Loading ...
Sorry, an error occurred while loading the content.
 

Re: [VLF_Group] Re: Intrinsic mode functions

Expand Messages
  • Paul
    ... Yes. It seems in this case that the noise is not strong enough compared with the sferic, so that during the sferic, the only extrema are those of the
    Message 1 of 17 , Dec 1, 2010
      Lionel wrote:
      > Ideally, low order IMF should contain only noise, and not
      > part of the useful signal as here.

      Yes. It seems in this case that the noise is not strong enough
      compared with the sferic, so that during the sferic, the only
      extrema are those of the sferic - there are no smaller ones
      riding on top.

      Another puzzle - the remainder is supposed to be monotonic but
      it isn't in this test. Does that mean there should be another
      IMF to extract? Should I have subtracted the mean from the
      raw signal before applying EMD? Something is a bit wrong there.

      > You are correct. There's over-simplification in the Wikipedia
      > description.

      Thanks.
      --
      Paul Nicholson
      --
    • Lionel Loudet
      Paul, ... The word monotonic is a bit misleading here. It means here no more than one minimum or maximum . The process would also have stopped with only
      Message 2 of 17 , Dec 1, 2010
        Paul,

        > Another puzzle - the remainder is supposed to be monotonic but
        > it isn't in this test. Does that mean there should be another
        > IMF to extract?

        The word "monotonic" is a bit misleading here. It means here " no more than one minimum or maximum".
        The process would also have stopped with only one minimum and one maximum.
        You can find an example of similar residue in Huang's paper, page 922.

        > Should I have subtracted the mean from the
        > raw signal before applying EMD? Something is a bit wrong there.

        This is not necessary. As stated by Huang "Even for data with zero mean, the final residue can still be different from zero; for data with a trend, then the final residue should be that trend."

        Best Regards,
        Lionel.
        --
      • Paul
        ... Yes, I see now. I looked at the residue on p922 and it looked monotonic, but on closer inspection there appears to be a minimum at 17 seconds (and maybe
        Message 3 of 17 , Dec 1, 2010
          Lionel wrote:
          > The word "monotonic" is a bit misleading here. It means here "
          > no more than one minimum or maximum". The process would also
          > have stopped with only one minimum and one maximum. You can
          > find an example of similar residue in Huang's paper, page 922.

          Yes, I see now. I looked at the residue on p922 and it looked
          monotonic, but on closer inspection there appears to be a minimum
          at 17 seconds (and maybe even a slight max at 25 seconds?).

          I continue to experiment on sferics. One thing to try is to
          apply EMD to the spectrum instead of to the time domain waveform.
          --
          Paul Nicholson
          --
        Your message has been successfully submitted and would be delivered to recipients shortly.