Here is the background noise diurnal for 2013 at 29.5kHz
http://abelian.org/vlf/tmp/29499_noise_2013.png
The noise is averaged over a 500Hz band and converted to units
of fT per root Hz.
The noise is before sferic blanking. Divide the noise
amplitude by 10 to estimate the noise after blanking.
Multiply the noise amplitude by the square root of your
bandwidth to obtain your RMS noise floor. Or instead, divide
the noise by the square root of your Fourier transform width
in seconds.
For example, from the plot the noise during the night in early
March is in the purple but on poor nights has a hint of red.
Lets say a bad night then is 15 fT/root(Hz).
After blanking that would be 1.5 fT, and in a 600 second Fourier
transform the floor would be 1.5/sqrt(600) = 0.06 fT in each
Fourier bin.
Compare that with the noise level between 19:00 and 21:00 in
http://abelian.org/vlf/tmp/29499_140307a.gif
and we see the estimate from the noise map is in the right
ballpark.
Markus wrote (in a nearby thread):
> saw practically nothing here on 29.499 last night.
Signal level from Bob is lower. NAA maintains normal level.
> are you continuously saving all raw data from your VLF loops?
Only the lower 24 kHz of 3 channels (2xH + vertical E) are
saved raw, spectrum data is saved for 0 to 96kHz. Storage of
the timestamped flac-compressed data is not too onerous, at
the last count it was heading for 18 Tbytes.
Hard disks are big and cheap these days fortunately.
Also reliable too, in the last 3 years I've only had one disk
error (which was revealed by md5 checksum verification) which
turned out to be a single bit flip in one byte of one file.
I strongly recommend recording signals to disk and doing signal
analysis by post-processing. Then you don't miss anything
and you don't have to guess in advance the optimum antenna
pointing, spectrogram settings, and so on. It is easy to
keep a cache of 10 days or a month of signal. The fact is,
almost always you get to hear about interesting things *after*
they've happened, especially with natural radio events.
You really are very limited and will miss a lot if you only
work with real time signals.
--
Paul Nicholson
--
|
