Thread: SAC Help Downsampling

Started: 2015-06-25 19:03:00
Last activity: 2015-06-28 14:42:02
Topics: SAC Help
Matteo Bagagli
2015-06-25 19:03:00
Hello to all!
I'm having some trouble with the downsampling of my data.
I have a lot of time series sampled at 1 sec (delta), and I need those files
to be sampled every 600 sec (10 minutes). Now I tried to apply manually a
lowpass IIR filter at the new nyquist frequency (1/1200 Hz) first , and
later on
used the function "INTERP DELTA 600" to achieve the results.
Unfortunately the results seems aliased. The function decimate (which I
need to call
several times) creates a lot of "border effects" (ringing) that makes my
data too noisy.
Anyone has experienced the same problem? My workflow for this step force
me to
remain in SAC, I just want to know if there's a good method before using
another program.

Hoping to hear from you soon, have a great day!
Matteo

  • Milton Plasencia
    2015-06-26 02:34:27
    Hi Matteo,

    You can try first apply rmean command, perhaps it help to avoid the border effects,
    if not try taper command.

    Try:
    r sac_file
    rmean
    decimate 6
    p # to see the effects on the waveform.
    decimate 5
    decimate 5
    decimate 4

    You can use p after each decimate command.
    Here decimation factor 6*5*5*4=600.

    HTH,

    Milton
    Attachments
  • George Helffrich
    2015-06-28 14:42:02
    Dear All -

    After looking at the raw data, it was obvious that it contained glitches — sudden, brief changes by a factor of 10**6 in the signal level. These should be removed before decimating. The anti-alias properties of the downsampling filters carried the glitches into the downsampled data — as they should. When the glitches were removed, the downsampling provided smooth data even at the large decimation required for this use.

    Simply put, one should remove all glitches in data before downsampling it. Don’t expect downsampling to fix data for you.

    On 25 Jun 2015, at 18:42, Matteo Bagagli <matteo.bagagli<at>gmail.com> wrote:

    Hello to all!
    I'm having some trouble with the downsampling of my data.
    I have a lot of time series sampled at 1 sec (delta), and I need those files
    to be sampled every 600 sec (10 minutes). Now I tried to apply manually a
    lowpass IIR filter at the new nyquist frequency (1/1200 Hz) first , and later on
    used the function "INTERP DELTA 600" to achieve the results.
    Unfortunately the results seems aliased. The function decimate (which I need to call
    several times) creates a lot of "border effects" (ringing) that makes my data too noisy.
    Anyone has experienced the same problem? My workflow for this step force me to
    remain in SAC, I just want to know if there's a good method before using another program.

    Hoping to hear from you soon, have a great day!
    Matteo


    ----------------------
    SAC Help (http://ds.iris.edu/message-center/topic/sac-help/)

    Sent via IRIS Message Center (http://ds.iris.edu/message-center/)
    Update subscription preferences at http://ds.iris.edu/account/profile/



    George Helffrich
    george.helffrich<at>bris.ac.uk


10:46:56 v.22510d55