(meteorobs) Radio Meteor "Double Pings"

Steve Harrison wrote:

>With that notion having been introduced, I'd like to bring up another such
>observation regarding radio meteors. Quite often during the late summer,
>Shel Ennis and I will hear a meteor "ping" that was what we've termed a
>"double"; that is, one ping quickly followed by another, spaced a few
>hundredths or less milliseconds apart. It seems likely to me that those
>radio observers listening with FM radios may not even notice this since the
>reflections at lower frequencies last much longer than at the frequency
>where Shel and I have head these double pings, 144 MHz (that is, the second
>of the double ping may be obscurred by the long-lasting first of the double
>ping). We aren't the first to notice them, either; radio amateurs have long
>heard double pings since the beginnings of radio amateur meteor scatter
>work in the early 1950's.

Hello Steve,

Based upon the above paragraph, I can relate two phenomena to you which
might fit the description of causing successive peaks in an underdense
trail reflection which are a "few hundred or less milliseconds apart" --
unless you really meant "few hundredths of a millisecond or less," in which
case the below models won't help.  Your time description above is a bit
ambiguous.  Nonetheless, the potential cause that I see for the double
peaks is that you are witnessing an interference pattern between the
different parts of the meteor trail as (1) the trail is initially formed by
the meteor, OR (2) the trail rapidly expands following its initial
formation.  These should be more noticeable phenomena at your operating
wavelength (about 2.01 meters) than at the more standard FM and TV bands.
Note that I am being hypothetical here [insert standard disclaimer], and am
throwing out guesses more than being definitive.  

The simple model that I am looking at involves a path midpoint type of
reflection, with the meteor trail normal to the radio wave plane of
propagation.  the "reflective" part of the meteor trail is a rapidly
expanding cylindrical shell of free electrons whose density is proportional
to the meteor mass and velocity.  The primary portion of an underdense
radio wave reflection will come from the electrons in this cylinder over a
length of about 1-3 km centered on the point in the trail where the
forward-scatter geometry is properly met (called the first Fresnel zone).
 The length of this zone is dependent upon link distance and operating
frequency.  As the meteor initially approaches the center of the first
Fresnel zone, called F1 below, the trail forming in the meteor's wake will
begin to reflect the radio wave from the transmitter to the receiver with
ever increasing signal strength (over the course of about 50 msec or so for
a Perseid type meteor).  However, each part of the trail will have a
slightly different total radio path length, creating  an interference
pattern to be seen in the power level or phase of the received signal.
This will appear as a rapidly fluctuating signal (oscillating on the order
of milliseconds) near or just above the background noise level -- until the
meteor crosses the middle of the F1 zone, whose center is called the t0
point.  At time t0, an extremely sharp peak in echo power is observed,
generally tens of dB above the previous tiny oscillating signal.  The
reflection from the F1 part of the trail will begin immediately to
exponentially decay, as the parent meteor travels on to the F2, F3, and so
on regions.  This will cause another interference pattern, called the
post-t0 pattern, to be seen in the received signal, and this is what I
suspect you are witnessing.  

In most systems, the very strong signal from the F1 region will only show a
shallow oscillation during the post-t0 interference pattern time (again
less than about 50 msec).  However, if the wavelength for the link is
short, and the meteor is fast (with a high trail altitude and quick
diffusion coefficient), it might be possible that in some instances, the
destructive interference created as the meteor crosses the boundaries
between Fresnel zones (called T1, T2, T3, etc. below) could reduce the
signal from the F1 region enough to effectively (although momentarily)
block it out.  As the meteor reaches the center of F2, the F1 signal and F2
signal will add, creating a second observed peak and overriding the
destructive interference from T1.  This same type of pattern could continue
as the meteor moves on to T2 (destructive) and F3 (constructive).  

Note that this entire process occurs only during the initial trail
formation, about 100 msec in time, or less, and is indicative of the link
frequency, link distance, and meteor speed.  the meteor back-scatter radars
have used both the pre-t0 and post-t0 interference patterns (either power
or phase) in order to obtain meteor speeds.  It can be done to a much less
accurate extent in forward-scatter.  To give you an idea, below is a
portion of a maple worksheet which calculates the times and meteor flight
path distances involved in such a interference pattern.  Based upon the
input that you have listed:

link distance;  1300 km
operating frequency; 144 MHz
meteor speed (Perseid);  59.4 km/sec

(A hot spot, link mid-point, normal trail reflection is assumed)

Now look at these zone sizes in a row:

> evalf(f1t1_km);
> evalf(t1f2_km);
> evalf(f2t2_km);
> evalf(t2f3_km);

The times to cross each zone in milliseconds is:

> evalf(f1t1_km / metvel_msec);
> evalf(t1f2_km / metvel_msec);
> evalf(f2t2_km / metvel_msec);
> evalf(t2f3_km / metvel_msec);


In other words, this would give about 20 msec between the first, large t0
(or F1) peak and the next F2 peak, then about 8 msec between the F2 peak
and the F3 peak.  Note that the oscillation period rapidly reduces with
distance from the t0 point.  As i said, in most systems, these oscillations
are superimposed upon the exponential decay of the large F1 peak and
generally don't override it, but I am curious if this is possibly what you
are seeing.  If you can give me some more definite link / system / shower
time parameters, and some measured times off an audio spectrogram of one of
your wave files, we might be able to obtain a rough measure of meteor speed
and verify what you are seeing.

A second possibility for the successive peaks is an interference pattern
cause by the different radio path lengths  from a reflection from the top
of the meteor trail cylinder and the bottom of the meteor trail cylinder.
It is just such interference which causes the initial exponential decay of
the underdense trail in the first place, but I am wondering aloud here (and
adding salt and pepper to my shoe) if your short wavelength could expedite
the rapid decay of the initial signal to the point that a second
constructive interference peak can be observed before the trail has
dissipated to the point of returning no useful signal at all.  The observed
oscillation period would be proportional to the diffusion rate at the given
meteor altitude, and would be notably longer than the trail formation
oscillations described above.  A rough idea as to the oscillation period is
given in the below quick-and-dirty  worksheet;


> restart;

Initial inputs:

> r := 6378;
r := 6378
> h1 := 100;
h1 := 100
> za := (90 - 5.123437139) * (Pi/180);
za := .4715364603*Pi
> lambda :=  2.01 / 1000;
lambda := .2010000000e-2
> c1 := 7.4;
c1 := 7.4

find approx. distance to meteor trail (km);

> dis := sqrt((r^2) * (cos(za)^2) + (2*r*h1) + (h1^2)) - (r * cos(za));
dis :=
> dis := evalf(dis);
dis := 699.2936713

find approx. Earth angle (deg);

> theta := evalf(arccos((dis^2 -r^2 -(r+h1)^2) / (-2 * r * (r+h1))));
theta := .1077259506
> evalf((180/Pi)*theta);

find approx. diffusion coefficient (m/sec);

> d := 10^((0.067 * h1) -5.6);
d := 12.58925412

find approx. initial radius for meteor trail (m);

> r1 := 10^((0.075 * h1) -c1);
r1 := 1.258925412

Find approx. height for 180 deg out of phase reflection (km)

> eh1 := (dis+ (lambda/2))^2 = r^2 + eh^2 - (2*r*eh*cos(theta));
eh1 := 489013.0443 = 40678884+eh^2-12682.05562*eh
> sol1 := solve(eh1, eh);
sol1 := 6204.050690, 6478.004930
> h2 := sol1[2] - r;
h2 := 100.004930

find difference in trail radius from initial (r1) to out-of-phase radius
((r2), m);

> dh := 1000 * (h2-h1);
dh := 4.9300
> r2 := dh/2;
r2 := 2.465000000
> dr := r2 - r1;
dr := 1.206074588

find difference in time, based upon diffusion rate;

> dt := dr/d;
dt := .9580190983e-1


This would give a very rough peak to peak oscillation period of 192 +/- 127
msec, depending upon the initial trail radius calculation used (the above
worksheet shows an average of the models).  Note that the oscillation
period here is much longer than during trail formation, and does not vary
significantly (although not much more than 1 peak should be observed).
Note that I have not before encountered such an interference pattern as
this one, either personally or in the literature, so I am not very
confident about this second possibility.

Additionally, recall that large overdense trails can also display a
oscillating pattern, on the ordr of seconds or tenths of a second, caused
by upper atmospheric winds breaking apart the still-reflecting portions of
the meteor trail.  These are quite common on 6-meters, but should still be
prevalent even on 2-meters.  But your initial time description sounded much
shorter than this and more closely related to underdense "pings," so I
discounted this possibility.

Let me know which of the three above possibilities look better (if any),
and supply me with some more definite link information and measurements
from an audio spectrogram if you would like to try to pin this down further.

Best regards,


James Richardson
Tallahassee, Florida

Operations Manager / Radiometeor Project Coordinator
American Meteor Society (AMS)

To UNSUBSCRIBE from the 'meteorobs' email list, use the Web form at: