Experimental Log: Roger Bland, January 8, 2002
Find RAFOS peaks automatically.
( logbook |2001 data | ifremer on RAFOS |Garfield at NPS |raftimes )

10:30 Tuesday January 8, 2002. Yesterday I compared two randomly selected signals from the 4:00Z RAFOS beacon, and the times agreed to within a couple of seconds or better. I need an automated method for finding the peaks and estimating their arrival time. This shouldn't be too difficult.

16:49 I have got a manual-scanning program going, a PERL script, rafscan, calling an IDL program called scn1raf, which can also be called stand-alone. I am tabulating times for the (4:00Z, 16:00Z) beacon for all the data we have. The variance looks to me like about 0.2 sec, pretty good!

I have run across one file which seems to have FOUR signals! Now, we don't believe that there were ever more than three. I will just note the times as best I can read them from the large-scale graph:

Run r0130315.55d
16.21 hours, 16.57 hours, 17.13 hours, 17.565 hours
For comparison, the times expected from the known positions of sources and PSM are
16.208 hr, 16.564, 17.144, 17.568
This is pretty close! I wonder if this is after Toby's source was immersed. I think that it is. Later on I will have to look for all peaks in a systematic manner.

First, though, I have to establish what a "good" RAFOS peak looks like, so that I can teach the program to find them. The problem is that right now I am seeing a peak-doubling phenomenon with the 4:00Z signal which I don't understand at all.

17:35 I think I see the 5.13 hours signal in r0131003.39d.

17:51 Crazy beeping every 5 minutes (rather regular) on the RAFOS channel! Is a sperm whale trying to mate with a RAFOS beacon? (run r0131315.40d) Actually this is probably the NPS tests, which were due to start at 18:00 GMT on day 313.

18:26 Here is the summary of the arrival times for the signal which comes about 12 minutes after 4:00 or 16:00 GMT every day. The signal is doubled, for reasons unknown, so both arrival times are tabulated and graphed. The rms of the arrival times is about 0.2 seconds. There is no trend with time evident over the 23-day run (day 290-day 313) (Oct. 17-Nov. 9). The change in arrival time might be estimated at less than 0.3 sec, out of a total travel time of 750 sec. This translates into a change of temperature of about 0.25 degrees Celsius. I don't know what change should have been anticipated, but I bet that over a year there are changes easily measured this way. Of course, a few CTD drops would accomplish more or less the same thing.

18:36Good day's work. Go Home.