Introduction

Infrasonic wave episodes of long-duration high coherency wave trains with very high trace-velocities have been observed, in the passband from 0.015 to 0.10 Hz, over the past 35 years at many different high latitude infrasonic arrays in Alaska, Canada, Sweden and Antarctica. These high trace-velocity infrasound episodes are often directly associated with periods of geomagnetic and auroral activity. They have recently been observed throughout the year at the IMS/CTBT infrasonic arrays at I53US in Fairbanks and I55US in Antarctica. They have a maximum frequency of occurrence around the equinoxes.

At infrasonic arrays in Alaska, in Inuvik NWT, Canada, and at Kiruna in Sweden many impulsive large amplitude auroral infrasound signals were found to be related to specific auroral arcs in the auroral displays overhead. These signals were identified as bow-waves that are generated by the supersonic motion of auroral arcs that contain strong electrojet currents, (see Wilson, C.R., Auroral Infrasonic Waves, J. Geophys. Res, Vol 74, 1812-1836, 1969). These auroral infrasonic bow-waves were named AIW. The AIW infrasound is highly anisotropic propagating as a bow-wave moving in the same direction as that of the auroral arc motion. The AIW trace velocity across the microphone array is the same as lateral motion velocity of the supersonic auroral arc. Because of the anisotropic nature of the propagation of infrasonic bow-waves, it is not possible to triangulate on the auroral AIW source region by the use of data from two highly separated infrasonic arrays sites where the same AIW signals are observed.

In 2002 when I53US was established in Fairbanks, a new and different type of high trace-velocity auroral infrasound signal episode was discovered. These new high trace-velocity auroral signal episodes were named PAIW because they appear to be associated with pulsating-aurora displays. The characteristics and morphology of both the AIW and the PAIW auroral infrasound events are described below using examples from the I53US array.

AIW Bow Waves from Auroral Electrojet Arcs Supersonic Motions

An example of the wave-train of a very large AIW bow-wave observed at I53US on Sept. 11, 2005 (day 254) at 08:47 UT is shown in Figure 1. The pressure versus time traces from all eight sensors at I53US have been phase-aligned and superimposed in Figure 1 to show the high coherence MCCM = 0.979 for the signal across the array. The observed AIW trace velocity and back-azimuth are 0.559 km/sec and 46.7 deg. respectively. The peak-to peak amplitude of this AIW is about 0.9 Pascal. There was daylight at I53US at the time of the signal in Figure 1 thus no aurora video data is available. The geomagnetic data from the Poker Flat magnetic observatory, 30 km north of I53US, show large fluctuations in the H and Z magnetic components that indicate the presence of strong auroral electrojet currents that could have been the source of this AIW.

The geomagnetic traces of the H, D and Z components from Poker Flat observatory for Sept.11, 2005 are shown in Figure 2 . Very large fluctuations can be seen in H (north-south) , D (east-west) and Z (vertical) components that are typical of those associated with strong auroral electrojet currents during an auroral substorm. The presence of an auroral electrojet arc moving across the zenith at Poker Flat toward the I53US infrasonic array at Fairbanks can be inferred from an analysis of the magnetic data. If the auroral electrojet current is modeled as a line current that is of great length relative to its height above the surface, both the speed of lateral motion of the arc and its direction of motion can be determined from the time- variations in H, D and Z. For a westward flowing auroral electrojet current, as the arc moves laterally toward the magnetic observatory, the H component should show a southward change that becomes a maximum as the arc crosses the zenith. At zenith crossing time the Z component should change from minus (upward) to plus (downward) for a line current.

The magnetic data from the interval 08:38 to 08:45 UT was de-trended to remove the average values from H, D, and Z in order to separate the permanent geomagnetic field from the magnetic induction resulting from the auroral line current. The Total Horizontal Disturbance vector was formed from the de -trended H and D components as THD = Sqrt ( H ^2 + D^2). In Figure 3 the THD component and the de-trended Z component are plotted to show that as THD becomes a maximum the Z comp changes from negative to positive as the line current crosses the zenith at Poker Flat at about 08:41 UT.

The direction of the Total Horizontal Disturbance vector is perpendicular to the auroral line current axis and should therefore be parallel to the direction of the propagation of the bow-wave created by the supersonic motion of the auroral arc as is predicted by the simple AIW generation model.

A plot is shown in Figure 4, by a blue line, of the end-point of the THD vector in the geomagnetic D H Plane for the period 08:38 to 08:45 UT at Poker Flat for Sept. 11, 2005. The black arrow gives the direction of the THD vector at the time of its maximum value. The red arrow gives the direction of propagation of the AIW signal as it crossed the I53US array at 08:47 UT. For this particular example of an AIW associated with an auroral electrojet source there is excellent agreement between the directions of the THD vector at maxima and that of the AIW bow-wave propagation. The time of arrival of the AIW at I53US at 08:47 UTis consistent with the propagation time expected from a source auroral arc that crosses the Poker Flat zenith at 08:40:45 UT for a source height of 110 kilometers.

It is also possible to determine the speed of the lateral motion of the line current from its magnetic field components variations with time. For a line current at a height h above the surface the speed of lateral motion is given by V = h * (d/dt)(Bz/Bh) where Bz is the vertical component and Bh is the total horizontal component THD. ( see Charles R. Wilson, Infrasonic Waves From Moving Auroral Electrojets, Planet. Space Sc. 1969, Vol. 17, pp 1107 to 1120). This determination of arc speed has been made for the AIW event on Sept. 11, 2005. The results are shown in Figure 5 as a plot of the ratio of the Z component to the THD component for the period 08:39:40 to 08:41:40 UT for Poker Flat de-trended magnet data. In Figure 5 the blue line is a plot of the ratio of Z to THD as a function of time in seconds. A linear curve fitting has been done to this blue data line. This linear curve is shown as a red line with a slope of 0.0046 (1/sec.). The speed is then given by the assumed height of 110 km for the line current multiplied by the slope of the linear curve. This product gives a lateral motion speed of 0.506 km/sec. According to the model of AIW bow-wave generation the speed of the arc motion should be the same as that of the trace-velocity of the received AIW signal which is shown in Figure 1 to be 0.559 km/sec. These two numbers 0.506 and 0.559 agree well within the uncertainty of +/- 0.047 km/sec for the determination of the trace-velocity of the AIW across the I53US microphone array.

Many similar examples of AIW and their associated auroral line-current magnetic fluctuations have been found at I53US since its operation began in 2002.

High Trace Velocity PAIW Infrasound Signals

Recently a new type of auroral infrasound signal has been identified at both I53US in Alaska and I55US in Antarctica that is often associated with the pulsating auroras occurring during an auroral substorm’s final hours following the dynamic “break-up” phase of the bright auroral arcs. This class of auroral infrasound signals has been provisionally named PAIW.

The pulsating aurora infrasound source is quasi-stationary in the lower ionosphere, at a height of about 120 km, periodically turning on and off, at the same location in the sky radiating infrasound isotropicaly. Thus it is feasible that the PAIW source regions could be identified by triangulation from two separated infrasonic arrays. The ultimate source of the PAIW in the pulsating auroras patches is assumed to be the periodic heating of thin layers of the neutral gas in the atmosphere due to the auroral electron precipitation. The 1960 infrasonic observations from Washington D.C. indicated that there was an infrasound source, moving across the northern auroral regions from east to west throughout the night, during times of strong geomagnetic disturbance. The exact nature of this polar infrasound source was not known. In 1972 at UAF in a study of infrasound observed from four stations, Roland Johnson (Johnson, R.E., Planet and Space Sci., Vol 20, 313-329, 1972) found that it was possible to trace the east-west motion of polar substorms using auroral infrasonic data from four widely spaced infrasonic arrays at Washington D.C., Boulder, Colorado, Pullman, Washington, and College, Alaska. It is our contention that this infrasound source, which was observed at lower latitude infrasonic arrays and moved westward across the northern sky, was from the PAIW signals associated with westward moving auroral substorm activity.

Because of the large positive temperature gradient with increasing height in the atmosphere above the 100 km level, infrasound radiated from pulsating aurora patches at 120 km will be refracted strongly downward. Ray tracing, using an atmospheric model with a stratified temperature profile and no wind, indicates that a PAIW wave packet from a pulsating auroral patch at the 120 km level will arrive at the surface with a horizontal trace velocity that increases rapidly to infinity as the horizontal distance of the source patch from the array zenith point decreases to zero. Thus PAIW with observed trace velocities of (a) 0.5 km/sec, (b) 1.0 km/sec, (c) 1.5 km/sec, and (d) 2.0 km/sec will have come, respectively, from source patches that have distances, from the infrasonic array zenith, of (a) 63 km, (b) 26 km, (c) 18 km, and (d) 12 km. Thus the observed trace velocity of a PAIW wave packet can vary from about 0.40 km/sec for a source patch that is 100 km away from the zenith to infinity for a source that is directly overhead. Therefore one can not assign a characteristic value for PAIW trace velocity. However, from a study of twenty one PAIW events observed at I53US in all of year 2003 for PAIW signals, with velocity greater than 1.00 km/sec and less than 5.0 km/sec, the average horizontal trace velocity was found to be 1.698 km/sec with a Std = 0.27 km/sec. The PAIW ray path travel times from the 120 km level to the surface vary from 250 seconds for a source overhead to 320 sec. for a source 100 km from the zenith. The only way to unambiguously identify a very high trace velocity infrasonic wave packet with a particular pulsating auroral source would involve the analysis of auroral all-sky video data in conjunction with ray-tracing information using the measured temperature and wind profile of the atmosphere at the time in question.

At I53US on November 19, 2003 (day 323) at Fairbanks there was a large geomagnetic and auroral substorm that produced high trace velocity infrasound signals from 08:00 to 18:00 UT during a time when the clear night sky allowed good all-sky video data of the auroras to be recorded. Specific PAIW signals, as shown in Figure 6 , were observed from 10:40 to 10:50 UT that are associated with visible pulsating aurora forms that were positioned within a circular area of the sky of radius 50 kilometers above Fairbanks. The infrasonic waveform data in Figure 6 were passband filtered from [0.02 to 0.08 Hz). The peak-to-peak amplitude of these signals is about 0.2 Pa. Their high trace velocity, (Vt), of 0.619 km/sec indicates that the ray path of the PAIW wave packet arrived at the surface at an angle of 57 degrees above the horizontal. The phase-aligned overlay of the waveforms from all 8 microphones in Fig. 6 indicates a coherence of 0.769 for the PAIW signals. Signals similar to the PAIW shown in Figure 6 continued for 10 hours well into the daylight hours at Fairbanks. The mean trace velocity for the signals in this PAIW event was 0.980 km/sec for all signals for which Vt > 0.60 km/sec.

The spectrum of the PAIW wave train of Figure 6 is shown in Figure 7 with maxima at periods of 30, 22 and 15 seconds respectively. Examples of PAIW auroral infrasound signals of the continuous type are given in Figure 6 and Figure 9 from both the Northern and Southern hemispheres at I53US in Fairbanks and I55US in Antarctica.

Spectral analysis of the aurora video all-sky-camera time-series data, for day 323,2003 in the interval from 10:38:30 to 10:47:00 UT, indicates that there is an enhancement of the auroral luminous energy in the passband from ( 0.1 to 0.5 Hz) during periods when visible pulsating auroras are in the field of view of the camera. Therefore a search was made for coherent auroral infrasound in this same high frequency passband in hour 10 UT of day 323. Figure 8 below shows an example of the high frequency PAIW signals that were found from 10:44 to 10:49 UT with a high Vt = 1.145 km/sec. These PAIW signals were coming from the same general azimuth 213 deg. as that of the longer period PAIW signals for the same time interval as shown in Figure 6. These high frequency PAIW signals were found to be highly coherent (coh. = 0.831) when only the inner three sensors of the 8 sensor I53US pentagon-triangle array were used in the MCCM analysis of the waveforms. This loss of spatial-coherence in the high frequency PAIW for large microphone separations (1.7 km) is similar to that found for microbaroms of frequency 0.2 Hz as observed at I53US. This high-frequency PAIW event on day 323 continued for at least an hour while pulsating auroras were active. The principle energy peak in the spectrum of the signals shown in Figure 7 occurs at about 0.18 Hz with a period of 5.6 seconds that is similar to that of the ubiquitous microbaroms. Fortunately the very high trace velocity of 1.145 km/sec for the high-frequency PAIW is three times larger than that of microbaroms. This allows for unambiguous identification of the PAIW high frequency signals.

Simultaneous observation of PAIW at both I53US in Alaska and I55US in Antarctica

Continuous type PAIW auroral infrasound signals, as distinct from the AIW bow waves, are also observed at I55US in Windless Bight, Antarctica. The geomagnetic latitude of a particular site determines its location with respect to the auroral oval where auroral infrasound is generated. The geomagnetic latitude of I55US in Antarctica is 79.72 deg South while I53US in Fairbanks is at 65.12 deg North. Thus I55US is almost 15 deg. farther toward the South magnetic pole than I53US is from the North magnetic pole. Therefore one should not expect the auroral infrasound to be identical at I53US and I55US. However on October 29, 2003, during a global magnetic storm following a very large solar flare, in the time period from 15 to 17 UT the simultaneous observations of auroral infrasound occurred at both I53US and I55US. An example of PAIW auroral infrasound of high trace velocity is shown in Figure 9 from I55US. In late October in Antarctica at I55US there is 24 hours of daylight and thus no aurora, if present, could have been observed. The infrasound wave train shown in Figure 9 has a trace velocity, coherence and a spectral content that is similar to that seen at I53US during the same time interval from 15 to 16 UT. Thus these I55US infrasound signals are almost certainly associated with the auroral and magnetic storm that produced similar PAIW signals in Alaska at I53US. Studies of auroral infrasound will be continued at I53US and I55US over the next few years during their winter auroral seasons.

Video Time Series of Pulsating Auroras at Fairbanks and PAIW Signal Compared

On December 5, 2003 (day 339) the skies above I53US were clear with an extensive auroral display that ended with a pulsating aurora event that lasted many hours. Excellent aurora optical data were obtained that night with an image-enhanced video camera that is located at the Poker Flat rocket range 25 kilometers north of I53US. The pulsating aurora patches covered the entire sky for several hours. During the time interval on day 339 from 16:39:33 to 16:45:40 UT there was a very regular pattern of auroral pulsations that turned on and off, at the zenith above the camera, with a period of about 18.4 seconds (freq = 0.054 Hz). A time series, as shown in Figure 10 below, of this fluctuating auroral intensity was obtained from the video data. The spectrum of the video data showed a prominent peak at 0.05 Hz. The optical data from the all-sky image of the video camera was taken from a circular area of radius 35 Km centered on the zenith point above the camera.

At the I53US array high trace velocity coherent PAIW infrasonic signals were received in association with the auroral activity from hours 12 through 23 UT on day 339. The propagation time from the pulsating aurora source overhead at a height of 120 kilometers to the surface is about 5 to 6 minutes. A six minute interval of the infrasonic waveform is shown in the Datascan analysis plot from 16:46 to 16:52 UT in Figure 11 below. The infrasonic data were bandpass filtered from (0.02 to 0.10 Hz). The trace velocity for the signal was 0.684 km/sec with a coherence value of 0.905 from an azimuth of 323 deg.

The infrasonic wave field at the surface is, of course, the sum of all the infrasound produced by the pulsating auroral patches that are distributed all over the sky. Having acknowledged this difficulty it is, however, still useful to try and compare the auroral optical time series data from the zenith area and the infrasonic best-beam data at the surface to see if they have a significant correlation. The Matlab functions COHERE, that estimates the magnitude squared coherence function between two signals, was used to compare the optical and infrasonic data sets. The infrasonic data interval was delayed by five minutes with respect to the optical data to account for acoustic propagation time to the surface. Both the infrasonic and the optical time series data were bandpass filtered from (0.02 to 0.10 Hz). The coherence Cxy is a function of the power spectra Px of the optical data and the power spectra Py of the infrasonic data and of the cross spectrum Pxy of x and y. The value of Cxy, ( a number between 0 and 1), is plotted in Figure 12 below as a function of frequency for the PAIW event pictured in Figure 11 and the video data shown in Figure 10 .

This process was repeated to obtain the frequency domain coherence Cxy between the optical and infrasonic data with no time shift between the two data samples. The result was a decrease in the average value of the Cxy by a factor of ½ from that shown in Figure 12 where the propagation time was allowed for.

Summary of PAIW characteristics

• High trace velocity infrasound (PAIW), with Vt > 0.60 km/sec, is often observed at Fairbanks when pulsating auroras are visible in the dark skies overhead. A typical average trace velocity for PAIW event is 1.52 Km/sec.
• PAIW infrasound occurs in the passband from 0.02 to 0.5 Hz with p-t-p amplitudes up to 0.30 Pascal.
• PAIW events at I53US are observed most often in the morning hours after the auroral “break-up” phase, however they have been observed at all times of the day.
• PAIW signals come from all azimuths when the source region is centered above Fairbanks.
• The seasonal variation in PAIW events is the same as that of the auroral activity.
• The coherence, as a function of frequency, of the video data of pulsating auroras overhead and the PAIW waveform data from the surface becomes a maximum when a 5 to 6 minute time delay is used for the infrasonic data with respect to the auroral video data. This time delay allows for the propagation time for the infrasound from the source to the surface.
• PAIW events occur several times a month at Fairbanks throughout the year. There have been more than 80 PAIW events at Fairbanks over the past two years at I53US.