.5 s—overdense scattering; and (3) from 1.5 to 2.0 s—underdense scattering. Similarly, the echo shown in Fig. 2b can be categorized.

Figure 3a, b are RTI plots and light curves of a typical fragmenting meteor events that show “beat pattern”. The beat pattern can be noticed even without the aid of the light curve. Figure 3a shows beat pattern echo at the height range of nearly 92 km, while in Fig. 3b at 114 km, and also notice a weak head echo in the same height range, before beat pattern echo. The light curve show interference effects among the fragments, but are otherwise smooth. Earlier, the presence of beat pattern has already suspected to be due to a fragmentation of meteoroid, first by Evans (1965) and later by Elford (2001), provided the first strong evidence for beat pattern, which is the signature of meteoroid fragmentation. Kero et al. (2008) provide an example of a beat pattern light curve event and interpret it as being due to interference from two distinct scattering centers. In present study, we ignore the low-SNR events (SNR \(\ge\) 2) as in the case even small change in received power might be wrongly interpreted as beat pattern due to fragmenting meteor event. We argue that the pulsations or beat pattern in light curves are generated by interference between radio wave reflections from more than one distinct ionized region due to two meteoroid fragments simultaneously present in the transmitter radar. These fragments may have slightly unequal velocities with respect to each other, hence produce diverse interference patterns. Fragmentation consistent with two or three major fragments, each producing head echo that in turn produce distinctive interference pattern are often observed.

Fig. 3
figure 3

The RTI plot and light curve of a typical “beat pattern” echo of 2011 Draconids detected at a 22:52:08 in W-20 and b 23:45:24 in N-20 respectively

Full size image

Meteor occurrence rate variation over a 1.2 km interval of radar range resolution for the Draconid shower is as shown in Fig. 4, where the solid line represent the Draconids and dotted line represents the sporadic meteors. The mean height of occurrence is 95.0 km, and we find that the Draconid trails are formed on an average a few kilometers higher than sporadic meteors of comparable SNR or radio brightness. Higher altitude detection of Draconids have also been found previously among photographic (Jacchia et al. 1950) and video graphic meteors (Suzuki et al. 1999; Fujiwara et al. 2001; Koten et al. 2007). From RTI plots of down the beam head echoes, it can also be note that the beginning heights as a function of intensity (SNR), the beginning heights of the Draconid meteors lie between 100 and 110 km. The vast majority of other meteors start their luminous trajectories lower than the Draconids. If the higher beginning height reflects the meteoroids structure, then the high altitude detection confirms the extreme fragility of the Draconid meteoroids as outlined by Jacchia et al. (1950).

Fig. 4
figure 4

The height range (km) variation of the observed meteor echoes on 8/9 October, 2011. Here the solid line represent the Draconids and dotted line for the sporadic meteors

Full size image

4 Discussion

Meteoroid fragmentation is a very complex process, but it is very important in understanding how the meteoroid mass deposition takes place and thus how it affects the aeronomy of the upper atmosphere. For years together this subject is of much discussion and debate in research community. In literature, a few papers deal with theoretical light curves, thus allowing in principle, a comparison of observationally determined light curves.

The classical light curve produced by a compact solid meteoroid body is assumed to be smooth and reaches the maximum luminosity near the end of the trail. However, observations indicated that the solid-body ablation model may not be holds good (Jacchia 1955). According to this study, the majority of the Draconid meteoroids are porous and fragile bodies that can easily breaking into fragments. This is in accordance with the icy-conglomerate comet model of Whipple (1951), that later known as meteoroid dustball model (Öpik 1955). In consistent with the dustball model, Hawkes and Jones (1975) introducing the concept of a “glue”, explained that the meteoric bodies composed of numerous small grains with a high melting point are held together by a lower melting point “glue”. In this model, the grains were released after the glue is evaporated, i.e. before the grain ablation started. According to Hawkes and Jones (1975), the dustball model of meteoroids should produce nearly symmetrical light curves. However, the light curves were found to vary greatly from meteor to meteor, but on average were nearly symmetrical (e.g. Murray et al. 1999), which is generally considered as proof of the dustball model. Beech (1986) analyzed photographic data of Jacchia et al. (1950), of the 1946 Draconids and found good agreement with the dustball model of Hawkes and Jones (1975). Later, Fisher et al. (2000) stated that most meteoroids consists of hundreds to thousands of fundamental minute grains, at least some of which are released prior to the onset of intensive ablation and become aerodynamically separated during atmospheric entry. Campbell et al. (2000) argued that fragmentation occurs before the process of ablation starts: the grains are released, and then they undergo intensive ablation. The most advanced dustball meteoroid model was created by Campbell-Brown and Koschny (2004). They modeled heating of dustball meteoroids and considered ablation before the boiling temperature was reached, using the Clausius–Clapeyron formalism.

The dustball model predicts the weak dependence of the beginning height, maximum height and trail length of the visible meteor on the mass of the meteoroid (Campbell et al. 2000; Beech 1986) and the production of wake (grains of different masses ejected from the meteoroid) will decelerate at different rates, producing physical wake from the spread in the grains (Fisher et al. 2000). Oscillations or beat pattern in echo power are also seen from ALTAIR radar, which interpreted as interference of echo returns from closely spaced fragments (Close et al. 2004, 2005). Kero et al. (2008) has noticed pulsations in received echo power in European Incoherent Scatter (930 MHz EISCAT) radar data and suggested that plasma effects or meteoroid rotation as other possible causes. In terms of the dustball meteoroid model, by assuming the power law distribution of meteor mass, Beech and Murray (2003) modeled the meteor light curves and compared with observational data recorded during the 1998–2001 Leonid storm. This study found no correlation between the light curve shapes and age of the meteoroids.

In reaching our conclusions, we present “type specimen” meteor events that serve to define the presence of meteoroid fragmentation processes. We utilize the Figs. 1, 2 and 3 events defining the presence of fragmentation in the 2011 Draconid meteor that leads to conclude that meteoroid fragmentation is a dominant process in meteoroid interaction with the atmosphere. In broad outline, the detection of meteor fragmentation and its interference hypothesis presented here is in consistency with that of Elford and Campbell (2001). Our results suggest that meteoroid arrival, and mass loss in the upper atmosphere as observed by the Gadanki MST radar is a complex process in which fragmentation mechanisms play a dominant role. Though, the percentage of meteors showing fragmentation is only a quarter of the others, this percentage is dominate, because, the mass deposition process by fragmentation that presumably introduce nanometer sized dust into the meteor detection altitudes of classical HF/VHF low power small aperture radar (Elford and Campbell 2001). We argue at an observational level that meteoroid fragmentation and terminal flaring play an important role in meteoroid mass deposition in the upper atmosphere. As the Draconids are slow moving meteors, as they enter into any of the radar range gates, the trail completely get distorted immediately in only one or two range gates. This may cause that the most of the head and trail echoes, we detected, are in one or two range gates only as shown in Fig. 2. The oscillating or beat pattern echoes shown in Fig. 3 can be interpreted as being due to alternate in-phase and out of phase scattering or change in separation between multiple particles as per the modeling results given by Mathews et al. (2010) and Malhotra and Mathews (2011). The beat pattern echoes observed in the present study are in consistent with that of Kero et al. (2008) and also with the theoretical model.

5 Conclusions

The MST radar observations of the 2011 Draconid meteor shower activity on 8/9 October show a number of head echoes with RSTE reflections in multiple range gates along with “beat pattern” echoes, those are consistent with meteoroid fragmentation. The head echoes that followed by a long duration trail echo, extends over many range gates are observed when the meteor trajectory is perpendicular to the radar beam direction. Observed head echoes with long duration trail reflections, can be explained by the blob theory together with the meteoroid fragmentation concept, where the fragmented meteor form numerous ionization centers that eventually reflect radar signals. The range aliased meteor trails that completely get distorted immediately in one or two range gates are because of slow moving nature of the Draconids. The presence of beat pattern in the power profiles (ie., SNR plots) may be due to alternate in-phase and out of phase scattering or change in separation between multiple particles.

The 2011 Draconids light curves showed a wide scatter among them, which may be due to high fragile nature of the Draconid meteor particles. The mean height of occurrence of Draconids obtained from the present observations is 95.0 km, and we find that the Draconid trails are occurred, on the average few kilometers higher than sporadic meteors of comparable SNR or radio brightness. From the observations it is found that the beginning heights of the Draconids lie between 100 and 110 km. If the higher beginning height reflects the meteoroids structure, then the high altitude detection confirms the extreme fragility of the Draconid meteoroids as outlined by Jacchia et al. (1950). If the higher beginning height does not reflect the meteoroid structure, then the deceleration of meteor occurs before the start of substantial ablation. As the higher beginning height meteor do not penetrate very deep into the atmosphere and therefore suffer less deceleration, hence, it gives the true structure of the meteoroid. The beginning height of the meteor can also affected by the different zenith distance of the radiant as well as the different initial velocities. Hence, large zenith distance meteoroids have higher average beginning heights (Kero et al. 2008). Higher beginning height of head and trail echoes as shown in RTI plots, confirm the extreme fragility of Draconid meteoroids. The prolongation of the meteor as terminal reflection towards the end is qualitatively in agreement with the observations. Our observations confirm the fragile nature of meteoroids produced by comet 21P/Giacobini–Zinner.