Re: a bug from specx_anal?

There is no bug in 'specx_anal'.

You have set the 'jave' argument to zero ( 0 ).
This means that no averaging window will be
used on the raw spectral estimates. The result
is a periodogram. Each estimate would have 2 degrees
of freedom (dof). However, you have set the 'pct' argument
to 0.1 which means the data have been tapered.
Hence, the dof is reduced slightly ===> 1.895

[1] you always want to taper any series that is not periodic
to reduce 'leakage'.

[2] You should always set 'jave' to, at least, 3.

Xuguang Sun wrote:
> Dear everyone,
>
> I use the function specx_anal() to do spectrum analysis, however, after the step like
>
> sdof = specx_anal(ts,0,0,0.1)
> print(sdof)
>
> I always got the same value of sdof, i.e., 1.895535, whatever I used various time series. Is it a bug or should it be like that?
>
> Thanks.
>
> Regards,
> Xuguang
>
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Fri Aug 28 2009 - 09:31:26 MDT