Dynasonde AGC algorithms

A short report on some tests made 7-8 March

I ran two 8-pulse soundings after each other every 12 minutes from about 17 UT on 7 March to about 13 UT on 8 March 1997. The first in the pair was with AGC=4, the standard one now, developed at BAS to set the AGC level in the absence of echoes to a level determined by the noise of say interfering stations. The second of the pair was with AGC=1, a simple algorithm which tends to set the attenuation to zero when echoes are not found. I wanted to compare the two quantitatively to confirm that AGC=4 is the better one.

The motivation for this was the test of non 100 Hz PRF (e.g 8192us IPP) which showed that AGC=4 is unusable in the presence of PRE/Digisonde interference because this interference sets the attenuators too high in the absence of valid echoes, whereas the simple AGC=1 ignores the background interference level and gave better results.

The results are summarised in the EXCEL spreadsheet which includes a plot of the results. I used the size of the soundings as an indicator of the number of echoes. Of the 84 pairs of soundings, the average ratio of the sizes of soundings with AGC=4 compared to AGC=1 is 1.45 (with a st dev of 1.15) showing that on average we get more echoes with AGC=4.
The plots are interesting though, because there are times when the ratio is less than 1. The ratio is also largest when the number of echoes is smallest, presumably at night when there are many interfering stations and weak echoes.