Research opportunities at the geospace facilities exemplified by a large meteor echo

A small 30-MHz coherent scatter radar imager has been deployed to the HAARP facility near Gakona, Alaska, for studying the effects of ionospheric modifications on polar mesospheric summer echoes (PMSEs). In initial tests scheduled during the 2025 Perseids shower, the radar observed a remarkably long-lived non-specular meteor trail. Although not optimized for the purpose, the radar mode used was sufficiently general to permit analysis of the echoes. This paper explores the results of the analysis and highlights the adjacent science that can be performed in the backdrop of multi-purpose geospace research facilities like HAARP. Lessons learned from the exercise and perspectives on the role and value of the geospace facilities are discussed.

1 Introduction

The HAARP ionospheric modification facility is the newest and largest instrument purpose built for studying the effects of high-power radio waves on the ionosphere. Those effects include Ohmic heating, very low frequency (VLF) wave generation, airglow excitation, stimulated electromagnetic emission (SEE), field-aligned irregularity generation, artificial periodic inhomogeneity generation, and strong Langmuir turbulence creation leading to electron acceleration and secondary ionization. The literature on ionospheric modification experiments is extensive (e.g., Duncan and Gordon, 1982; Robinson, 1989; Stubbe, 1996; Gurevich, 2007; National Research Council, 2014).

Ionospheric modification effects can be subtle, and sensitive instrumentation is required for measurements. The most important diagnostics include airglow imagers and spectrometers, VLF receivers, SEE receivers, ionosondes and other HF transceivers, and radars up through the scale of incoherent scatter radars (ISRs). Major geospace facilities like ISRs and ionospheric modification instruments serve as magnets for other instruments which tend to cluster around them for scientific and logistical reasons. Clustering plays an important role in experimental research as the synergy between instruments increases the value of all of them in the aggregate.

Ionospheric modification experiments are logistically demanding and usually conducted in brief campaigns a few times per year. ISRs are more flexible but still conduct campaigns, for example, during Coordinated World Day runs. The diagnostic instruments need not be idle outside the campaigns, however. Nor need their observing modes be designed purely for supporting the major instruments at the center of the cluster. Ideally, the instrument clusters and the experimental modes they run should be designed to provide comprehensive measurements and to capture rare but revealing scientific phenomena when they occur.

An example of a rare event involving unusually strong, long-lived non-specular meteor trail echoes is described below. The echoes were observed during initial testing of a small radar being deployed at HAARP for studying polar mesospheric summer echoes (PMSEs). The echoes occurred when HAARP was running. The meteor trail was remarkably good fortune, and it was possible to develop methods for generating high-resolution, three-dimensional images of both the meteor head and trail echoes from the radar observations. However, neither the HAARP experiment, the PMSE radar, nor the local instrument cluster were optimized to take full advantage of it. The purpose of this paper is to encourage the designs of experiments and instrument clusters for rare, serendipitous occurrences which might include irregularities due to neutral and plasma instabilities and sudden responses to space-weather drivers from above and below in addition to meteors.

1.1 Background

The electron number density becomes significant at mesospheric heights where the electrons exhibit fluctuations driven by the neutral atmospheric turbulence. The D-region ionosphere involves complex chemistry with positive and negative ions and exotic species and may be characterized as a dusty plasma (e.g., Plane et al., 2015; Scales and Mahmoudian, 2016). The deposition of free neutral metals in the mesosphere by meteor ablation (see below) contributes to the chemistry. In the polar summer mesosphere, extremely low mesopause temperatures cause ice crystal formation leading to polar mesospheric clouds, and the electron density gradients in the vicinity of the charged ice grains can be significant. Heavy, multiply-charged ions in the D region can greatly retard ion diffusion, leading to long-lived irregularities and fluctuations (Rapp and Lübken, 2004; Varney et al., 2011).

A byproduct are PMSEs which occur between about 80 and 90 km altitude. These arise from mesospheric turbulence and associated fluctuations in electron density. The echoes can be much stronger than those detected routinely by mesosphere-stratosphere-troposphere (MST) radars in the polar summer mesosphere when the scattering cross section is enhanced by electron density gradients associated with ice grains which also arrest diffusion in the plasma at small scales. The scattering cross section increases with increasing radar wavelength, making HF and VHF systems the most sensitive. PMSE echoes are sensitive tracers of vertical velocity and are also diagnostic of ion composition. Furthermore, PMSEs can be modulated by ionospheric modification by powerful HF waves (Havnes et al., 2003). Ionospheric modification of PMSE is an incisive tool for quantifying complex charging processes in dusty plasmas (Scales and Mahmoudian, 2016).

Meteor echoes, meanwhile are frequent and ubiquitous. Meteor echoes occur in three varieties (Kero et al., 2019). Head echoes arise from the plasma sheath that forms ahead of the incoming meteoroid. The mechanism is essentially Rayleigh scatter. Head echoes are the most difficult to detect and have mainly been studied using high-power, large-aperture radars (e.g., Chau et al., 2007; Hedges et al., 2024). The ionization trails left behind the meteoroids, meanwhile, produce Fresnel scatter which becomes highly visible to radars when the trail is aligned in such a way that the condition for specular reflection is met. Specular meteor echoes can be observed by small radar systems and provide information about the neutral wind speed and direction together with other parameters which can be inferred from their decay rate (e.g., Hocking et al., 2001; Holdsworth et al., 2004). Finally, non-specular or range-spread echoes are the third variant. In the case of non-specular echoes observed with scattering wavevectors perpendicular to the magnetic field, the scattering mechanism is Bragg scatter from field-aligned plasma density irregularities. The irregularities are associated with instabilities due to conductivity gradients and strong polarization electric fields along the trail (Oppenheim and Dimant, 2015). These are the most common form of non-specular echo, and they are highly magnetic aspect sensitive.

Long-lived non-specular meteor trails are rare but have been observed in a number of occasions with low- and high-power radar systems. Chau et al. (2021) studied a bolide event, tracking the position of the meteoroid responsible with a network of meteor radars. Vierinen et al. (2022) reviewed multiple datasets documenting the Pajala fireball event and the associated plasma and neutral dynamics.

The mechanism responsible for non-specular non-field-aligned meteor echoes is not well understood. Close et al. (2011) observed cross-polarization in head and trail echoes from a large meteoroid. They attributed their trail observations to multiple scattering from steep gradients at the edges of a high Schmidt-number dusty plasma (see also Kelley, 2004). Chau et al. (2014b) proposed that the echoes could be due to a mechanism similar to the one responsible for PMSEs only with the role of charged ice crystals being played by charged meteoric material. In either case, the slow multipolar diffusion mechanism described by Varney et al. (2011) could explain the long echo lifetimes. If the hypothesis is correct, the non-specular echoes should exhibit hysteresis similar to PMSEs when heated.

2 Methods

We present data from a small 30-MHz coherent scatter radar interferometer deployed in Gakona, Alaska, at the HAARP ionospheric modification facility (62.2378$°$N, 145.1014$°$W, 590 masl). The radar incorporates a transmitter emitting 12 kW of peak power with a 7.5% maximum duty cycle, 200 $\mu$s maximum pulse length, and 100 kHz bandwidth. The intended application of the radar is PMSEs. Its nominal operating mode for PMSE employs 16-bit complementary coded pulses with a 400 Hz pulse repetition frequency. A typical PMSE experiment might involve coherently summing the returns from 50 pulses, subsequently forming 16-point spectra, and incoherently integrating 4 spectra for an overall experimental cadence of once every 8 s. This affords a considerable signal-to-noise ratio improvement over a faster experiment while being rapid enough to capture changes in the PMSE scattering cross section arising from ionospheric modification.

The radar uses software-defined radios (SDR) for transmission and reception. The antenna array combines 16 three-element vertically-pointing Yagi antennas into $\mathit{\text{N}}$ = six groups or modules. Two of the modules are used for transmission, and all six for reception. Figure 1 shows the layout of the array in plan view. The modules are situated such that the $N(N-1)/2$ = 15 interferometry baselines available are non-redundant. Individual Yagi antennas are turned 45$°$ with respect to the array axes to reduce coupling. The axes of the array are rotated 20$°$ clockwise with respect to geographic coordinates, closely matching the alignment of the Ionospheric Research Instrument (IRI) array at HAARP.

Figure 1

Figure 1. 30-MHz imaging radar antenna array configuration (plan view). Six distinct antenna groups (modules) are distinguished by different colors. The module spacings are multiples of 1.25$\lambda$ in either cardinal direction. Modules (a) and (b) are used for transmission whereas all modules are used for reception. The array axes are rotated 20$°$ clockwise from the cardinal geographic coordinates at the radar site.

In PMSE applications, the echoes are truly three dimensional (e.g., Sommer and Chau, 2016; Urco et al., 2019; Chau et al., 2020), and the antenna design favors a vertical pencil beam and antenna modules presenting a diversity of baselines in the horizontal plane. This configuration implies shorter interferometry baselines than would be used for studying quasi-2D echoes from field-aligned irregularities, for example,. In the case of the array in Figure 1, the longest baseline is only about 4.5$\lambda$.

Early tests of the VHF radar in mid-August 2025, yielded no PMSE echoes as the season for strong PMSEs had already passed. However, a very long-lived non-specular meteor echo was observed within the first hours of testing. This is depicted in Figure 2. These data were acquired in the nominal PMSE transmission mode. Processing was performed using eight coherent integrations, eight spectral bins, and four incoherent integrations for an overall cadence of one profile every 1.56 s.

Figure 2

Figure 2. Range time intensity (RTI) plot of a long-lived, non-specular meteor trail observed on 12 August 2025, starting at 23:49:26 UT. The brightness, hue, and saturation of the pixels represent signal-to-noise ratio, Doppler shift, and spectral width according to the legends shown. The color bar indicates HAARP heating and probing intervals in yellew and green, respectively.

The non-specular meteor echo spanned altitudes between 70 and 110 km altitude and persisted for more than 5 min. The signal-to-noise ratio was as great as 50 dB at times. Doppler shifts varied strongly with altitude and were as large as 50 m/s. In the figure, the trail apparently emerged first at the highest altitudes, then at the lowest, and only later at intermediate altitudes.

Echoes from the times preceding the emergence of the meteor trail are depicted in Figure 3. In this case, there was no coherent integration, and eight spectral bins were incoherently integrated twice for an overall experimental cadence of 25 profiles per second. Remarkably, the small radar system captured not only the non-specular meteor trail but the meteor head echo. The top panel in the figure depicts the results of conventional pulse decoding. As the complementary nature of the coded pulse is not being exploited, we may expect relatively high range sidelobe levels for this mode. The panel shows a thin trail emerging almost immediately at about 110 km altitude. Note that a strong, unrelated specular meteor echo occurred at 120-km range at precisely 23:49:27.

Figure 3

Figure 3. RTI depiction of meteor head echo plotted following the conventions of Figure 2. (top) Results using conventional pulse decoding. (bottom) Results using decoding with a progressive phase shift compensating for meteor head line-of-sight velocity.

Because of its high velocity, the meteor head echo requires additional processing for proper pulse decompression (Chau and Woodman, 2004; Kero et al., 2012). The bottom panel of Figure 3 shows the results of decoding the signals with the introduction of a progressive phase shift corresponding to a line-of-sight velocity of 40 km/s. This processing reduces the range sidelobes substantially. The resulting meteor head echo can be seen to span about 40 km in range over a duration of about 1 s. The range span of the head echo matches that of the trail that formed later.

Spectra from the meteor head echo were found to be broad and inconsistent from range to range and time to time. A pulse-to-pulse correlation analysis found that the inter-pulse coherence (modulus of complex correlation) was small, about 0.2–0.3, preventing precise Doppler shift determination (e.g., Woodman and Hagfors, 1969). This is unfortunate, as information derived from the range rate of the echo together with its (frequency aliased) Doppler shift could be used to estimate its precise speed and deceleration rate (Schult et al., 2017). The head-echo signal was evidently not statistically stationary in this experiment, precluding more detailed time series analysis.

2.1 Phase calibration

Multiple receive antennas were deployed for aperture-synthesis radar imaging which produces true imagery of the radar backscatter in the directions normal to the radar line of sight. Imaging and other spaced-antenna methods require the phases of the signals from the different antennas to be calibrated regularly as the receive systems tend to introduce spurious phase offsets that can vary over time. Phase calibration with the Gakona VHF radar has been performed in three steps as explained below.

2.1.1 Radio source calibration

Signals from radio stars can be used for phased-array calibration if they are made sufficiently narrow-band by filtering (Palmer et al., 1996). At high latitudes, calibration can be performed using emissions from Cas-A which is not a star but a supernova remnant (Chau et al., 2014a). While it is a bright signal source which appears nearly overhead of the Gakona radar daily, it nonetheless presents a weak signal to the compact antenna modules used.

Calibration data were recorded after-the-fact on Oct. 17, 2025, with the radar transmitter deactivated. The complex cross-correlations (covariances) of signals representing all 15 antenna baselines acquired at 10 $\mu$s intervals were computed on the basis of 4 s averages. The corresponding data are shown in Figure 4 along with their astronomical predictions. For clarity, only the five baselines incorporating reference module (a) are shown in the figure.

Figure 4

Figure 4. Radio star calibration data for the Gakona radar on Oct. 17, 2025. Five colors represent results for five module baselines (cyan: a-b, green: a-c, red: a-d, violet: a-e blue: a-f). From top to bottom, the panels show coherence, phase angle, and predicted phase angle for Cas-A. Spurs at 7.5 min intervals are interference from the nearby ionosonde.

The phases of signals from antennas (b) through (f) have been adjusted for optimal agreement between the measured and predicted phase angles for Cas-A. The optimization used unweighted data but incorporated data only from times after 7 UT when signal quality was best.

The calibration data are clearly noisy, a consequence of the limited signal-to-noise ratios for Cas-A emissions. The noisiness is reflected in the rather low coherences of the interferometric covariances which are less than 0.2. Covariance estimates with the highest coherences are the most accurate (e.g., Farley, 1969). Interference is present in the signals from all the modules as well, especially module (f) which is situated near two electronics shelters. Other radio sources in the sky also contribute to the interference. The radiation patterns of the various modules are too broad to exclude celestial and anthropogenic sources that are not the intended one.

2.1.2 Internal consistency calibration

The precision of the aforementioned phase calibrations can be improved by using strong radar echoes as sources and enforcing internal consistency. We regard the echoes from the non-specular trail as arising from a point source in each range gate early in the trail’s evolution. This is confirmed by the measured coherences of the echoes which are nearly unity in every Doppler bin, range gate, and time interval where the earliest echoes were strong.

We process the echo data using two coherent integrations, eight spectral bins, and four incoherent integrations as described above. The measured phase angle of the covariance term for baseline $ij$ can be expressed as:

${\varphi }_{i,j}=Q_{ij}\cos \eta +R_{ij}\cos \zeta +{\mathrm{\Delta }}_i-{\mathrm{\Delta }}_j(1)$

where $k$ is the wavenumber, $Q_{ij}\equiv k(d_{xj}-d_{xi})$, $R_{ij}\equiv k(d_{yj}-d_{yi})$, and ${\mathrm{\Delta }}_{i,j}$ are the phase angle biases for modules $i$ and $j$, respectively, introduced by the hardware. In Equation 1, $\cos \eta$ and $\cos \zeta$ are the direction cosines of the signals measured with respect to the horizontal $(x)$ and vertical $(y)$ axes in Figure 1. The terms $Q$ and $R$ contain the horizontal and vertical baseline lengths $d$ for baseline $ij$.

The goal is to estimate the various phase biases $\mathrm{\Delta }$ on the basis of the measured data. We take ${\mathrm{\Delta }}_a$ to be zero without loss of generality. ${\mathrm{\Delta }}_b$ and ${\mathrm{\Delta }}_d$ are taken from the radio star calibration outlined above. That leaves the constant offsets ${\mathrm{\Delta }}_c$, ${\mathrm{\Delta }}_e$, and ${\mathrm{\Delta }}_f$ to be determined together with the direction cosines which vary with range.

For every range gate and Doppler bin, we define two objective functions for each interferometry baseline:

${ϵ}_{I;i,j}=\cos \left(Q_{ij}\cos \eta +R_{ij}\cos \zeta +{\mathrm{\Delta }}_i-{\mathrm{\Delta }}_j\right)-\cos {\varphi }_{i,j}(2)$

${ϵ}_{Q;i,j}=\sin \left(Q_{ij}\cos \eta +R_{ij}\cos \zeta +{\mathrm{\Delta }}_i-{\mathrm{\Delta }}_j\right)-\sin {\varphi }_{i,j}(3)$

In Equations 2, 3, the cosine and sine variants of the objective functions are introduced to account for aliasing of the periodic phases. The solution vector is found by minimizing the sum of the squares of all the $N(N-1)$ objective functions for $i\in [1,N]$, $j\in [i+1,N]$ where $N$ is the number of sensors. Absent the aliasing issue and the requirement for the sine and cosine functions, the problem could be posed and solved linearly. In this form, the problem is solved using a nonlinear least squares solver, e.g., Levenberg Marquardt.

Finally, when averaging the ${\mathrm{\Delta }}_i$ estimators thus obtained over multiple range and Doppler bins for improved statistical precision, it is necessary to average their sines and cosines separately and to define the operational phase angle offset estimators from the arctangent (i.e., $⟨{\mathrm{\Delta }}_i⟩\equiv \mathrm{a}\mathrm{r}\mathrm{c}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{2}(⟨\sin {\mathrm{\Delta }}_i⟩,⟨\cos {\mathrm{\Delta }}_i⟩)$. This too avoids averaging problems associated with angle aliasing. The resulting phase offset estimators have been found to be consistent, precise, and robust. The direction cosine estimators available for each range gate are not required for calibration but are useful byproducts.

2.1.3 Global calibration

The internal calibration improves the precision of the phase offset estimates obtained from the radio source calibration but does not improve the accuracy. For that, a final, global method is used. This involves considering the totality of non-specular meteor trail echoes collected over a long time interval–several hours in practice. The non-specular meteor echoes in ranges less than 100-km altitude, for example, are expected to arrive from small zenith angles preferentially. The phase angle offsets for all the receivers relative to the reference receiver (a) can therefore be adjusted to “steer” the aggregate observations toward zenith. (Steering here implies adding a progressive shift to the $\mathrm{\Delta }$ estimators proportional to the displacement of the given antenna in both cardinal directions.) Given enough observations, the histogram of non-specular meteor detections vs. the direction cosines should resemble the two-way antenna radar radiation pattern.

Figure 5 depicts all the non-specular meteor echo detections over 3 hours of measurements spanning Aug. 12, 13, and 14. Non-specular meteor echoes are defined as exceeding a minimum signal-to-noise ratio threshold over three or more contiguous range bins. The detections are binned and accumulated in direction cosine space. Note that a single meteor echo can give rise to multiple counts in this figure if it occupies multiple range, Doppler, and time bins. The total number of detections represented by Figure 5 is approximately a half million.

Figure 5

Figure 5. Bearings of non-specular meteor echoes observed over 3 hours of observations spanning Aug. 12, 13, and 14, 2025. Grayscales represent the relative number of detections binned by the two direction cosines $\cos \eta$ and $\cos \zeta$. Red and blue curves denote the radiation patterns of the Gakona VHF radar and the nominal HAARP IRI at 5.2 MHz, respectively. Contours at the −3 dB, −6 dB, −10 dB, and −20 dB levels are plotted.

Since the shortest baselines in the array depicted in Figure 1 are 1.25$\lambda$, only direction cosines between $\pm$0.4 can be unambiguously determined without additional knowledge of the target. The visible region thus defined is indicated by the square in the figure. As red contours, Figure 5 shows the two-way radiation pattern of the Gakona radar assuming transmission on modules (a) and (b) together and reception on either (a) or (b). The figure shows that the radiation pattern is down more than 20 dB where $\cos \eta =\pm$0.4 but only by about 7 db where $\cos \zeta =\pm$0.4. Echo angle aliasing in the $\cos \zeta$ axis therefore remains a possibility and a potential source of ambiguity.

Starting with the phase angle offsets found through the preceding two procedures, the final optimization procedure involves refining them by steering the meteor echoes in Figure 5 to minimize the number of detections occurring outside a −7 dB contour. The procedure accounts for angle aliasing and reintroduces detections that vanish from one side of the visible region to the other side. The procedure is iterative and converges after a few tens of iterations.

3 Results

The covariances ordered by the spatial separation of the antennas $(k{d}_x,k{d}_y)$ are known as the visibilities, and the received power sorted by the direction cosines $(\cos \eta ,\cos \zeta )$ is known as the brightness distribution. Under certain restrictions, according to the van Cittert-Zernike theorem, these quantities are related simply by Fourier transform (Thompson, 1986). The restrictions include the requirements that the antenna modules be coplanar, the range of included direction cosines be sufficiently narrow, and the signals arising from different locations in space be statistically independent.

3.1 Aperture synthesis imaging

The correspondence principle defined above permits the construction of true images of the radar backscatter from visibility measurements in the time or frequency domain. Due to the sparseness of the measurements, however, it is generally impractical to transform from one domain to the other directly. Instead, aperture synthesis imaging usually involves solving an optimization problem. The first methods for imaging star fields used a greedy method to deconvolve the point spread function from the measured visibilities (e.g., Clark, 1980). This works well for point targets but not for continua. In the STAP (space-time adaptive processing) paradigm widely used in operational radar applications, the weights of the signals from each module are adjusted to minimize the signal power in each steering vector while simultaneously maintaining fixed gain (e.g., Capon, 1969). This is a special case of a linearly-constrained minimum variance (LCMV) or minimum-variance distortionless recovery (MVDR) method. A more common method in radio astronomy involves finding brightness distributions consistent with the measured visibilities within a specified confidence limit while maximizing the entropy of the distribution (MaxENT, e.g., Skilling and Bryan, 1984; Wilczek and Drapatz, 1985; Hysell and Chau, 2006). This method incorporates confidence estimates for the visibility measurements, something normally neglected with the other methods, and is the method followed here because it is effective in low- and high-SNR regimes (see below). Another method in compressive sensing minimizes the $L_1$ norm of the brightness distribution instead of the entropy (e.g., Harding and Milla, 2013; Hysell et al., 2019).

The MaxENT algorithm is a super-resolution method (surpassing the diffraction limit) that has by one measure the finest spatial resolution possible (Kosarev, 1990). Relating the imaging problem to the Shannon Hartley channel capacity theorem, Kosarev showed that the improvement in angular resolution of the MaxENT over the diffraction limit (what would be recovered using a Fourier transform directly for inversion) is proportional to the base-2 logarithm of 1 + SNR. Unlike many estimators in signal processing where performance improvements are negligible beyond SNR$\sim$10, the performance of the MaxENT algorithm improves indefinitely with the signal-to-noise ratio. This makes the method highly suitable for studying the meteor trail echo.

Images of the radar backscatter are generated as functions of range, time, Doppler shift, and bearing. Rendering the imagery invites compromises. Here, information about the signal-to-noise ratio, Doppler shift, and spectral width are conveyed through pixel brightness, hue, and saturation as in Figure 3. Time information is conveyed through snapshots. The images themselves are presented in two dimensional cuts through constant $\cos \eta$ and constant $\cos \zeta$ surfaces. These surfaces are cones with axes of rotation aligned with the $x$ and $y$ axes depicted in Figure 1, respectively.

Figure 6 depicts images of the head echo at timesteps spaced by 120 ms. The panels on the left and right of the figure represent constant $\cos \zeta$ and $\cos \eta$ surfaces, respectively. It turns out that all the meteor echoes fell very nearly on a constant $\cos \eta$ surface, and so the panels on the right represent the echoes fairly completely. The echoes only sometimes intersect the panels on the left, meanwhile, which are more sparsely populated. Nonetheless, the method gives a clear picture of the progression of the meteor head echo. Together, the images depict a compact target descending rapidly while moving south-southwestward.

Figure 6

Figure 6. Cuts through volumetric images of the meteor head echo at different times (top to bottom) staggered by 120 ms. Panels on the left represent spans of $\cos \eta$ space for constant $\cos \zeta =0.2$. Panels on the right represent spans in $\cos \zeta$ space for constant $\cos \eta =0.1$. The brightness, hue, and saturation of the image pixels represent signal-to-noise ration, Doppler shift, and spectral width, respectively, following the legends in Figure 3. The panel axes are range in km and angle in degrees.

Figure 7 shows the evolution of the meteor trail that appeared after the head echo. The color conventions are the same as in Figure 2. The trail is precisely collocated physically with the path of the head. The imagery shows that trail echoes appeared almost immediately at all ranges between about 75 and 110 km. After a few seconds, however, echoes at the extremes of the trails dominated. Only after a few more seconds did echoes from the middle of the trail intensify, as shown in Figure 2. The Doppler shifts at different ranges vary significantly, indicating strong shear in the meridional lower thermospheric winds mainly. Over several minutes, different parts of the trail at different altitudes can be seen to drift in different directions.

Figure 7

Figure 7. Cuts through volumetric images of the meteor trail echo at different times (top to bottom) staggered by 3 s. Plotting conventions are the same as in Figure 6. The brightness, hue, and saturation of the image pixels represent signal-to-noise ration, Doppler shift, and spectral width, respectively, following the legends in Figure 2.

3.2 Ionospheric modification

On Aug. 12, HAARP was being used to characterize hysteresis in PMSE should it occur. In previous experiments using the EISCAT heating facility, it has been shown that the echoes can both diminish and grow in amplitude when heating is initiated and terminated, respectively (e.g., Gunnarsdottir et al., 2023 and references therein). Hysteresis has been interpreted in terms of charging models for the dusty plasma formed in the vicinity of ice crystals in the polar summer mesosphere (Rapp and Lübken, 2000; Havnes, 2004; Scales and Mahmoudian, 2016). The experimental findings are broadly consistent with theory but highly and enigmatically variable. Experiments at HAARP were begun to expand the database and work towards improved closure with theory.

The radiation pattern of the HAARP pencil beam used is depicted by the blue contours in Figure 5. The experiment ran from 2330 to 0015 UT in cycles. For the first 24 s of each cycle, CW X-mode emissions at 5.2 MHz were transmitted. For the next 16 s, pulses with 10 $\mu$ pulse widths and a 20 ms interpulse period (IPP) were transmitted. The heater was then turned off for 80 s, completing the 2-min cycle. The pulsing was an implementation of an HF radar mode incorporating a dedicated HF receiver connected to two orthogonal loop antennas. The HF receiver experienced interference from the nearby VHF radar, so that part of the experiment was unsuccessful.

Figure 5 shows that the non-specular meteor trail fell well outside the HAARP main beam. According to the model described by La Rosa and Hysell (2025), the electron temperature increase in the vicinity of the meteor trail induced by the HAARP emissions would have been on the order of about 25 K peaking at 75-km altitude. We find no clear evidence of any corresponding effect in Figure 2 during or following the three heating cycles that occurred during the duration of the trail.

4 Discussion

The non-specular meteor echo provided a rare opportunity to test the effects of ionospheric modifications on what could be another manifestation of dusty plasma physics in the mesosphere and lower thermosphere. While the cause of non-specular, non-field-aligned, long-lived meteor trails is unknown, they can be expected to be subject to the same kind of charging and transport mechanisms at work within PMSEs. The opportunity to test this hypothesis was not fully exploited, however, because of a number of deficiencies in the experiment which was designed narrowly for PMSE. Capturing rare, episodic phenomena epitomized by this one in ionospheric modification experiments will require a different experimental approach going forward. Some lessons learned from the event are the following:

1. A high pulse repetition frequency (PRF) should be prioritized over other experimental considerations in order to recover usable stationary meteor head echo spectra. In the present experiment, the PRF could have been doubled to 800 Hz if combined with a 13-bit Barker coded pulse with a 1-km bit width. This bit width exceeds the bandwidth rating of the transmitter but appears to be workable in practice.

2. Pursuant to the previous item, broader bandwidth and improved range resolution will be important for studying narrowly-layered phenomena like PMSE and meteor echoes going forward. The nominal 100 kHz bandwidth seems inadequate in view of the data collected. (The 1-km bit width target implies a bandwidth of 150 kHz)

3. The radiation pattern of the HAARP IRI should be broadened/spoiled when the VHF radar is supporting experiments. The twisted beam configuration in particular would fill the region illuminated by the VHF radar beam (Bernhardt et al., 2016; Hysell et al., 2018). This would both increase the detectable impacts of HF modifications in the VHF radar data and increase the likelihood of capturing meteor and other episodic events in the modified volume.

4. The VHF radar array should be enlarged so that the radiation pattern falls entirely within the visible region. Interference sources that impede radio-source calibration should be identified and mitigated.

5. Additional filtering is required to minimize interference between the VHF system, the HF receiving system, and the ionosonde at HAARP.

The meteor event also highlights systemic improvements that could be made to any of the geospace facilities to promote instrumental synergy and facilitate studies of rare events that do not occur to schedule. Firstly, the addition of a modest radar system to HAARP exposed an otherwise invisible class of naturally occurring phenomena to study, broadening the facility’s disciplinary reach. The addition was straightforward since the required infrastructure existed already. Ionospheric modifications are not strictly necessary for pursuing mesospheric research with the radar which can be run remotely outside of campaign times. Moreover, heating experiments designed for other applications would nonetheless affect the mesospheric observations in an informative way in many cases. In that sense, heating time for PMSE studies is effectively available at no cost so long as heating experiments are coordinated and made somewhat general. Campaign time is scarce, but cross-cutting scientific objectives and a well designed suite of diagnostics can increase the value of a campaign multiplicatively.

Secondly, some additional, inexpensive instrumentation would materially enhance the capabilities of the facility further both during and outside campaigns. A simple commercial webcam, for example, would permit comparisons between optical and radio meteor detections as well as meteor radar afterglow, a phenomenon at the forefront of discovery research in aeronomy and space physics (e.g., Obenberger et al., 2020). More capable but still tractable optical instrumentation such as all-sky cameras would support additional spinoff science, for example, the study of the relationship between PMSEs and polar mesospheric clouds. Resonance lidars would facilitate even more incisive research into, for example, D region composition, an already complicated subject made more complicated by the increasing rate of space vehicle reentries (Murphy et al., 2023). This would be valuable research outside of campaigns, with campaigns adding incisive experimental modalities when they occur.

There are roles for additional radars at the geospace facilities as well. A multistatic meteor radar network running continuously would establish wind driven transport in the mesosphere and D region during heating experiments as well as in connection to launches from the Poker Flat research range and ISR experiments using the Poker Flat Incoherent Scatter Radar (PFISR). The Modular UHF Ionospheric Radar (MUIR) at HAARP, proven to be capable of observing ion and plasma lines during heating experiments, would be a valuable asset if restored to service. Finally, the formidable HF transmitter at HAARP has most of the components required to function as an equally formidable HF radar. The equipment shelters already possess directional couplers for antenna access as well as timing signals and networking needed for SDR receivers. An HF radar capability would expand the research scope of the system from a heater to a general-purpose laboratory for studying radio science, aeronomy, and space physics.

The remarks above apply equally well to all the geospace facilities, which span latitudes from the polar cap through the equatorial ionosphere, and their clusters of supporting instrumentation. The facilities are capable of extremely detailed, high-fidelity investigations of narrowly-targeted scientific objectives. Increasingly, the facilities are able to perform extended runs in reduced average power modes, and components of the clusters can operate continuously. Given coordinated observations involving multi-application experiments and complete, long-duration databases for retrospective studies, the geospace facilities are the most auspicious resources available for discovery research.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author contributions

DH: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing – original draft. AK: Investigation, Writing – review and editing. GV: Investigation, Writing – review and editing. JC: Investigation, Methodology, Validation, Writing – review and editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. The work was supported by NSF award AGS-2146187 and DARPA through subcontract HR00112490535 from Ohio State University. The Subauroral Geophysical Observatory (SAGO) is supported by NSF Award AGS-2054361. JC acknowledges support from the European Office of AFOSR 717 (EOARD) grant FA8655-23-1-7017.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Bernhardt, P. A., Siefring, C. L., Briczinski, S. J., McCarrick, M., and Michell, R. G. (2016). Large ionospheric disturbances produced by the HAARP HF facility. Radio Sci. 51, 1081–1093. doi:10.1002/2015rs005883

CrossRef Full Text | Google Scholar

Capon, J. (1969). High-resolution frequency-wavenumber spectrum analysis. Proc. IEEE 57, 1408–1418. doi:10.1109/proc.1969.7278

CrossRef Full Text | Google Scholar

Chau, J. L., and Woodman, R. F. (2004). Daytime vertical and zonal velocities from 150-km echoes: their relevance to F-region dynamics. Geophys. Res. Lett. 31, L17801. doi:10.1029/2004GL020800

CrossRef Full Text | Google Scholar

Chau, J. L., Woodman, R. F., and Galindo, F. R. (2007). Sporadic meteor sources as observed by the jicamarca high-power large-aperture VHF radar. Icarus 188 (1), 162–174. doi:10.1016/j.icarus.2006.11.006

CrossRef Full Text | Google Scholar

Chau, J. L., Renkwitz, T., Stober, G., and Latteck, R. (2014a). MAARSY multiple receiver phase calibration using radio sources. J. Atmos. Solar-Terrestrial Phys. 118, 55–63. doi:10.1016/j.jastp.2013.04.004

CrossRef Full Text | Google Scholar

Chau, J. L., Strelnikova, I., Schult, C., Oppenheim, M. M., Kelley, M. C., Stober, G., et al. (2014b). Nonspecular meteor trails from non-field-aligned irregularities: can they be explained by presence of charged meteor dust? Geophys. Res. Lett. 41, 3336–3343. doi:10.1002/2014GL059922

CrossRef Full Text | Google Scholar

Chau, J. L., Urco, J. M., Avsarkisov, V., Vierinen, J. P., Latteck, R., Hall, C. M., et al. (2020). Four-dimensional quantification of kelvin-helmholtz instabilities in the polar summer mesosphere using volumetric radar imaging. Geophys. Res. Lett. 47, e2019GL086081. doi:10.1029/2019GL086081

CrossRef Full Text | Google Scholar

Chau, J. L., Urco, J. M., Vierinen, J., Harding, B. J., Clahsen, M., Pfeffer, N., et al. (2021). Multistatic specular meteor radar network in Peru: system description and initial results. Earth Space Sci. 8, e2020EA001293. doi:10.1029/2020EA001293

CrossRef Full Text | Google Scholar

Clark, B. G. (1980). An efficient implementation of the algorithm ‘CLEAN. Astron, Astrophys. 89, 377–378.

Google Scholar

Close, S., Kelley, M., Vertatschitsch, L., Colestock, P., Oppenheim, M., and Yee, J. (2011). Polarization and scattering of a long-duration meteor trail. J. Geophys. Res. Space Phys. 116. doi:10.1029/2010JA015968

CrossRef Full Text | Google Scholar

Duncan, L. M., and Gordon, W. E. (1982). Ionospheric modification by high power radio waves. J. Atmos. Terr. Phys. 44 (12), 1009–1017. doi:10.1016/0021-9169(82)90013-7

CrossRef Full Text | Google Scholar

Farley, D. T. (1969). Incoherent scatter correlation function measurements. Radio Sci. 4, 935–953. doi:10.1029/rs004i010p00935

CrossRef Full Text | Google Scholar

Gunnarsdottir, T. L., Poggenpohl, A., Mann, I., Mahmoudlan, A., Dalin, P., Haeggstroem, I., et al. (2023). Modulation of polar mesospheric summer echoes (PMSEs) with high-frequency heating during low solar illumination. Ann. Geophys. 41, 93–114. doi:10.5194/angeo–41–93–2023

CrossRef Full Text | Google Scholar

Gurevich, A. V. (2007). Nonlinear effects in the ionosphere. Uspekhi Fiz. Nauk. 177 (11), 1145–1177.

Google Scholar

Harding, B. J., and Milla, M. A. (2013). Radar imaging with compressed sensing. Radio Sci. 48, 582–588. doi:10.1002/rds.20063

CrossRef Full Text | Google Scholar

Havnes, O. (2004). Polar mesospheric summer echoes (PMSE) overshoot effect due to cycling of artificial electron heating. J. Geophys. Res. 109. doi:10.1029/2003JA010159

CrossRef Full Text | Google Scholar

Havnes, O., Hoz, C. L., and Naesheim, L. L. (2003). First observations of the PMSE overshoot effect and its use for investigating the conditions in the summer mesosphere. J. Geophys. Res. 30, 2229.

Google Scholar

Hedges, T., Lee, N., and Elschot, S. (2024). Meteor head echo detection at multiple high-power large-aperture radar facilities via a convolutional neural network trained on synthetic radar data. J. Geophys. Res. 129, e2023JA032204. doi:10.1029/2023JA032204

CrossRef Full Text | Google Scholar

Hocking, W., Fuller, B., and Vandepeer, B. (2001). Real-time determination of meteor-related parameters utilizing modern digital technology. J. Atmos. Solar-Terrestrial Phys. 63, 155–169. doi:10.1016/S1364-6826(00)00138-3

CrossRef Full Text | Google Scholar

Holdsworth, D. A., Tsutsumi, M., Reid, I. M., and Nakamura, T. T.Tsuda (2004). Interferometric meteor radar phase calibration. Radio Sci. 39, RS5012. doi:10.1029/2003RS003026

CrossRef Full Text | Google Scholar

Hysell, D. L., and Chau, J. L. (2006). Optimal aperture synthesis radar imaging. Radio Sci. 41. doi:10.1029/2005RS003383RS2003

CrossRef Full Text | Google Scholar

Hysell, D. L., Munk, J., and McCarrick, M. (2018). VHF radar images of artificial field-aligned ionospheric irregularities in the subauroral E region. Radio Sci. 53, 334–343. doi:10.1002/2017RS006497

CrossRef Full Text | Google Scholar

Hysell, D. L., Sharma, P., Urco, M., and Milla, M. A. (2019). Aperture-synthesis radar imaging with compressive sensing for ionospheric research. Radio Sci. 54, 503–516. doi:10.1029/2019RS006805

CrossRef Full Text | Google Scholar

Kelley, M. C. (2004). A new explanation for long-duration meteor radar echoes: persistent charged dust trains. Radio Sci. 39. doi:10.1029/2003RS002988

CrossRef Full Text | Google Scholar

Kero, J., Szasz, C., Nakamura, T., Terasawa, T., Miyamoto, H., and Nishimura, K. (2012). A meteor head echo analysis algorithm for the lower VHF band. Ann. Geophys. 30, 639–659. doi:10.5194/angeo-30-639-2012

CrossRef Full Text | Google Scholar

Kero, J., Campbell-Brown, M. D., Stober, G., Chau, J. L., Mathews, J. D., and Pellinen-Wannberg, A. (2019). Radar observations of meteors. Cambridge University Press, 65–89.

Google Scholar

Kosarev, E. L. (1990). Shannon’s superresolution limit for signal recovery. Inverse Probl. 6, 55–76. doi:10.1088/0266-5611/6/1/007

CrossRef Full Text | Google Scholar

La Rosa, B. H., and Hysell, D. L. (2025). Modeling and analysis of artificial periodic inhomogeneities in the ionosphere. Radio Sci. 60, e2025RS008226. doi:10.1029/2025RS008226

CrossRef Full Text | Google Scholar

Murphy, D. M., Abou-Ghanem, M., Cziczo, D. J., Froyd, K. D., Jacquot, J., Lawler, M. J., et al. (2023). Metals from spacecraft reentry in stratospheric aerosol particles, Proc. Natl. Acad. Sci. U. S. A. Washington, DC: National Academies Press, 120. doi:10.1073/pnas.2313374120

CrossRef Full Text | Google Scholar

National Research Council (2014). Opportunities for high-power, high-frequency transmitters to advance ionospheric/thermospheric research: report of a workshop. Washington, DC: The National Academies Press.

Google Scholar

Obenberger, K. S., Holmes, J. M., Ard, S. G., Dowell, J., Shuman, N. S., Taylor, G. B., et al. (2020). Association between meteor radio afterglows and optical persistent trains. J. Geophys. Res. Space Phys. 125, e2020JA028053. doi:10.1029/2020JA028053

CrossRef Full Text | Google Scholar

Oppenheim, M. M., and Dimant, Y. S. (2015). First 3-D simulations of meteor plasma dynamics and turbulence. Geophys. Res. Lett. 42, 681–687. doi:10.1002/2014GL062411

CrossRef Full Text | Google Scholar

Palmer, R. D., Vangal, S., Larsen, M. F., Fukao, S., Nakamura, T., and Yamamoto, M. (1996). Phase calibration of VHF spatial interferometry radars using stellar sources. Radio Sci. 31, 147–156. doi:10.1029/95RS02319

CrossRef Full Text | Google Scholar

Plane, J. M. C., Feng, W., and Dawkins, E. C. M. (2015). The mesosphere and metals: chemistry and changes. Chem. Rev. 115, 4497–4541. doi:10.1021/cr500501m

PubMed Abstract | CrossRef Full Text | Google Scholar

Rapp, M., and Lübken, F.-J. (2000). Electron temperature control of PMSE. Geophys. Res. Lett. 27, 3285–3288. doi:10.1029/2000gl011922

CrossRef Full Text | Google Scholar

Rapp, M., and Lübken, F.-J. (2004). Polar mesosphere summer echoes (PMSE): review of observations and current understanding. Atmos. Chem. Phys. 4, 2601–2633. doi:10.5194/acp-4-2601-2004

CrossRef Full Text | Google Scholar

Robinson, T. R. (1989). The heating of the high latitude ionosphere by high power radio waves. Phys. Rep. 179 (2), 79–209. doi:10.1016/0370-1573(89)90005-7

CrossRef Full Text | Google Scholar

Scales, W. A., and Mahmoudian, A. (2016). Charged dust phenomena in the near-Earth space environment. Rep. Prog. Phys. 79, 106802. doi:10.1088/0034–4885/79/10/106802

PubMed Abstract | CrossRef Full Text | Google Scholar

Schult, C., Stober, G., Janches, D., and Chau, J. L. (2017). Results of the first continuous meteor head echo survey at polar latitudes. Icarus 297, 1–13. doi:10.1016/j.icarus.2017.06.019

CrossRef Full Text | Google Scholar

Skilling, J., and Bryan, R. K. (1984). Maximum entropy image reconstruction: general algorithm. Mon. Not. R. Astron. Soc. 211, 111–124. doi:10.1093/mnras/211.1.111

CrossRef Full Text | Google Scholar

Sommer, S., and Chau, J. L. (2016). Patches of polar mesospheric summer echoes characterized from radar imaging observations with MAARSY. Ann. Geophys. 34, 1231–1241. doi:10.5194/angeo-34-1231-2016

CrossRef Full Text | Google Scholar

Stubbe, P. (1996). Review of ionospheric modification experiments at tromsø. J. Atmos. Terr. Phys. 59, 349–368.

CrossRef Full Text | Google Scholar

Thompson, A. R. (1986). Interferometry and synthesis in radio astronomy. New York: John Wiley.

Google Scholar

Urco, J. M., Chau, J. L., Weber, T., and Latteck, R. (2019). Enhancing the spatiotemporal features of polar mesosphere summer echoes using coherent MIMO and radar imaging at MAARSY. Atmos. Meas. Tech. 12, 955–969. doi:10.5194/amt-12-955-2019

CrossRef Full Text | Google Scholar

Varney, R. H., Kelley, M. C., Nicolls, M. J., Heinselman, C. J., and Collins, R. L. (2011). The electron density dependence of polar mesospheric summer echoes. J. Atmos. Sol. Terr. Phys. 73, 2153–2165. doi:10.1016/j.jastp.2010.07.020

CrossRef Full Text | Google Scholar

Vierinen, J., Aslaksen, T., Chau, J. L., Gritsevich, M., Gustavsson, B., Kastinen, D., et al. (2022). Multi-instrument observations of the pajala fireball: origin, characteristics, and atmospheric implications. Front. Astronomy Space Sci. 9, 1–18. doi:10.3389/fspas.2022.1027750

CrossRef Full Text | Google Scholar

Wilczek, R., and Drapatz, S. (1985). A high accuracy algorithm for maximum entropy image restoration in the case of small data sets. Astron. Astrophys. 142 (9).

Google Scholar

Woodman, R. F., and Hagfors, T. (1969). Methods for the measurement of vertical ionospheric motions near the magnetic equator by incoherent scattering. J. Geophys. Res. 74, 1205–1212. doi:10.1029/ja074i005p01205

CrossRef Full Text | Google Scholar