Some time ago, I posted a request for help in understanding a bowl shaped spectral feature I saw in recordings of aircraft flyovers. This feature can be seen in the Air Tractor spectrogram. First my son and subsequently, commenter Alberto Armani both identified the cause of the spectral feature as a multi-path interference pattern, specifically due to ground reflection. Mr. Armani suggested that if the microphone was placed on the ground rather than being held above it, the bowl shaped feature would not occur. Sometime after his comment, I had the opportunity to record the sound of the Air Tractor as it was seeding some fields near our house. I found a position on top of a hill, in a mowed lawn and placed my Canon Powershot G10 on the ground for one recording.
The spectrogram of the fly over with the microphone on the ground is shown below:
Just as Mr. Armani had suggested, the bowl-shaped feature is gone. In its place, presumably obscured by the ground reflection in the previous study, is a beautiful family of Doppler shifted lines. What could these be?
A quick examination showed these curves to be harmonically related with an apparent fundamental frequency of about 103 Hz. Documents on the Air Tractor website show the propeller speed to be in the range of 1700 to 2200 RPM. The photograph of the aircraft shows a three blade propeller. The blade passing frequency is calculated by multiplying its speed in revolutions per second by the number of blades. In this instance, assuming a speed of 2000 RPM, I calculated the blade passing frequency to be 100 Hz. The family of curves I see in the spectrogram now is probably the Doppler shifted propeller blade passing frequency and several of its harmonics.
I extracted several time and frequency pairs from a zoomed in on the propeller blade passage tone around 100 Hz in the Raven Lite spectrogram display. This data was then entered into Igor Pro where the Doppler shift analysis as described previously. The curve fit to the propeller data is shown below:
PROPELLER BLADE PASSAGE TONE Doppler shift analysis results:
Coefficient values ± 95% Confidence Interval
f =100.63 Hz ± 0.307
v =177.36 ft/sec ± 4.68 = 120.93 miles/hour ± 3.19
t0 =26.084 sec ± 0.0875
h =408.04 ft ± 41.4
I also curve fit to the turbo blade passage tone as I had done in the previous study using the same technique. Those results are also shown:
TURBO BLADE PASSAGE TONE Doppler shift analysis results:
Coefficient values ± 95% Confidence Interval
f =2854.6 Hz ± 5.36
v =179.71 ft/sec ± 3.05 = 122.53 miles/hour ± 2.08
t0 =26.092 sec ± 0.0495
h =419.72 ft ± 24.7
The reader may have also noticed another feature in the spectrogram centered on the 50 second mark in the recording and showing a family of peaks. These are also Doppler shifted sounds but for a much different flight path than the one in my model. Shortly after aircraft passed overhead toward the east it banked hard to the north and then west to make a seeding run parallel and to the north of its approach path that brought it over my position. The peaked signature carries information not only about the relative direction that the aircraft is flying but also of engine speed changes required to make those tight seeding maneuvers.
Murmurs from the Earth... 