Thread: Re: decimation of SAC files

Started: 2014-05-16 16:23:31
Last activity: 2014-05-16 16:23:31
Topics: SAC Help
Sheila Peacock
2014-05-16 16:23:31
Dear Nasir Mahmood,

Thanks for your email.

I am copying my reply to the sac-help email list so that others might
see the question and reply. Do you belong to the sac-help list?
If not, then you can join it by visiting and looking
under "Data and Software", then under the sub-menu "Support",
then "list of all mailing lists", then click on "sac-help"
and follow the instructions.

You asked:

On 05/16/2014 07:39 AM, nasir mahmood wrote:
Dear Professor
hopefully you will be fine and good. I have small question about the SAC, if you can help me it will be great.
I want to run a moment tensor program, but fortran code did not recognize my data. I attache ISBA and KSR station SAC files. I want to know how I can change my data files for ISBA station ( with sampling rate 100sps) exactly similar to KSR station( with 20SPS) which work fine available inside program example.
How I can filter the data to the common freq. range to avoid possible aliasing effect due to decimation, and how I can make the data in the same physical unit. I mean how I can do this in SAC. I mean I want to make ISBA similar to KSR station.
Your help in this regard will be great for me.
my best wishes ever for you
Nasir, from Pakistan.

To try to answer your questions:

1. I think you need to deconvolve the instrument response from
your data to convert the seismograms into common units of
displacement. For this you need the "transfer" command in SAC,
and information on the instrument responses. If the data were
given to you in SEED format then the instrument responses can
be extracted with "rdseed" into files with names including
the word "RESP". You can then use the command
"transfer from evalresp subtype <RESP-filename> to none"

Otherwise you need to find the response of the instrument,
most probably from the manufacturer's data sheet, in the
form of poles and zeros. If you put the poles and zeros
into a file in the correct SAC format, then the command to
use is

"transfer from polezero subtype <polezero-filename> to none"

It is likely that you will have to use the option "freqlimits"
in the "transfer" command to prevent noise from overwhelming
the seismogram in frequency bands where the instrument response
is weak. Use the command "help transfer" to find out the
options to the "transfer" command.

2. Once you have deconvolved the instrument response from
both seismograms, they will be in the same units (displacement)
and you can then reduce the sampling frequency of the seismograms
from ISBA. For this you need the command "decimate", which will
apply an anti-alias filter before reducing the sampling rate.
"help decimate" will show the options for this command.

Hope that helps.

Sheila Peacock.

PS beware that the poles and zeros in a manufacturer's data
sheet are likely to be in the velocity or acceleration
domain, so you will have to add one or two "zeros" at (0.0,0.0)
to convert them to displacement. Check the data sheet carefully,
and ask the manufacturer if it is not clear. Also if they are Guralp
seismometers, the poles and zeros are given in Hz not
radians in the data sheet, and have to be multiplied by 2pi to conform with
SAC convention. If you have to change the poles or zeros for
this reason, you need to re-calculate the constant. This
unfortunately needs a specialist computer program or
some careful algebra if you do it by hand.

(Is there anyone on the sac-help list who can point us all to
a website that describes CLEARLY how to derive the
"constant" from a set of poles and zeros, so that enquirers
like Nasir Mahmood can see how it has to be done?)

Below is a typical SAC-format pole-zero file.
-0.14860E+00 0.14860E+00
-0.14860E+00 -0.14860E+00
-0.11310E+04 0.00000E+00
-0.10053E+04 0.00000E+00
-0.50265E+03 0.00000E+00
0.00000E+00 0.00000E+00
0.00000E+00 0.00000E+00
0.00000E+00 0.00000E+00
CONSTANT 0.91364E+08

06:06:15 v.eb79165e