High-resolution radar observations of meteoroid fragmentation and flaring at the Jicamarca Radio Observatory Free

Abstract

1. INTRODUCTION

As fast micrometeoroids (micro to milligram in mass; speeds ∼11–70 km s⁻¹) enter the Earth's upper atmosphere, significant atmospheric interactions in the form of differential ablation (Janches et al. 2009), fragmentation (Mathews et al. 2001; Stokan & Campbell-Brown 2014) and sputtering (Spurný et al. 2000; Gao & Mathews 2015a,c) occur, yielding observable radar and optical meteors over the ∼70–200 km altitude range. Details concerning these atmospheric interactions remain controversial in the meteor community as will be discussed below.

The sputtered or ablated atoms and ions surrounding the meteoroid ionize the air molecules they encounter thus forming a region of plasma around the meteoroid along with the trail plasma. With a sufficiently sensitive (HPLA: high-power, large-aperture) radar, the region of plasma surrounding the meteoroid and travelling with the meteoroid at approximately the same velocity is observed as the radar meteor head-echo (Mathews 2004). The trail of plasma left in the wake of the meteoroid is often observed as the radar meteor trail-echo. When the radar beam is pointed perpendicular to this ionization trail, the classical under-dense trail-echo is often observed (McKinley 1961) – this is also known as the specular trail-echo. Also observed are echoes from evolving ‘point’ flare plasmas that occur when the meteoroid or a particle from the meteoroid terminally flares (explodes) creating an intense but rapidly diffusing and recombining plasma that also includes meteoroid dust and smoke (Briczinski, Mathews & Meisel 2009; Mathews et al. 2010; Malhotra & Mathews 2011). The various trail- and flare-echoing plasma regions are embedded in the atmosphere.

Although radar meteor head-echoes were first observed in the 1930s (Mathews 2004 and Baggaley 2009 give histories of meteor radar), they were not seriously studied until they were observed via HPLA radars in the 1990s (Chapin & Kudeki 1994b; Pellinen Wannberg & Wannberg 1994; Mathews et al. 1997; Zhou, Mathews & Nakamura 2001). Since then, meteor head-echo observations have been successfully applied in determining meteoroid velocities (Janches et al. 2000; Janches, Nolan & Meisel 2003), mass flux (Janches, Meisel & Mathews 2001; Mathews et al. 2001; Close 2005; Fentzke et al. 2009), and radiants (Chau, Woodman & Galindo 2007; Kero, Szasz & Nakamura 2011). However, studies of meteoroid mass-loss – and thus meteoroid mass flux – based on meteor head- and specular trail-echo observations using various radars remains a source of intense discussion in the community. Issues include the relative roles of meteoroid fragmentation – including flaring (seen as an evolving echo due to particle ‘instantaneous’ destruction to plasma) – and ablation (including differential ablation).

The physical mechanism(s) behind fragmentation of micrometeoroids is poorly understood. However, we can roughly divided fragmentation into two categories, gross and continuous fragmentation (Ceplecha et al. 1998). Gross (macroscopic) fragmentation is defined as the process during which the meteoroid progressively splits into a small number of discernible fragments that may form separate head-echoes. Alternatively, the meteoroid may undergo catastrophic one-step disintegration yielding a great number of small fragments that may form a very complex radar scattering region. Conversely, continuous (microscopic) fragmentation, whereby numerous small particles continuously detach from the parent meteoroid and then ablate or flare, likely does not form easily interpretable radar meteor interference patterns (Mathews et al. 2010). Regardless of origin, larger fragments that terminally flare tend to create altitude-narrow flare-trail echoes that are motionless relative to the head-echo formed by the parent meteoroid. Continuous fragmentation may produce enhanced regions of the light curve that resemble differential ablation (Janches et al. 2009). Gross fragmentation may be more commonly observed for bright radar meteors and be a result of greater initial mass meteoroids that are perhaps composed of many similar-sized constituent grains.

Considering the brightness and resolution requirements, photographic and video techniques have been more sensitive to the larger and faster meteoroids that result in the brighter meteors (e.g. Spurný et al. 2000). Thus, although fragmentation has been observed and studied in the optical meteor community since the 1950s, the mechanism for fragmentation in faint meteors – in general resulting from the meteoroids less massive than 10⁻⁴ kg – is poorly understood.

To explain the anomalous observations from faint meteors photographed with the Super-Schimidt cameras, Jacchia (1955) provided a description of fragmentation which, we suggest, remains useful:

‘…fragments can be detached from the surface of larger meteor bodies without destroying their unity; but if fragments of similar size are detached from small bodies, this may mean their complete disruption into a cluster of fragments. Larger meteors, then, will disintegrate only toward the end of their trajectories, while among fainter meteors the breakup may occur at earlier stages, even at the very beginning of the visible trail. What we obtain by integrating the brightness of a faint meteor is not the mass of a single body but a function of the total mass of all the fragments.’

In addressing Jacchia's concerns, Hawkes & Jones (1975) proposed the ‘dustball’ model for the overall structure of meteoroids. This model suggests that ‘meteoric bodies are composed of grains that are held together by a lower boiling point glue. Once the boiling point of the glue is reached grains are detached. Light is assumed to be produced only by detached grains’. This model has been shown to be consistent with the observed light curves of both fainter and brighter optical meteors, thus further strengthening the case for fragmentation. Recently, image intensified optical studies have shown several examples of gross fragmentation in Leonid (Murray, Hawkes & Jenniskens 1999) and sporadic (Hawkes et al. 2004; Stokan & Campbell-Brown 2014) meteoroids in this size range, but the development of a quantitative model for fragmentation has been hindered by a small number of observations.

The radar meteor community has not considered fragmentation as an important mechanism with the implicit assumption that radar-echo-producing meteoroids are ‘too small to fragment’ (Madiedo & Trigo-Rodriguez 2008). As a result of this assumption, few if any radar meteor observations considered fragmentation an issue until the 2000s. Elford & Campbell (2001) suggested that the absence of Fresnel oscillations in many meteor specular trail echoes was due to interactions between multiple ablating fragments of the same parent meteoroid dispersing along the trajectory, rather than the ‘overdense’ trail plasma resulting in specular ‘kinks’ along the trail. With validation from Very High Frequency (VHF) radar observations and simulations, Elford & Campbell (2001) concluded that fragmentation is a common feature of meteoroids prior to or during ablation in the atmosphere.

Using the tri-static 930 MHz European Incoherent Scatter Scientific Association (EISCAT) Ultra High Frequency (UHF) radar system, Kero et al. (2008) provided the first strong evidence of a submillimetre-sized meteoroid breaking apart into two distinct fragments. They suggested that the presence of a beat pattern or pulsations in the received power of meteor echoes is consistent with two particles and thus of fragmentation. Roy et al. (2009) also argued that the beat pattern in radar meteors observed using the Poker Flat (Alaska) Incoherent Scatter Radar operating at 449.3 MHz was generated from multiple, closely spaced bodies travelling at nearly identical speeds. They employed a genetic-algorithm-based procedure to determine the properties of the individual meteoroid fragments including relative scattering cross-section, speed and deceleration.

Based on common-volume observations using the V/UHF radars at the Arecibo Observatory, Mathews et al. (2010) concluded that meteoroid fragmentation is a dominant process in meteoroid interaction with the atmosphere, as manifested by intrapulse and pulse-to-pulse interference (in the manner of Young's two-slit optical experiment) between multiple meteor head-echoes and between head-echo and impulsive flare-echoes. Malhotra & Mathews (2011) statistically studied the relative roles of simple ablation, differential ablation and fragmentation in interpretation of the meteor observations by HPLA radar. They argued that the meteoroid mass-loss mechanisms are complex and involve all three mechanisms mentioned above, instead of the emphasis of one or the other mechanisms as indicated in the previous references.

As variously noted above, it is challenging to record and study faint optical meteors and associated processes. We argue that observing the corresponding radar meteors with radar interferometry (imaging) promises new information on fragmentation. Moreover, with interferometry, it becomes possible to image the fragments in three spatial dimensions. Ultimately, the goal is to bring new understanding of the fragmentation mechanisms for small meteoroids and their interaction with the atmosphere as well as gain knowledge of meteoroid structure and composition.

In this study, we present and interpret five representative radar meteor events observed at the Jicamarca Radio Observatory (JRO) in Peru. For this study, we introduce a hybrid interferometric-CS method (H-ICS), which is a significant extension to 3D of the compressed sensing (CS) method discussed in Zhu et al. (2015). Our study builds in part on the insights of Volz & Close (2012) who convincingly introduced CS to the study of radar meteors. They applied the technique to JRO 13-baud Barker code meteor results yielding higher range and Doppler frequency resolution of radar meteors than the matched filter based method.

Applying H-ICS processing to the five typical events we have selected has revealed unprecedented details (and further puzzles), such as the spread of meteoroid fragments due to gross fragmentation, both perpendicular and parallel to the meteoroid trajectory. Field aligned irregularity (FAI) development related to fragmentation is also analysed. These observational studies will form the basis for future modelling efforts for meteoroid mass-loss and fragmentation mechanisms. The observational setup and parameters are explained in Section 2. The results and their interpretations are presented in Section 3. A summary and discussion of this investigation is presented in Section 4.

2. OBSERVATION SETUP

The radar meteor events discussed in this paper are all from observations carried out at the JRO on 2010 April 15th. Early results and technique details from these observations are given in Gao & Mathews (2015a,b). For these observations the antenna beam was positioned near zenith pointing ∼1.5° geomagnetic north from the vertical and towards the geomagnetic equator. The full Jicamarca antenna array can be divided into four quarters (Fig. 1a). The west (B) and east (not labelled) quarters of the Jicamarca antenna array were combined for transmission. For reception, three quarters A, B and C, as shown in Fig. 1(a), were provided with independently sampled receivers for interferometry. The phase centres of the A,B and B,C legs are separated by a distance of 147 m. The coordinate system used for the 3D spatial imaging is shown in Fig. 1(b). In order to match the Y-axis to geomagnetic north (approximately parallel to the geomagnetic B field direction), we computationally rotate the interferometry baselines (BC and BA) counter-clockwise by θ = 51.07^o to obtain the X–Y axis showed in the Fig. 1(b).

Figure 1.

(a). Full JRO antenna array and receiver configuration for meteor observations. Three quarter-array sections, A, B and C, are used for interferometric receiving. (b). The 3D spatial imaging coordinate system used in the analysis of observation results.

Open in new tabDownload slide

Pointing is calibrated by taking advantage of the presence of the highly aspect sensitive equatorial electrojet (EEJ) scattering, that defines the k ⊥ B (k is radar wave number; B is geomagnetic field vector) locus at JRO and thus the geomagnetic equator (as shown in Fig. 1(b); Kudeki & Farley 1989). Other pointing calibration paradigms are given by Gao & Mathews (2015b). The transmitting half-power beam width is 2.2° perpendicular to the connecting diagonal and 1.1° along the diagonal (Fig. 1a), while the three quarter arrays each have individual full width at half-maximum beam widths of approximately 2.2° (see Gao & Mathews 2015c, figs 2 and 3, for all beam patterns).

Figure 2.

(a) & (b). RTI plots of Event 20100415_023541 before and after H-ICS processing with normalized SNR indicated by the colour bar. (c) The light curves of the three receivers: A, B and C. Note the strong interference (beat) pattern as the meteor enters the beam followed the terminal flaring. (d) The spatial imaging results shown in the 3D X–Y–Z domain and its 2D projections with time (s) given by the colour bar. As shown in the left-bottom plot (the X–Y plane) in (d), this event exhibits a range-rate ν_f = 7.5 km s⁻¹ while the vector velocity ν ≈ [−8.0 4.9 −7.4] km s⁻¹ with magnitude |ν| ≈ 12 km s⁻¹ at a zenith angle θ ≈ 51.8°. The relative mean-square error (e) of the fitted line is ∼0.002%. Note the evolution of the single flare-trail echo.

Open in new tabDownload slide

Figure 3.

(a) & (b). RTI plots of Event 20100415_025203 before and after the H-ICS processing with normalized SNR indicated by the colour bar. (c) The light curves of three receivers: A, B and C. This event is similar to the Fig. 2 event but with both intermediate and terminal flaring. (d) The spatial imaging results shown in the 3D X–Y–Z domain and its 2D projections with time (s) given by the colour bar. As shown in the left-bottom plot (the X–Y plane) in (d), this event exhibits a range-rate ν_f = 15 km s⁻¹ while the vector velocity ν ≈ [11.8 7.2 −15] km s⁻¹ with magnitude |ν| ≈ 20.3 km s⁻¹ at a zenith angle θ ≈ 42.7°. The relative mean-square error (e) of the fitted line is ∼0.002%. Note the evolution of the two flare-trail echo regions.

Open in new tabDownload slide

A 20 μs uncoded pulse sequence with an interpulse period (IPP) of 2 ms yielding a 300 km unaliased range window was employed. The oversampled, uncoded pulse yields superior phase and Doppler sensitivity compared with, say, Barker codes. The complex (in-phase and quadrature) received signals were sampled at 1 MHz yielding the basic 150 m range resolution. The principal radar parameters for our observations are summarized in Table 1. More observational details are given in Gao & Mathews (2015a,b).

3. RESULTS AND DISCUSSION

In this section, we discuss five diverse but, in our experience, typical meteor events from this ∼4 hr 2010 April 15 observation set. Events similar to these appear to comprise the vast majority (∼74%) of the JRO meteors observed below ∼90 km for this dataset. The three-receiver observational data for each of these events is analysed using the H-ICS technique. In this approach, we combine the CS paradigm with interferometry providing very high 3D spatial resolution. Detailed mathematical and algorithmic descriptions are given in Zhu et al. (2015) and references therein; however a brief overview follows.

CS is an efficient signal processing method based on the non-linear sampling theorem (Candes 2006). It can achieve enhanced resolution with sub-Nyquist sampling given the condition that the desired signal is sparse or approximately sparse in an appropriate basis (Candes & Romberg 2006). In this paper, by introducing aspects of CS, we achieved high spatial resolution for meteor observations with only three receivers (one baseline measurement for each X and Y direction, see Fig. 1). Better imaging results are expected using the new observations conducted in 2014 with seven receivers (more baseline measurements for higher resolution horizontal imaging).

Each meteor event discussed below is identified using the format 20100415_XXXXXX, indicating the date (2010 April 15) and the local time XX:XX:XX in hours, minutes, and seconds from local midnight of the record containing the event. The processed results for each event are given in four different plots: the unprocessed range–time–intensity (RTI) plot where the intensity is given by the signal-to-noise ratio (SNR in dB; colour bar), the RTI plot after the H-ICS processing (SNR in dB; colour bar), the combined light curves (event range integrated SNR in dB versus time) from the three receivers (A, B and C), and the 3D spatial imaging results (after integration in the Doppler frequency domain) shown in 3D as well as in 2D projections, colour-encoded by discrete time (seconds). The 3D imaging Cartesian coordinate system is shown in Fig. 1(b).

3.1 Event 20100415_023541

Fig. 2 shows a relatively simple but representative meteor event. This event appears to show that atmospheric ‘processing’ of even relatively small meteoroids is more complex than expected via classical ablation (Bronshten 1983). The unprocessed RTI plot (Fig. 2a) shows a clear interference pattern in both the head- (before ∼0.15 s) and trail- (after ∼0.15 s) echo regions. This deep, periodic modulation of the SNR is also strikingly manifest in the Fig. 2(c) light curves. While a rise and then sudden drop of the received signal can be interpreted as differential ablation (Janches et al. 2009), this common, periodic pattern in the head-echo can be better explained as multiple fragments/scattering centres – in this case already present when the meteor appeared in the beam – travelling with slightly different velocities. That is, the received head-echo signals from these multiple, slowly separating fragments (that act as point scatters in that they exhibit well-defined phase centres) alternately add in- and out-of-phase as the net path to the receivers varies over half a wavelength (Mathews et al. 2010). Fig. 2(b) shows the RTI plot after the H-ICS processing which yields higher range resolution but does not resolve the separate fragments. The flare-echoing region (after ∼0.15 s) is also not resolved.

The 3D spatial imaging results are given in Fig. 2(d) as four separate spatial projections of this event as a function of the time given by the progressive colour scale. The spread of the points in Fig. 2(d) is thought to result from mutual interaction (interference pattern) between multiple fragments (head-echoes) comprising the meteor.

This event exhibits a range-rate of ∼7.5 km s⁻¹ while the magnitude of the velocity vector is ∼12 km s⁻¹ at a zenith angle of 51.8° (as shown in the left-bottom plot of Fig. 2d) on an approximate SE-to-NW trajectory in the magnetic coordinates shown in Fig. 1.

As stated previously, the Fig. 2 event appears to be consistent with at least two major initial fragments generating interfering head-echoes over the observed trajectory. As seen in Fig. 2(c), flaring started at ∼0.08 s with the final terminal flare occurring at ∼0.12 s. An interference pattern within the flare echoes is seen, suggesting that at least two individual flares occurred. The flaring generates altitude-narrow trails, referred as the terminal flare-echoes (Mathews et al. 2010). Flaring apparently results in a small ‘blob’ of plasma imbedded in the atmosphere thus yielding a range-narrow, trail-like echo (with a well-defined phase centre) that evolves as the blob expands via parallel-B diffusion (Malhotra, Mathews & Urbina 2007a; Malhotra, Mathews & Urbina 2007b). This evolving FAI structure (Fig. 2d) exhibits a weak classical Fresnel oscillation (McKinley 1961; Mathews 2004) as it expands. Fig. 2(d) shows that evolution of the flare-echo region (after ∼0.15 s) is parallel-B (along the Y-direction) as most clearly seen in the horizontal (X–Y) projection of the event trajectory. The vertical extent of this flare region indicates the diffusion rate in the Y direction is faster than in the X direction.

To obtain the exponential decay time τ, we use the light curves shown in Fig. 2(c). We first find the initial peak (marked by the black dashed line) and fit a line to the decay curve (marked by the red/blue/pink dashed lines) from this peak, then from the slope, find τ and thus D via equation (2). The effective diffusion rate of this event is ∼1.79 m² s⁻¹ near 86 km altitude.

3.2 Event 20100415_025203

Event 20100415_025203 (Fig. 3) resembles the previous event, 20100415_023511 (in Fig. 2), in that as the meteor enters the beam, it displays the same characteristic periodic interference pattern that is consistent with two or more fragments. However, in contrast to the Fig. 2 event, it then undergoes a bright intermediate flare (∼0.1 s) and then a final terminal flare (∼0.17 s) as seen in the unprocessed RTI plot (Fig. 3a) and the H-ICS RTI plot (Fig. 3b). These flares are characterized by two separate short flare-echoes at heights between 87 and 90 km. The semiperiodic interference pattern continues after the first flare suggesting that at least two particles continue to the final flare. The entire event exhibits an average radial speed of ∼15 km s⁻¹ with a net speed of ∼20.35 km s⁻¹ (Fig. 3d). The meteoroids travelled from SW-to-NE at 42.7° zenith angle.

In Fig. 3(a), it is important to also note the tapered interference pattern between the first flare-echo and the continuing head-echoes which reinforces the point that both the altitude-narrow flare-echo and the head-echo can be treated as point targets exhibiting well-defined scattering (phase) centres. In Fig. 3(d), we can see clearly that each flare event generates a small blob of plasma that results in an evolving specular (FAI) flare-echo. As noted before, from the light curve of these evolving flare-echoes (Fig. 3c), we find that the apparent diffusion rate at ∼89 km is ∼1.69 m² s⁻¹, which is approximately the same as for the Fig. 2 event. This result is expected as the time interval between these events is less than half an hour. Note that both the head-echo and flare-echo regions are wider in the Y direction than in the X direction apparently due to parallel-B diffusion, as indicated in the X–Z and Y–Z projections of Fig. 3(d). This suggests that we are actually observing the trail-echo development, i.e. the transition from head-echo plasma to trail-echo plasma (e.g. figs 10 and 11, Stokan & Campbell-Brown 2015), in addition to flare-echo (FAI) regions.

3.3 Event 20100415_051949

Event 20100415_051949, depicted in Fig. 4, shows a clear beat pattern in the light curve at all three receivers, as illustrated in Fig. 4(a) and (c). However, no obvious flaring is observed and the event features are substantially different from the earlier examples. The range-rate speed of this meteor is ∼9 km s⁻¹ indicating a very shallow trajectory into the atmosphere. We split this event into two segments, before and after crossing the null in the antenna beam pattern at ∼0.07 s, and compute separate trajectory linear fits for both segments. The first segment yields a net speed of ∼24.2 km s⁻¹ at zenith angle ∼66° and the second segment a net speed ∼21.5 km s⁻¹ at zenith angle ∼65.3° (fitting errors are given in the caption). Overall, this event appears to have two or more components travelling from NE-to-SW.

Figure 4.

(a) & (b). RTI plots of Event 20100415_051949 before and after the H-ICS processing with normalized SNR indicated by the colour bar. (c) The light curves of three receivers: A, B and C. (d) The spatial imaging results shown in the 3D X–Y–Z domain and its 2D projections with time (s) given by the colour bar. As shown in the left-bottom plot (the X-Y plane) in (d), this event has two components: the initial Z > 88 km component (event 1) exhibits a range-rate ν_f = ∼9 km s⁻¹ while the vector velocity ν ≈ [−16.0 −11.0 −9.0] km s⁻¹, and its magnitude |ν| ≈ 21.5 km s⁻¹ at a zenith angle θ ≈ 65.3°. The relative mean-square error (e) of the fitted line is ∼0.003%. The second event (event 2) exhibits a range-rate ν_f = 9.5 km s⁻¹ while the velocity ν ≈ [−18.4 −12.2 −9.9] km s⁻¹, and its magnitude |ν| ≈ 24.2 km s⁻¹ at a zenith angle θ ≈ 66°. The relative mean-square error (e) of the fitted line is ∼0.981%. Note the head-echo interference or beat pattern and that no flaring is seen. As discussed in the text, the early part of this net event appears to be a separate and likely unrelated event.

Open in new tabDownload slide

The apparent beat pattern seen in Figs 4(a) and (c) appears, simply, to resemble the characteristics of the previous events, shown in the Figs 2 and 3. However, the spatial 3D imaging results as well as the linear fit results discussed above and shown in Fig. 4(d) reveal that while the main (after ∼0.07 s) event is indeed likely comprised of two or more fragments, there is a curvature in the overall trajectory. We can see clearly the change of direction (slope) between the two segments, and the larger mean square error in the fit (given by e in Fig. 4d) for the longer second segment, which indicates the progressive change of direction (slope). Additionally, the initial Z > 88 km component is Y-elongated (∼1 km) compared with the X-extent (∼0.4 km) suggesting that this is due to an otherwise unobserved flare-echo of a completely separate event.

The deep beat pattern shown in Figs 4(a) and (c) suggests that at least two major meteoroid particles move through the beam yielding the major features of this event. This conclusion is supported by the progressive change of slope (tilting) at individual range gates (separated by ∼150 m) in the X–Z and Y–Z projections in 4(d). This may be due to the slightly different velocities and decelerations of individual scattering centres that are also horizontally separated, as the net (multiparticle) event progresses.

3.4 Event 20100415_045609

Event 20100415_045609, shown in Fig. 5, is another example of a possible multiparticle meteor similar to the Fig. 4 event. Figs 5(a) and (c) show a mild interference pattern although the nulls near ∼0.07 s and ∼0.16 s result from the meteor entering antenna pattern nulls. The spatial imaging of the event, shown in Fig. 5(d), reveals two or more separate and perhaps unrelated events with little evidence of flaring. We investigate by fitting the trajectory of the event before and after the null with a line, as for the previous event.

Figure 5.

(a) & (b). RTI plots of Event 20100415_045609 before and after the H-ICS processing with normalized SNR indicated by the colour bar. (c) The light curves of three receivers: A, B and C. (d) The spatial imaging results shown in the 3D X–Y–Z domain and its 2D projections with time (s) given by the colour bar. As shown in the left-bottom plot (the X–Y plane) in (d), this event has two components: the initial component exhibits a range-rate ν_f = 4 km s⁻¹ while the velocity ν ≈ [3.7 −16.5 −3.6] km s⁻¹, and its magnitude |ν| ≈ 17.2 km s⁻¹ at a zenith angle θ ≈ 78.1°. The error of fitting to a line e ≈ 0.001%; the second component exhibits a range-rate ν_f = 3.1 km s⁻¹ while the velocity ν ≈ [2.0 −11.9 −3.1] km s⁻¹, and its magnitude |ν| ≈ 12.4 km s⁻¹ at a zenith angle θ ≈ 75.5°. This error of fitting to a line e ≈ 1.015%. As discussed in the text, this net event is likely two nearly overlapping but separate events.

Open in new tabDownload slide

As with the previous events, it seems that two separate meteors have been observed for this event. This is evident from the change in trajectory, comparing the segment of this event before the null crossing (0.07 s) and after. A mild interference pattern is seen near the centre of the return (∼0.11 s in Fig. 5a), which also suggests two overlapping events. The net direction of these events is magnetic north with the Y-width at each range gate being significantly more than the corresponding X-width. A mild flare may have occurred towards the end. If so, it is partially masked by the continuing head-echo. These events are at ∼77° zenith angle, have very small radial speeds of 3–4 km s⁻¹ and net speeds of ∼12 and ∼17 km s⁻¹. This event does not appear to be related to space debris re-entry.

3.5 Event 20100415_041635

Event 20100415_041635, shown in Fig. 6(a), looks at first like a classical, specular meteor trail-echo with rapid onset as the meteoroid crosses perpendicular to the radar beam (t₀; McKinley 1961; Mathews 2004) followed by rapid (Fresnel) oscillations as the trail further unwinds. This interpretation appears to be supported by the CS result shown in Fig. 6(b) although the return is two or three range gates in extent. Oscillations in the light curve, Fig. 6 (c), are shallow, and returns from the three receivers show significant differences. The Fig. 6(d) 3D projections show the returns from this event are unexpectedly wide in both X and Y dimensions but wider in the Y (parallel-B) direction. As with the previous two events, it seems that this event is comprised of two distinct events, separated by 150 m (one range gate) vertically, and 100 m in the X–Y plane. The observed strong fading of the signal in the interference pattern (Fig. 6a and c) again suggests that the individual scattering centres must be small (and distinct) compared to the separation between them.

Figure 6.

(a) & (b). RTI plots of Event 20100415_041635 before and after the H-ICS processing with normalized SNR indicated by the colour bar. (c) The light curves of three receivers: A, B and C. This event appears to be a classical trail-echo but, as discussed in the text, is in fact much more complex. (d) The spatial imaging results shown in three 2D projection domains with time (s) given by the colour bar. Note that the different time spans have been displaced slightly in range so that the evolution of these echoes can be seen more clearly.

Open in new tabDownload slide

The Fig. 6(c) light curves yield an apparent diffusion rate of this ‘trail’ of ∼4.2 m² s⁻¹ which is over twice that of the events shown in Figs 2 and 3 that occurred about two hours earlier. We examined the diffusion rate from other flare-echoes at about the same time and altitude range as event 20100415_041635 (Fig. 6) – these yielded ∼1.7 m² s⁻¹ apparent diffusion rates. We interpret this event to be similar to the event shown in Fig. 3 (in which a clear head-echo and embedded flare- and/or terminal-flare echoes are observed), but with the progenitor meteor head-echo having been missed due to its having occurred outside of the antenna beam. That is, it appears that we are observing the rapidly evolving, net scattering features from two slightly separated flare-trail scattering centres. Thus, as the flare plasma evolves it is advected and/or diffuses into the beam in a manner like that discussed in Malhotra et al. (2007a,b).

It is important to note that the time (colour scale) and vertical pattern of two range-gate groups is different from the previous figures. In Fig. 6(d) for each time segment (colour) at both range gates, a slight vertical displacement has been included so that the net echo time evolution can be clearly distinguished. Using this approach, we note that echo horizontal diversity is largest in Y-extent at the beginning and that it gradually becomes narrower and centred to the south with little change in the X-extent but with a tilting (Z-separation) due to actual slight altitude differences the different scattering centres.

It is difficult to understand why the Fig. 6 interference pattern is so different from those from the events in Figs 2 and 3. As the combined interference among the various scattering centres evolves, it generates a complex Fresnel interference pattern that is similar to a classical meteor echo (Elford & Campbell 2001; Roy et al. 2009). However, this could be the result of two (or more), slightly offset but parallel classical trail events. In this interpretation note that, as the JRO antenna beam is directed at near zenith, a classical meteor trail-echo would imply that the meteoroid is travelling tangential to the surface of the Earth. This would seem unlikely, though certainly not impossible, at these low altitudes.

4. CONCLUSION

The five meteor events discussed in Section 3 appear to be representative of the majority (∼74% of the total of 78 events in this 4 hr dataset) of the low altitude (≲90 km) events observed at JRO. Among these 78 events, 21% show the multiple trail-echoes that resemble event 20100415_025203 (Fig. 3), 13% show interference patterns that resemble Figs 4 and 5. While these events may appear to be relatively simple from an RTI perspective, 3D imaging reveals that they are complex, exhibiting possible fragmentation and flaring, while two of the events may have actually been comprised of overlapping, independent events.

While meteoroid fragmentation is observed and accepted for optical meteors, the same cannot be said for radar meteors. This appears to be due to the perception that the generally smaller meteoroids attributed to radar meteors are thought to be homogeneous and so not susceptible to fragmentation although differential ablation is well accepted (Janches et al. 2009). However, the micrometeoroid flaring and fragmentation observations provided in this paper, which seem to resemble the ‘nebulous’ and fragmenting meteors shown by Murray et al. (1999, figs 11 and 12) and the striking faint optical meteor fragmentation results given by Stokan & Campbell-Brown (2014, figs 2, 5 and 7), demonstrate that radar meteors are generated by processes considerably more complicated than simply classical ablation – these observations perhaps offer some direction to the study of fragmentation.

All five events presented here occurred over the 85–90 km altitude range at the lower end of the radar meteor zone (Briczinski et al. 2009). Further, it is important to note that the first four events showed clear evidence of fragmentation as these events first entered the beam. That is, the fragmentation process began above 90 km altitude, where the density of atmosphere is much lower. Additionally, multiple, slightly separated scattering centres with different velocities/decelerations could generate the curved (not linear) trajectories shown in Figs 4(d) and 5(d). The optical meteor analogue would be those transversely distributed meteors given in Stokan & Campbell-Brown (2014).

The mechanisms responsible for the faint and complex radar meteors observed in this study must be investigated further. Meteoroid fragmentation and flaring and, in two cases, apparently independent, nearly overlapping events appear necessary to explain the observations. However, we certainly need more sophisticated radar meteor observations – including common-volume optical observations – to evaluate and improve the existing physical models. The ‘dustball’ model mentioned in the introduction (Hawkes & Jones 1975) seems at this point to be a an excellent beginning.

Finally, we emphasize that the majority of the JRO low altitude meteor events are complex exhibiting both fragmentation (interference patterns) and flaring (flare-echoes). This is in contrast to the Stokan & Campbell-Brown (2014) results for which 311 of some 1800 optical meteor events exhibit gross fragmentation results. As a result, we suggest that coherent radar imaging is a powerful and novel method for investigating small meteoroid fragmentation and flaring.

This effort was supported under NSF grants ATM 07–21613 and AGS 12–02019 to The Pennsylvania State University.

REFERENCES