John Diebold provided some source signatures for a 6 GI gun linear array
towed at 10m, 12m, 15m, 18m, and 20m.

Download the SEGY signatures:

JD's plot (UAF_array_10m.pdf)
The time domain plot of all signatures is sig1.png.
sig1a.png is an expanded plot of all 5 signatures.
The power density frequency plot is sig2.png.
Comments:
1) The ghost at 10m is 20m / 1500m/s = .013333s
   The ghost at 20m is 40m / 1500m/s = .026666s
2) The ghost at 10m causes a notch in the frequency spectrum at 75Hz (1./.01333)
   The ghost at 20m causes a notch in the frequency spectrum at 37.5Hz.

Filters:
sig1b.png is a plot of 10m and 20m signatures with 10x90, 20x90, 30x90, 40x90 filters.

NMO stretch:
A reflector at 1.250 TWT (two way travel-time) with a 6km streamer would look like this. 
sig3a.png is an expanded plot of range 0 and 1000.
After NMO this. (Enlarged here.)
NMO stretches or lowers the frequency content.  (plot)
Every range stretches, or delays, the ghost, so stack will destructively add.

AUTOCORRELATIONS:
sig-acorr.png is a plot of the autocorrelation of all five signatures.

SPIKING DECON (pdist 0 fillen .2 sedts 0 .3)
plot of 10 meter signature with spiking decon and various prewhiteners.
plot of 20 meter signature with spiking decon and various prewhiteners.
plot of all signatures with spiking decon with prewhi 0.
plot of autocorrelations.
Comments:
0)  Spiking decon (pdist 0) uses the prediction error as the filter.
1)  Is spiking decon the same as designature?

Water depth of 100m synthetic with 10m signature.
Used to remove the water bottom multiple.  Using the 10m signature, a water depth of 100m,
and a water bottom reflection coefficient of .6, then, the water bottom comes in at .133s
with amplitude 1., the first multiple is at .266 with amplitude -.6, the second multiple is
at .4 with amplitude .36, the third at .533 with amplitude -.216, the fourth at .666 with
amplitude .1296.

UCK:
1)  These signatures do not taper to zero, which causes a discontinuity.  To solve this,
filter the signature with a 10x110 bandpass filter.
2)  All these signatures have significant bubbles.  See sig1.png.

Water depth of 225m synthetic with 10m signature.
Changing the example to reflectors at .1s, .4s, .7s and a filtered 10m signature (plot)
plot of the autocorrelation.
Plot with spiking decon (fillen .2 sedts 0 .5 pdist 0 prewhi 0 double yes)
plot of the autocorrelation after spiking decon.
Plot with predictive decon (fillen .6 pdist .3 sedts 0 1. prewhi 0 double yes)
plot of the autocorrelation after predictive decon (fillen .6 pdist .3 sedts 0 1. prewhi 0 double yes)