Paul,
I'm guessing that your 3 sigma to 4 sigma estimate is based on a comparison of
the 2970Hz pixels' amplitudes to surrounding pixels' amplitudes during the time
that Stefan was transmitting; is that a correct guess?
If so, do you know of (or suspect that there could be) a numerical adjustment
to confidence level based on/off time coherence (the close agreement between
observed received signal on and off times and known transmitter on and off
times, where off times include times long before and long after transmission)?
In other words, could there be a numerical adjustment to confidence level, that
includes acknowledgement of the observed close temporal alignment of long
absence of received and transmitted signals?
(the above independent of the knowledge of Stefan's TX frequency)
In a similar context, do you suspect that there could be a numerical adjustment
to confidence level based on (a) time-agnostic knowledge that energy at 2970Hz
is likely to be higher than at surrounding frequencies because of a known
transmission at 2970Hz, combined with (b) correlation of transmit spectrum with
received spectrum (analogous perhaps to a temporal correlation)
I have as many reasons for thinking that both of the above double-count (i.e.
that they have already been accounted for) as for thinking that they have
independent validity; can you comment?
Thanks,
Jim AA5BW
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Paul Nicholson
Sent: Saturday, February 11, 2017 10:56 AM
To: [email protected]
Subject: Re: ULF: 5 wavelengths on the 101 km band? Valid or not?
I scraped the pixels off the Cumiana spectrogram, (summing each row) and did my
best guess of reversing the mapping of power to pixel brightness.
http://abelian.org/vlf/tmp/170211a.png
2970 is the strongest line. At least 3 sigma, maybe 4, depending on how you
treat the lumpy floor. A physicist would insist on 5 sigma but the fact that
the peak is at exactly the right frequency is significant in itself.
Markus just wrote:
> In my humble opinion, this is clearly a successful > detection.
I was doubtful looking at the spectrogram but having plotted the pixels I am
convinced.
Spectrograms aren't good for this sort of thing. Oh
for a spectrum plot! I couldn't get anything from
the stream recording, too many timing breaks on the uplink.
Best I can get in Todmorden is 2 and a bit sigma using just the daytime signal
in 3.9 uHz. Not significant at all. Would need at least another 7 days of
transmission.
--
Paul Nicholson
--
|
