readme

su 的源代码库

The demos for sudivstack and supws give examples on program usage but 
(unfortunately) do not demonstrate where one of the different stacking 
algorithms performs best.

Phase-weighted stacking (supws) uses a coherency measure, i.e. the phase 
stack, to weight the traces of an ordinary stack. Thus, the signal must 
be (very) coherent for good results. If you type "sudoc supws" on the 
command line (without the quotes), you'll find references to the 
algorithm near the bottom the documentation printed.

Diversity stacking down-weights traces with high-amplitude, incoherent 
noise and is superior to an ordinary true-amplitude stack, if the mean 
signal energy is smaller than the mean energy of the noise. Again you'll 
find references, if you type "sudoc sudivstack".

To compare the performance of the different stacking algorithm, you may 
play around with the Supws demo. You could change the signal-to-noise 
ratio in a copy of the Supws script to very poor values and run the 
modified script again. For example, replace the four lines beginning 
with "suaddnoise" by
   suaddnoise < record.su sn=0.9  > record.noise
   suaddnoise < record.su sn=0.5 >> record.noise
   suaddnoise < record.su sn=1.0 >> record.noise
   suaddnoise < record.su sn=0.8 >> record.noise

and compare the stacking results. In this case, the diversity stack 
yields the very best result. Also with real data "sudivstack" is often a 
good choice, possibly with a different window length ("winlen=" 
parameter) then the default value. Although I wrote "supws", it is 
difficult for me to say generally, when the phase-weighted stack 
performs best...


> Not 
> only that I will be extremely glad if you can explain the diffraction 
> (ndpfz in shell script) and how it can be overcomed by using phase 

To get information on the "ndpfz", you may have to look into the source 
code of "susynlv" ("$CWPROOT/src/su/main/susynlv.c"), and to explore the 
effect of different values for "ndpfz", you may  play around with the 
demo scripts in "$CWPROOT/src/demos/Synthetic/Kirchhoff/Susynlv".

Good luck,

Nils


-- 
Nils Maercklin
NORSAR
P.O. Box 53
2027 Kjeller, Norway
Phone: [phone]    E-Mail: [e-mail]
Fax:   [phone]    Web:    http://www.norsar.no