[ncl-talk] How to choose the value of "sm" in spec_5.ncl

[Dennis Shea]

re:  "a time-series which is a two month long daily data for 30 years
*concatenated*"

---two-months-of-daily-data-- *10-month-break* --
two-months-of-daily-data-- *10-month-break* -- etc, etc

The purpose of spectral analysis is to *decompose a time series into
periodic components*.   Concatenating non-contiguous time-series segments
does not allow proper decomposition into periodic components.

You can perform a spectral analysis on each two-month [60 day] segment and
average the spectra.

*https://www.ncl.ucar.edu/Document/Functions/Built-in/specx_anal.shtml*
<https://www.ncl.ucar.edu/Document/Functions/Built-in/specx_anal.shtml>
See Example 5

In this case, you would compute the periodogram of each segment, average
the resulting periodogram estimates at each frequency.
With 30 segments and two degrees of freedom for each periodogram you would
get about 60 degrees of freedom for each estimate.

The longest frequency/period would be 1/60 days [60 days].

ncl-talk is not appropriate for statistical questions. Please consult a
local statistician.

Good Luck


On Mon, May 11, 2020 at 5:47 AM Bidyut Bikash Goswami via ncl-talk <
ncl-talk at ucar.edu> wrote:

> Dear All,
> I have a time-series which is a two month long daily data for 30 years
> concatenated. So the length of the time series is 60days*30. My question is
> how do I choose the value of "sm" in spec_5.ncl
>
> I am attaching spectra of my time-series with sm=3,5,7,9,1,13,15 and 21.
>
> Thanks.
>
>
> --
>
>
> *Bidyut Bikash GoswamiResearch ProfessorIrreversible Climate Change
> Research Center (IRCC)*
> *Yonsei University*
> *Seoul, South Korea*
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20200511/977f5efd/attachment.html>