Thread: Transfer problem

Started: 2014-02-19 05:57:27
Last activity: 2014-02-21 21:50:38
Topics: SAC Help
Blaine Bockholt
2014-02-19 05:57:27
I am attempting to remove the instrument response of some data. I downloaded the RESPFILE from IRIS and while using the evalresp option of transfer, I get the following error:


"WARNING (norm_resp): computed and reported sensitivities[ET.CPCT.00.HHE] differ by more than 5 percent."


I thought the final sensitivity at the end of the file is the product of all the different gain/sensitivities of each stage. When I calculate it by hand, I get the sensitivity reported. I am not sure where this error is coming from, nor I am sure how SAC calculates the sensitivity.? Thanks.


Blaine


Here is the command I use


trans from evalresp fname ET.CPCT.resp to none freqlim 0.001 0.002 15 20

  • Brian Savage
    2014-02-20 21:44:20
    Dear Blaine,

    This error is generated by evalresp. You should be able to confirm this by running evalresp on your RESP file.

    For example
    % evalresp CPCT HHE 2010 001 0.001 20 -v -f RESP.ET.CPCT.HHE.2
    << EVALRESP RESPONSE OUTPUT V3.3.3 >>
    WARNING (norm_resp): computed and reported sensitivities differ by more than 5 percent.
    Execution continuing.
    --------------------------------------------------
    RESP.ET.CPCT.HHE.2
    --------------------------------------------------
    ET CPCT 00 HHE 2009,001,00:00:00 2011,026,00:00:00
    Seed units: Velocity(in)->Counts(out)
    computed sens=1.14868E+09 (reported=1.25829E+09) @ 5.00000E-02 Hz
    calc_del=8.92261E-01 corr_app=8.92261E-01 est_delay=8.92261E-01 final_sint=0.01(sec/sample)
    stage 1: LAPLACE A0=6.259322E+08 NZeros= 2 NPoles= 5 Sd=1.369329E+03
    stage 2: GAIN Sd=1.000000E+00
    stage 3: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 1 SamInt=1.953125E-06 Sd=8.388610E+05
    stage 4: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 36 SamInt=1.953125E-06 Sd=1.000000E+00
    stage 5: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 6 SamInt=1.562500E-05 Sd=1.000000E+00
    stage 6: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 7 SamInt=3.125000E-05 Sd=1.000000E+00
    stage 7: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 48 SamInt=6.250000E-05 Sd=1.000000E+00
    stage 8: FIR_SYM_2 H0=1.000000E+00 Ncoeff=126 SamInt=2.500000E-04 Sd=1.000000E+00
    stage 9: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=5.000000E-04 Sd=1.000000E+00
    stage 10: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=1.000000E-03 Sd=1.000000E+00
    stage 11: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=2.000000E-03 Sd=1.000000E+00
    --------------------------------------------------
    %

    Possibly someone at IRIS can comment on this further. I think the first gain value is being modified due to a Analog PoleZero.

    Brian
    savage<at>uri.edu

    On Feb 18, 2014, at 4:57 PM, Blaine Bockholt (bbckholt) wrote:

    I am attempting to remove the instrument response of some data. I downloaded the RESPFILE from IRIS and while using the evalresp option of transfer, I get the following error:

    "WARNING (norm_resp): computed and reported sensitivities[ET.CPCT.00.HHE] differ by more than 5 percent."

    I thought the final sensitivity at the end of the file is the product of all the different gain/sensitivities of each stage. When I calculate it by hand, I get the sensitivity reported. I am not sure where this error is coming from, nor I am sure how SAC calculates the sensitivity.​ Thanks.

    Blaine

    Here is the command I use

    trans from evalresp fname ET.CPCT.resp to none freqlim 0.001 0.002 15 20
    _______________________________________________
    sac-help mailing list
    sac-help<at>iris.washington.edu
    http://www.iris.washington.edu/mailman/listinfo/sac-help


    • Philip Crotwell
      2014-02-21 05:43:17
      Hi

      Responses are at best confusing, but you need to remember that the gain is
      not just a value, it is a value at a particular frequency.

      The last stage (stage zero) is supposed to be the overall gain and should
      be the product of the gains at each stage, but only if the frequency of
      each stage is the same. In this case the stage zero gain is +1.25829E+09,
      but it is given at freq of 5e-2. The stage 1 gain is 1.5e3, but at a freq
      of 1Hz. So to be proper, you must adjust the gain based on the poles and
      zeros from 1Hz to 5e-2 Hz before you multiply. Looking at the response plot
      here:

      http://service.iris.edu/irisws/evalresp/1/query?net=ET&sta=CPCT&loc=00&cha=HHZ&time=2009-06-07T00:00:00&units=def&annotate=true&output=plot

      it looks like 5e-2 is just barely beyond the flat part of the response,
      which explains why evalresp says that the stage 1 gain is 1.369329E+03
      instead of 1.5e3. It has rolled off just a bit at 5e-2 Hz compared to 1Hz
      which is in the flat part of the response.

      To be honest, I think that the mistake in this response is not in the gain
      numbers, but rather in the frequency reported in the stage zero. It
      probably should have the frequency in the stage zero as 1Hz instead of
      5e-3Hz . So, likely in this case the 1.25829E+09 is the gain you want to
      use even though evalresp complains.

      Hope this helps,
      Philip



      On Thu, Feb 20, 2014 at 1:44 PM, Brian Savage <savage<at>uri.edu> wrote:

      Dear Blaine,

      This error is generated by evalresp. You should be able to confirm this
      by running evalresp on your RESP file.

      For example
      % evalresp CPCT HHE 2010 001 0.001 20 -v -f RESP.ET.CPCT.HHE.2
      << EVALRESP RESPONSE OUTPUT V3.3.3 >>
      WARNING (norm_resp): computed and reported sensitivities differ by more
      than 5 percent.
      Execution continuing.
      --------------------------------------------------
      RESP.ET.CPCT.HHE.2
      --------------------------------------------------
      ET CPCT 00 HHE 2009,001,00:00:00 2011,026,00:00:00
      Seed units: Velocity(in)->Counts(out)
      computed sens=1.14868E+09 (reported=1.25829E+09) @ 5.00000E-02 Hz
      calc_del=8.92261E-01 corr_app=8.92261E-01 est_delay=8.92261E-01
      final_sint=0.01(sec/sample)
      stage 1: LAPLACE A0=6.259322E+08 NZeros= 2 NPoles= 5
      Sd=1.369329E+03
      stage 2: GAIN Sd=1.000000E+00
      stage 3: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 1 SamInt=1.953125E-06
      Sd=8.388610E+05
      stage 4: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 36 SamInt=1.953125E-06
      Sd=1.000000E+00
      stage 5: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 6 SamInt=1.562500E-05
      Sd=1.000000E+00
      stage 6: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 7 SamInt=3.125000E-05
      Sd=1.000000E+00
      stage 7: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 48 SamInt=6.250000E-05
      Sd=1.000000E+00
      stage 8: FIR_SYM_2 H0=1.000000E+00 Ncoeff=126 SamInt=2.500000E-04
      Sd=1.000000E+00
      stage 9: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=5.000000E-04
      Sd=1.000000E+00
      stage 10: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=1.000000E-03
      Sd=1.000000E+00
      stage 11: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=2.000000E-03
      Sd=1.000000E+00
      --------------------------------------------------
      %

      Possibly someone at IRIS can comment on this further. I think the first
      gain value is being modified due to a Analog PoleZero.

      Brian
      savage<at>uri.edu

      On Feb 18, 2014, at 4:57 PM, Blaine Bockholt (bbckholt) wrote:

      I am attempting to remove the instrument response of some data. I
      downloaded the RESPFILE from IRIS and while using the evalresp option of
      transfer, I get the following error:

      "WARNING (norm_resp): computed and reported sensitivities[ET.CPCT.00.HHE]
      differ by more than 5 percent."

      I thought the final sensitivity at the end of the file is the product of
      all the different gain/sensitivities of each stage. When I calculate it by
      hand, I get the sensitivity reported. I am not sure where this error is
      coming from, nor I am sure how SAC calculates the sensitivity. Thanks.

      Blaine

      Here is the command I use

      trans from evalresp fname ET.CPCT.resp to none freqlim 0.001 0.002 15 20
      _______________________________________________
      sac-help mailing list
      sac-help<at>iris.washington.edu
      http://www.iris.washington.edu/mailman/listinfo/sac-help



      _______________________________________________
      sac-help mailing list
      sac-help<at>iris.washington.edu
      http://www.iris.washington.edu/mailman/listinfo/sac-help



      • Robert Casey
        2014-02-21 15:33:55

        Thanks for the helpful reply, Philip. I had Mary Templeton looking at it and she came the same conclusion as you. She made a plot to illustrate:





        So the recommendation, I take it, is for the frequency of sensitivity, Blockette 58, stage 0, field 5 to be changed to 1.00000E+00.

        What I noticed was that the Normalization frequency of stage 1 is in fact set to 1 Hz, so I wonder if this is a product of merging the sensor to a datalogger response and then not seeing a trickle-down alteration of the frequency of sensitivity to stage 0? The PDCC tool doesn't handle this, for example, but I do recall having discussions about this and seeing the solution as ambiguous.

        So if Blaine edits the RESP file, is that the way to proceed?

        -Rob



        Hi

        Responses are at best confusing, but you need to remember that the gain is not just a value, it is a value at a particular frequency.

        The last stage (stage zero) is supposed to be the overall gain and should be the product of the gains at each stage, but only if the frequency of each stage is the same. In this case the stage zero gain is +1.25829E+09, but it is given at freq of 5e-2. The stage 1 gain is 1.5e3, but at a freq of 1Hz. So to be proper, you must adjust the gain based on the poles and zeros from 1Hz to 5e-2 Hz before you multiply. Looking at the response plot here:

        http://service.iris.edu/irisws/evalresp/1/query?net=ET&sta=CPCT&loc=00&cha=HHZ&time=2009-06-07T00:00:00&units=def&annotate=true&output=plot

        it looks like 5e-2 is just barely beyond the flat part of the response, which explains why evalresp says that the stage 1 gain is 1.369329E+03 instead of 1.5e3. It has rolled off just a bit at 5e-2 Hz compared to 1Hz which is in the flat part of the response.

        To be honest, I think that the mistake in this response is not in the gain numbers, but rather in the frequency reported in the stage zero. It probably should have the frequency in the stage zero as 1Hz instead of 5e-3Hz . So, likely in this case the 1.25829E+09 is the gain you want to use even though evalresp complains.

        Hope this helps,
        Philip



        On Thu, Feb 20, 2014 at 1:44 PM, Brian Savage <savage<at>uri.edu> wrote:
        Dear Blaine,

        This error is generated by evalresp. You should be able to confirm this by running evalresp on your RESP file.

        For example
        % evalresp CPCT HHE 2010 001 0.001 20 -v -f RESP.ET.CPCT.HHE.2
        << EVALRESP RESPONSE OUTPUT V3.3.3 >>
        WARNING (norm_resp): computed and reported sensitivities differ by more than 5 percent.
        Execution continuing.
        --------------------------------------------------
        RESP.ET.CPCT.HHE.2
        --------------------------------------------------
        ET CPCT 00 HHE 2009,001,00:00:00 2011,026,00:00:00
        Seed units: Velocity(in)->Counts(out)
        computed sens=1.14868E+09 (reported=1.25829E+09) @ 5.00000E-02 Hz
        calc_del=8.92261E-01 corr_app=8.92261E-01 est_delay=8.92261E-01 final_sint=0.01(sec/sample)
        stage 1: LAPLACE A0=6.259322E+08 NZeros= 2 NPoles= 5 Sd=1.369329E+03
        stage 2: GAIN Sd=1.000000E+00
        stage 3: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 1 SamInt=1.953125E-06 Sd=8.388610E+05
        stage 4: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 36 SamInt=1.953125E-06 Sd=1.000000E+00
        stage 5: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 6 SamInt=1.562500E-05 Sd=1.000000E+00
        stage 6: FIR_SYM_1 H0=1.000000E+00 Ncoeff= 7 SamInt=3.125000E-05 Sd=1.000000E+00
        stage 7: FIR_SYM_2 H0=1.000000E+00 Ncoeff= 48 SamInt=6.250000E-05 Sd=1.000000E+00
        stage 8: FIR_SYM_2 H0=1.000000E+00 Ncoeff=126 SamInt=2.500000E-04 Sd=1.000000E+00
        stage 9: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=5.000000E-04 Sd=1.000000E+00
        stage 10: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=1.000000E-03 Sd=1.000000E+00
        stage 11: FIR_SYM_1 H0=1.000000E+00 Ncoeff=501 SamInt=2.000000E-03 Sd=1.000000E+00
        --------------------------------------------------
        %

        Possibly someone at IRIS can comment on this further. I think the first gain value is being modified due to a Analog PoleZero.

        Brian
        savage<at>uri.edu

        On Feb 18, 2014, at 4:57 PM, Blaine Bockholt (bbckholt) wrote:

        I am attempting to remove the instrument response of some data. I downloaded the RESPFILE from IRIS and while using the evalresp option of transfer, I get the following error:

        "WARNING (norm_resp): computed and reported sensitivities[ET.CPCT.00.HHE] differ by more than 5 percent."

        I thought the final sensitivity at the end of the file is the product of all the different gain/sensitivities of each stage. When I calculate it by hand, I get the sensitivity reported. I am not sure where this error is coming from, nor I am sure how SAC calculates the sensitivity. Thanks.

        Blaine

        Here is the command I use

        trans from evalresp fname ET.CPCT.resp to none freqlim 0.001 0.002 15 20
        _______________________________________________
        sac-help mailing list
        sac-help<at>iris.washington.edu
        http://www.iris.washington.edu/mailman/listinfo/sac-help


        _______________________________________________
        sac-help mailing list
        sac-help<at>iris.washington.edu
        http://www.iris.washington.edu/mailman/listinfo/sac-help


        _______________________________________________
        sac-help mailing list
        sac-help<at>iris.washington.edu
        http://www.iris.washington.edu/mailman/listinfo/sac-help


        Attachments
09:26:47 v.01697673