Dependence of radar signal strength on frequency and aspect angle of nonspecular meteor trails

Click Here for Full Article JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113,, doi:10.1029/2007ja012647, 2008 Dependence of radar signal strength on frequency and aspect angle of nonspecular meteor trails S. Close, 1 T. Hamlin, 1 M. Oppenheim, 2 L. Cox, 1 and P. Colestock 1 Received 16 July 2007; revised 9 January 2008; accepted 18 February 2008; published 3 June 2008. [1] When a meteoroid penetrates Earth s atmosphere, it forms a high-density ionized plasma column immersed in the ionosphere between approximately 70 and 140 km altitude. High-power, large-aperture (HPLA) radars detect nonspecular trails when VHF or UHF radio waves reflect off structures in a turbulent meteor trail. These trails persist from a few milliseconds to many minutes and the return from these trails is referred to as nonspecular trails or range-spread trail echoes. In this paper, we present analysis of nonspecular trails detected with ALTAIR, which is an HPLA radar operating simultaneously at 160 MHz and 422 MHz on the Kwajalein Atoll. First, we investigate the aspect sensitivity of nonspecular trails and show that as the angle between the radar beam and the background magnetic field increases, the signal strength falls off 3 to 4 db per degree at 160 MHz. For ALTAIR, this means that the aspect angle must be within approximately 12 degrees in order to detect nonspecular trails using the chosen waveforms. Second, we compare and contrast the meteoroids that form nonspecular trails and find that the meteoroid energy causes much of the variability in the nonspecular trail s signal-to-noise ratio (SNR) for a given aspect angle. In addition, we show two range-resolved fragmentation events that also affect the SNR. Finally, we determine the dependence of SNR on wavelength using two wavelengths and show that the maximum nonspecular trail SNR scales as approximately l 6, with a variation that depends upon altitude. Citation: Close, S., T. Hamlin, M. Oppenheim, L. Cox, and P. Colestock (2008), Dependence of radar signal strength on frequency and aspect angle of nonspecular meteor trails, J. Geophys. Res., 113,, doi:10.1029/2007ja012647. 1. Introduction [2] Meteoroids entering the Earth s atmosphere ablate large numbers of atoms between 140 and 70 km altitude. These ablated atoms, traveling at more than 11 km/s, subsequently collide with air molecules, creating expanding columns of partially ionized plasma called meteors. Highpower, large-aperture (HPLA) radars, such as the Advanced Research Projects Agency (ARPA) Long-Range Tracking and Instrumentation Radar (ALTAIR), measure short duration returns called head echoes when the beam scatters off the dense plasma that surrounds and moves with the ablating meteoroid. Often, tens of milliseconds after the head echo vanishes, more persistent echoes return from the expanding plasma column left in the meteoroid s wake, referred to as trails. When the incident radar beam and trail are almost exactly perpendicular, these strong returns are called specular trails and arise from Fresnel scattering. Specular echoes form the basis for classical meteor radars [Sugar, 1963]. When the radar beam does not lie perpendicular to the trail but is rather quasi-perpendicular to the 1 Los Alamos National Laboratory, Los Alamos, New Mexico, USA. 2 Center for Space Physics, Boston University, Boston, Massachusets, USA. Copyright 2008 by the American Geophysical Union. 0148-0227/08/2007JA012647$09.00 background magnetic field, we refer to these trails as either nonspecular [Close et al., 2002] range-spread echoes [Mathews, 2004] or spread meteor echoes [Reddi et al., 2002]. These trails have been detected using both HPLA radars [Zhou et al., 2001; Close et al., 2002] and low-power radars [Haldoupis and Schlegel, 1993] [Reddi and Nair, 1998; Reddi et al., 2002] and result from Bragg scattering from field-aligned irregularities (FAI) in the trail. [3] The first observations of meteor trails generating irregularities that scatter at VHF were given by Heritage et al. [1962]. Further observations and analysis of nonspecular trails was presented by Chapin and Kudeki [1994] using the Jicamarca Observatory, which is an HPLA radar operating at 50 MHz, as well as Haldoupis and Schlegel [1993] and Reddi and Nair [1998] using low-powered radars operating at 50 MHz and 53 MHz, respectively, during the same time period. Haldoupis and Schlegel [1993] detected nonspecular trails, referred as meteorinduced backscatter, with abrupt onsets and lifetimes up to 3 min. Their data showed large velocities, including an ion acoustic velocity at 320 m/s. Reddi and Nair [1998] were the first to show the altitude-dependent time delay between the head echo and the trail and definitively conclude that the two-stream plasma instability is the mechanism that generates the field-aligned irregularities detected by radar. Chapin and Kudecki hypothesized that gradient drift and two-stream plasma instabilities driven by electrojet electric 1of8

fields caused these unique signatures. Oppenheim et al. [2000] and Dyrud et al. [2001] utilized plasma simulations and theory to show that trails, even without strong external fields, will often develop Farley-Buneman/gradient drift (FBGD) waves that become turbulent, generate fieldaligned irregularities (FAI), and exhibit anomalous crossfield diffusion, generated by the FBGD instability [Dyrud et al., 2001]. Zhou et al. [2001] used the MU radar at 50 MHz to show that trails could be observed when pointed perpendicular to the background magnetic field (B) (but not in the parallel to B configuration), and by Close et al. [2002] using the ALTAIR radar operating at 160 MHz and 422 MHz, simultaneously, when pointed nearly perpendicular to B. Other HPLA radars that do not point perpendicular to B, such as EISCAT, Arecibo, and Millstone Hill, do not routinely detect nonspecular trails. [4] More recently, Zhou et al. [2004] studied the aspect sensitivity using a simple model with a wavelength of 6.4 m, which assumes that trail turbulence is strictly field-aligned but localized spatially, making the turbulence visible from a range of aspect angles. They found that although the maximum power from nonspecular trails will be detected when the radar is pointed perpendicular to the magnetic field, substantial power can still be obtained in the offperpendicular direction, contrary to thin ionization layers. Specifically, they showed that for an FAI of 20m in length, the power observed 6 degrees off perpendicular to B is approximately 10 db below the perpendicular to B direction. However, even though the direction of incident radar energy, relative to the background magnetic field, clearly dominates the detectability of nonspecular trails, we stipulate that further dependencies must arise from variations in the incident electromagnetic wave (i.e., the frequency and polarization), as well as the parent meteoroid s size, shape, composition, trajectory, and velocity. [5] In this paper, we examine the aspect sensitivity of nonspecular trails detected by ALTAIR and associate these with meteoroid parameters. In particular, we expand upon the work of Zhou et al. [2001] in order to determine the cutoff for nonspecular trail detection by steering the AL- TAIR beam away from the perpendicular to B direction in small increments. In addition, we estimate the frequency dependence (at two frequencies) of nonspecular trails by calculating the maximum SNR at both 160 MHz and 422 MHz. Sections 1.1 and 1.2 contain the instrumentation and observation descriptions, respectively. Section 2 discusses the aspect sensitivity of nonspecular trails. Section 3 shows the frequency dependence at two frequencies on trail signal strength, and section 4 shows the correlation between meteoroid parameters and trail signal strength. Section 5 summarizes and concludes. 1.1. Instrumentation [6] ALTAIR is a 46-m diameter, high-power, twofrequency radar operating at 160 MHz (VHF) and 422 MHz (UHF), located in the central Pacific at 9 N and 167 E (geographic) on the island of Roi-Namur in the Kwajalein Atoll, Republic of the Marshall Islands. ALTAIR transmits a peak power of 6 MW simultaneously at the two frequencies with right-circularly (RC) polarized signal energy in a half-power beam width (6 db down) of 2.8 and 1.1 at VHF and UHF, respectively. The full beam width (to the first sidelobes) is 7 at VHF, and 2.8 at UHF. ALTAIR receives both right-circular and left-circular (LC) energy and has four additional receiving horns for the purpose of angle measurement (both azimuth and elevation), which gives the position of an object in three dimensions by inclusion of its range, and hence the 3-D velocity and deceleration as well. 1.2. Observations [7] Radar meteor data were collected at ALTAIR on 18 November 1998, during a 4-h period at fixed pointing in 2-min segments, and subsequently in 1999 during a 6.5-h period in 2-min segments resulting in 51.9 min of VHF data from the two experiments. ALTAIR showed a peak detection rate of 1.6 VHF head echoes per second during the 1998 shower, for a total of 749 detected head echoes. During the 1999 shower, a peak detection rate of 1 VHF head echo per second was observed for a total of 441 detected head echoes. A total of 41 nonspecular trails were detected or approximately one trail every 1.3 min. The pointing directions of the radar, with respect to the angle between the ALTAIR beam and B associated with the Leonid 1998 and 1999 showers, are given in Figure 1. These data show the experiment time in GMT for both years; note that ALTAIR is 12 h ahead of GMT. The sampling is a result of "following" the Leonid radiant as it traversed the sky. Note that we were also primarily pointing at the North Apex (sporadic/nonshower apparent) source and have low sensitivity for the other major (and slower) sporadic sources; at times 1515 and 2020 we were pointed off-radiant. The vast majority of our detections (>98%) are sporadic in origin. [8] Amplitude and phase data were recorded for each frequency and four receiving channels for altitudes spanning 70 to 140 km at VHF and 90 to 110 km at UHF. The beam width and the range sampling correspond to a total collecting area of 1502 km 3 at VHF and 58 km 3 at UHF. The two ALTAIR waveforms used to collect the data were a 40 ms VHF chirped pulse (30 m range resolution), and a 150 ms UHF chirped pulse (7.5 m range resolution). A 333 Hz pulse-repetition frequency was utilized for its sufficiently high sampling rate, which allows the calculation of 3-D velocity as a function of altitude. Using these waveforms, ALTAIR has a sensitivity of 64 db and 81 db at VHF and UHF, respectively, for a single pulse on a 1 m 2 target at 100 km range. Using a mean velocity of 56 km/s, this corresponds to a lower mass limit of approximately 6 10 7 g for these waveforms using the spherical scattering theory described in the work of Close et al. [2004]. [9] For the nonspecular trail data, we computed the mean noise (at both LC and RC) in each 1-s interval for the full altitude extent at VHF and UHF. We then converted the raw amplitude and phase data to power and divided the entire 1-s data interval by the noise power. 2. Aspect Sensitivity of Nonspecular Trails [10] Nonspecular trail detection is thought to be a direct result of the growth of plasma instabilities. The time sequence associated with this process was summarized by Oppenheim et al. [2000], who showed that an ambipolar E 2of8

Figure 1. Plot of the angle between ALTAIR boresite and the magnetic field (B) as a function of observation time. Both the 1998 and 1999 observing times are included in the plot. field (perpendicular to both the geomagnetic field and the trail) first develops, followed by growth of plasma concentration waves on the edges of the trail arising from the gradient drift/farley Buneman (GDFB) instability, and then, finally, plasma turbulence. This turbulence breaks the trail into multiple segments, which forms FAIs that radars detect as nonspecular trails. The aspect sensitivity of nonspecular trails, meaning the ability of radars to detect these FAIs is not completely understood. [11] In order to determine the cutoff for detecting nonspecular trails, we steered the ALTAIR beam from the perpendicular to B direction in small increments in azimuth and elevation and stared for 2 min at each particular aspect angle. We then calculated the number of trails that were detected during each 2-min segment. We detected a total of 41 nonspecular trails at VHF and only three at UHF, which were also simultaneously detected at VHF. Figure 2a shows the resulting data from the 41 VHF trails and includes the number of nonspecular trails detected per minute for each data collect as a function of the angle between ALTAIR pointing and B. Figure 2b shows the number of nonspecular trails detected per head echo for each data set as a function of the angle between ALTAIR pointing and B. The difference between Figures 2a and 2b most likely stems from the fact that we span approximately 8 h in time as we decrease the angle between ALTAIR and the magnetic field, which reflects the diurnal variation in meteor detection [Janches et al., 2006]. From these data, the cutoff for trail detection appears to be between 77.8 degrees and 78.4 degrees, or 12 degrees from perpendicular at 160 MHz; a coarser cutoff would lie between 71 and 78 degrees and cannot be discounted due to our small sampling. The variability of detection with angle, in particular the low numbers of trails detected at the very highest angles to the magnetic field, is most likely due to the small sampling of nonspecular trails. However, as noted from Figure 1, as our pointing becomes more perpendicular to the magnetic field, we are also progressing further toward daytime. Indeed, all of our near perpendicular to B pointing occurs after sunrise. As M. M. Oppenheim (personal communication, 2007) has shown, nonspecular trail detection is also highly dependent on time of day, with the largest number of trails being detected at night when the background ionospheric density is lowest. Therefore, as we progress toward daytime, we would expect our numbers to decrease, even though we are pointing more perpendicular to the background magnetic field. [12] In order to compare our observations to Zhou s numerical prediction, we computed the maximum LC power and duration from the 41 nonspecular trails in order to determine a dependence on aspect angle. We chose the maximum LC power from each trail and then further chose the maximum LC power across all trails detected at each particular aspect angle and plotted these as a function of angle and duration. These data are contained in Figure 3; the color relates to the maximum duration in seconds. Although the sparse data does not support a detailed examination, this plot suggests at a trend with aspect angle. Specifically, we find a 17 db drop in maximum LC power over 6 degrees. This agrees well with the 12 degree cutoff, meaning if we project another 17 db drop for the next 6 degrees, this brings us close to our noise value and would therefore be undetectable. By comparing to Zhou et al. [2004, Figure 2], Figure 2. Plot of (a) the number of nonspecular trails per minute as a function of the angle between ALTAIR boresite and the magnetic field and (b) the number of nonspecular trails per head echo as a function of the angle between ALTAIR boresite and the magnetic field. The aspect sensitivity cutoff, from these data, lies between 77.8 degrees and 78.4 degrees. 3of8

Table 2. Associated Head Echo Power Dependence on Frequency and Polarization LC RC VHF (db) UHF (db) x VHF (db) UHF (db) x 2042 41.3 30.5 2.6 35.3 19.2 3.8 2100 42.3 48.5 1.4 26.2 16.4 2.3 2118 44.9 33.7 2.7 43.4 30.2 3.1 Figure 3. Plot of maximum LC power from all trails detected at each aspect angle as a function of aspect angle and color-coded for duration (in seconds). this corresponds to a nonspecular trail length (along B) of 24 m. Since Zhou used a wavelength of 6.4 m (MU radar), his trail length/wavelength ratio is 3.75. For our wavelength of 1.9 m (at VHF), the corresponding trail length is approximately 7.1 m (Q. Zhou, personal communication, 2007). Note, however, that given the small statistic, these numbers may be subject to large errors and further work is needed in order to establish this trend. 3. VHF Versus UHF SNR of Nonspecular Trails [13] Of the 41 nonspecular trails detected at VHF, only three were detected simultaneously at UHF, giving a ratio of 14 VHF trails for every 1 UHF trail detected. This is in contrast to the VHF to UHF ratio of head echoes, which gives a ratio of 6 VHF head echoes for every 1 UHF head echo detected [Close et al., 2002]. The primary factors affecting the VHF to UHF ratio include the following: (1) relative volume sample between the VHF and UHF data sets, including 1502 km 3 and 58 km 3, respectively, (2) weighting due to antenna gain, (3) the k-perpendicular-b locus inside each of the two beams, (4) dependence between SNR and plasma density. Factors 1 and 2 can be immediately excluded, since these would also affect the head echo detection rate between VHF and UHF. In the next section, we will investigate factors 3 and 4 and show that the primary reason for the differences in detection rate between head echoes and trails appears to be that trails have a steeper fall-off of signal strength with radar frequency. [14] Table 1 contains the maximum trail power in db at both LC and RC for the three nonspecular trails detected at 2042, 2100, and 2118 simultaneously at VHF and UHF. Table 2 contains the data for the associated head echo. The Table 1. Meteor-Induced Backscatter Power Dependence on Frequency and Polarization LC RC VHF (db) UHF (db) x VHF (db) UHF (db) x 2042 44.7 18.5 6.2 39.3 10.5 6.8 2100 51.5 23.8 6.6 40.6 14.8 6.1 2118 55.3 24.0 7.4 39.7 21.8 4.2 column denoted with a cross symbol indicates the wavelength dependence to the measured signal strength, i.e., (SNR is proportional to l x ). While some variation is evident, it is clear that the nonspecular trail dependence on wavelength is much greater, on the order of x 6, while the head echo dependence is closer to 2. It is worth noting that the nonspecular scattering strength dependence on wavelength is similar to that reported by Eshelman [1954] for short wavelength trails (i.e., echo durations that are comparable with the trail s formation time). Eshelman also showed that signal strength is proportional to l 6. [15] To clarify this point further, we show a trail detected simultaneously at VHF and UHF and include its position relative to the main beam as well as the k-perpendicular-b crossing; the VHF trail was originally published in the work of Close et al. [2002]. These data are given in Figure 4 and show the VHF trail (Figure 4a), VHF trail zoomed (Figure 4b), UHF trail (Figure 4c), UHF trail zoomed (Figure 4d), position of the trail relative to the VHF beam, UHF beam, and k-perpendicular-b crossing, color-coded for altitude (Figure 4e), and position of the trail relative to the VHF beam, UHF beam, and k-perpendicular- B crossing, color-coded for LC SNR (Figure 4f). First, in Figure 4b and 4d it is evident that this meteoroid experienced a fragmentation event. The 30-m range resolution of the VHF data is suggestive; however, the 7-m range resolution of the UHF data shows conclusively that this meteoroid indeed fragments. These data also show that the maximum power for this trail is located near the center of both the VHF and UHF beam and is approximately 1.5 away from the k- perpendicular-b crossing. Therefore, we conclude that the observed peak power for this particular trail depends more heavily on position within the beam and we can exclude factor 3 (relative position to k-perpendicular-b) as a factor in the strong wavelength dependence. 4. Correlating Meteoroid Parameters With Nonspecular Trails [16] As noted in the introduction, the SNR of nonspecular trails is not only highly dependent on the radar pointing direction with respect to the magnetic field but also on the properties of the parent meteoroid. In this section, we investigate the properties of the meteoroids that create the nonspecular trails. [17] Figure 5 shows six images of nonspecular trails detected at 1820 GMT on 17 November 1998. The number on each image corresponds to the maximum power of each trail (range-spread signal) as well as its associated head echo (streak). Two features are immediately obvious. The first is the time delay between the head echo streak and the nonspecular trail that extends out to the right from each streak. 4of8

Figure 4. Trail detected at 2042 simultaneously at VHF and UHF showing (a) VHF, (b) VHF zoomed showing the fragmentation event, (c) UHF, (d) UHF zoomed showing the fragmentation event, (e) position of the trail relative to the VHF beam, UHF beam, and k-perpendicular-b crossing, color-coded for altitude, and (f) position of the trail relative to the VHF beam, UHF beam, and k-perpendicular-b crossing, color-coded for LC SNR. The lines in Figures 4e and 4f correspond to the k-perpendicular-b crossing and the k-perpendicular-b crossing in 1 increments. As noted in the work of Close et al. [2002], the range- Doppler coupling constant is approximately 2 ms and 3 ms for the VHF and UHF waveforms, respectively; therefore this effect does not account for the observed delays, which are typically larger than 20 ms (with an altitude dependence). Recall that Range-Doppler coupling is a property of a chirp-type pulse, where Doppler shifts of the radar echo cause an offset in the apparent range of the echo. Dyrud et al. [2001] attributed the delay between head and trail to the time needed for plasma instabilities to grow strong enough to be visible to the radar. They showed that for a simulated trail at 105 km, gradient drift Farley-Bunneman waves begin to appear after only 1 ms and turbulence begins at approximately 20 ms. The typical time delay of 20 30 ms between a head echo and a nonspecular trail is consistent with their theory that we are observing reflections from 5of8

Figure 5. Images of trails detected at 1820, which corresponds to an angle of 84.4 degrees between ALTAIR boresite and B. The number on each plot corresponds to the maximum RCS of the associated head echo for each trail. The trails were detected at (a) 1820:42, (b) 1820:54, (c) 1821:10, (d) 1821:24, (e) 1821:25, and (f) 1821:54. The strongest trail Figure 5d also happens to correlate to the head echo with the highest RCS. these segments. The second notable feature is the variation between the peak powers for each of the six trails. This collection time corresponds to an angle of 84.4 degrees between the ALTAIR beam and the magnetic field; therefore the aspect sensitivity is the same for all six trails and cannot account for this variability (although some variability will occur due to position of the trail within the beam). [18] Although the head echo power or radar cross section (RCS) is a possible choice for the differences in nonspecular power, the data shows that this is not the dominant trend. Figure 6 shows the head echo RCS as a function of altitude for the head echo/nonspecular trail pairs, including 41 head echoes, as well as for head echoes without trails, including 427 head echoes; these data only correspond to times when 6of8

Figure 6. Plot of altitude versus RCS for all 427 head echoes and the 41 head echoes detected with a corresponding nonspecular trail. These data indicate that there is no or little correlation between head echo RCS and nonspecular trail detection. nonspecular trails could be detected (i.e., angles less than 12 degrees perpendicular to the magnetic field). This plot shows that no such correlation exists with head echo RCS, meaning both head echoes and head echoes that had an associated trail were equally likely to create large or small RCS values. As noted in the work of Close et al. [2004], the head echo RCS is dependent upon many factors, including peak plasma density, plasma radius, radar parameters (polarization, frequency, etc.), and meteoroid parameters (mass, velocity, etc.). Therefore, this is an expected result. [19] In contrast, the meteoroid mass does seem to show some correlation with detection of a nonspecular trail. Figure 7 shows the meteoroid mass, calculated using the head echo spherical scattering theory [Close et al., 2004, 2007] as a function of altitude. Figure 7a shows all of the data, including 427 head echoes and 41 head echoes that have a trail pair. Figure 7b shows only data that were collected at an angle of 90 ±2.2 to the magnetic field. Both of these figures indicate that larger mass meteoroids have a higher tendency to create a nonspecular trail. There are a few notable exceptions, with three nonspecular trails corresponding to meteoroids with abnormally low masses (<5 10 6 grams) shown in Figure 7a. Of these three, two are contained in the 1820 data file and are shown in Figure 5c (mass = 1.9 10 6 grams) and Figure 5e (mass = 6 10 7 grams). Figure 5c again shows the lowest power, which corresponds to a low mass. The low-mass meteoroid calculated in Figure 5e is a result of fragmentation, as is evidenced in Figure 5d (note that the head echo shown in Figure 5e is a continuation of a possibly fragmented meteoroid shown in Figure 5c). This trail, in addition to the one shown in Figure 4, are perhaps only the second and third unambiguous (range resolved) fragmentation events reported [Mathews et al., 2008]. We leave an in-depth discussion of fragmentation and its effect on the determination of meteoroid mass and head/trail SNR to a future paper. The last meteoroid with a mass less than 5 10 6 gram has an extremely short duration, which causes the mass to be underestimated due to limited data for mass loss integration [Close et al., 2005]. [20] The most convincing relation, however, should be a combination of the meteoroid kinetic energy and altitude of formation which together dictate the resultant plasma concentration. Indeed, for the trails shown in Figure 5, the trail with the lowest power shown in Figure 5c was created by the meteoroid with the lowest energy of 1.1 10 8 gcm 2 /s 2 (of the six events shown). The trail with the highest power shown in Figure 5d was created by a high-energy meteoroid, with the energy exceeding 4.6 10 8 g cm 2 /s 2. Although, the trails shown in Figure 5a and Figure 5b were created by higher-energy meteoroids (9.6 10 8 gcm 2 /s 2 and 5.3 10 8 gcm 2 /s 2 ), these trails also formed at slightly higher altitudes than the one shown in Figure 5d. Therefore, a combination of meteoroid properties with altitude of formation (in addition to angle with B) helps dictate a nonspecular trail s peak power. 5. Summary [21] We have investigated the effects of radar frequency at two frequencies and aspect sensitivity on the ability to detect nonspecular trails or range-spread trail echoes. Using the 160 MHz and 422 MHz waveforms transmitted by ALTAIR, we pointed perpendicular to the magnetic field Figure 7. Plot of altitude versus meteoroid mass, calculated using the scattering theory. (a) Data includes 427 head echoes and the 41 head echoes detected with a corresponding nonspecular trail. (b) Data includes only aspect angles that are 87.8 degrees or more to the magnetic field. Both plots indicate that larger mass meteoroids tend to form nonspecular trails more often than their smaller mass counterparts. 7of8

and stepped in 1 2 degree increments from the perpendicular direction in order to determine the peak signal strength as a function of aspect angle. We have found that the peak SNR of trails drops by approximately 17 db over a 6 degree offset from perpendicular using our limited data set, which includes 41 nonspecular trails. In addition, we have found that the variability between the peak SNR over all trails detected at the same aspect angle can be primarily attributed to variations of the parent meteoroid (mass, velocity) and the altitude of formation. In addition to aspect sensitivity, we have also calculated the dependence between trail SNR and radar wavelength using the 160 and 422 MHz waveforms. We have found that the nonspecular SNR varies as l 6 using just these two frequencies, although this dependence also varies as a function of altitude. This is in contrast to the wavelength dependence of head echoes, which falls between l 2 and l 3. [22] Our future work includes examining the polarization ratios of nonspecular trails using a more extensive data set, determining diffusion coefficients for all of our trails, including specular and nonspecular, and investigating the dependence of SNR on frequency as a function of altitude. [23] Acknowledgments. The authors gratefully acknowledge the contributions from Michael Kelley, William Cooke, and Gary Bust. NASA Marshall Space Flight Center sponsored this work. [24] Amitava Bhattacharjee thanks John Mathews and another reviewer for their assistance in evaluating this paper. References Chapin, E., and K. Kudeki (1994), Plasma wave excitation on meteor trails in the equatorial electroject, Geophys. Res. Lett., 21, 2433 2437, doi:10.1029/94gl01705. Close, S., M. M. Oppenheim, S. Hunt, and L. P. Dyrud (2002), Scattering characteristics of high-resolution meteor head echoes detected at multiple frequencies, J. Geophys. Res., 107(A10), 1295, doi:10.1029/ 2002JA009253. Close, S., M. Oppenheim, S. Hunt, and A. Coster (2004), A technique for calculating meteor plasma density and meteoroid mass from radar head echo scattering, Icarus, 168, 43 52. Close, M., M. Oppenheim, D. Durand, and L. Dyrud (2005), A new method for determining meteoroid mass from head echo data, J. Geophys. Res., 110, A09308, doi:10.1029/2004ja010950. Close, S., P. Brown, M. Campbell-Brown, M. Oppenheim, and P. Colestock (2007), Meteor head echo radar data: Mass velocity selection effects, Icarus, 186, 547 556. Dyrud, L. P., M. M. Oppenheim, and A. F. vom Endt (2001), The anomalous diffusion of meteor trails, Geophys. Res. Lett., 28, 2775 2779, doi:10.1029/2000gl012749. Eshelman, R. (1954), Theory of radio reflections from electron-ion clouds, IRE Trans. Antennas Propag., 3, 32 pp. Haldoupis, C., and K. Schlegel (1993), A 50-MHz radio Doppler experiment for midlatitude E region coherent backscatter studies: System description and first results, Radio Sci., 28, 949 978. Heritage, J. L., W. J. Fay, and E. D. Bowen (1962), Evidence that meteor trails produce fields aligned scatter signals at VHF, J. Geophys. Res., 67, 953 959, doi:10.1029/jz067i003p00953. Janches, D., C. J. Heinselman, J. L. Chau, A. Chandran, and R. Woodman (2006), Modeling the global micrometeor input function in the upper atmosphere observed by high power and large aperture radars, J. Geophys. Res., 111, A07317, doi:10.1029/2006ja011628. Mathews, J. D., S. J. Briczinski, D. D. Meisel, and C. J. Heinselman (2008), Radio and meteor science outcomes from comparisons of meteor radar observations at AMISR Poker Flat, Sondrestrom, and Arecibo, Earth Moon Planets, 102, 365 372, doi:10.1007/s11038-007-9168-0. Mathews, J. J. (2004), Radio science issues surrounding HF/VHF/UHF radar meteor studies, J. Atmos. Sol. Terr. Phys., 66, 285 299, doi:10. 1016/j.jastp.2003.11.001. Oppenheim, M. M., A. F. Vom Endt, and L. P. Dyrud (2000), Electrodynamics of meteor trail evolution in the equatorial E region ionosphere, Geophys. Res. Lett., 27, 3173 3176, doi:10.1029/1999gl000013. Reddi, C. R., and S. M. Nair (1998), Meteor trail induced backscatter in MST radar echoes, Geophys. Res. Lett., 25, 473 476, doi:10.1029/ 98GL00139. Reddi, C., T. V. C. Sarma, and P. B. Rao (2002), Spatial doman interferometric VHF radar observations of spread meteor echoes, J. Atmos. Sol. Terr. Phys., 64, 339 347, doi:10.1016/s1364-6826(01)00107-9. Sugar, G. R. (1963), Radio propagation by reflection from meteor trails, Proc. IEEE Int. Soc. Opt. Eng., 52, 116 136. Zhou, Q. H., J. D. Mathews, and T. Nakamura (2001), Implications of meteor observations by the MU radar, Geophys. Res. Lett., 28, 1399 1402, doi:10.1029/2000gl012504. Zhou, Q. H., Y. T. Morton, J. D. Mathews, and D. Janches (2004), Aspect sensitivity of VHF echoes from field aligned irregularities in meteor trails and thin ionization layers, Atmos. Chem. Phys. Discuss., 4, 731 751. S. Close, P. Colestock, L. Cox, and T. Hamlin, Los Alamos National Laboratory, Los Alamos, NM 87545, USA. (sc_mars@yahoo.com) M. Oppenheim, Center for Space Physics, Boston University, Boston, MA 02215, USA. 8of8