[ncl-talk] power spectrum of filtered dataset

[Dennis Shea]

ncl-talk has no idea what you have done. The series has * *pentad** [5-day]
temporal sampling. The plots have the frequency axis labeled
'cycles/month'.

ncl 0> f = addfile("rainfall_pc_00000.nc","r")
ncl 1> ymd = cd_calendar(f->time,-2)
ncl 2> print(ymd)

(0)     19810303
(1)     19810308
(2)     19810313
(3)     19810318
(4)     19810323
(5)     19810328
(6)     19810402
(7)     19810407
(8)     19810412
(9)     19810417
(10)    19810422
(11)    19810427
(12)    19810502
(13)    19810507
.
.
It is *your responsibility *[**not ncl-talk's**]  to design a filter that
addresses the pentad time sampling to create a (say) 20-60 day filter.
ncl-talk's experience is that dealing with this type of issue leads to many
emails. We can no longer do that.

Good Luck

On Tue, Sep 15, 2020 at 6:43 AM Edwin Marco via ncl-talk <
ncl-talk at mailman.ucar.edu> wrote:

> Dear ncl users,
>
> I am trying to prove whether I have correctly filtered my dataset by
> plotting the power spectrum of the filtered data set.
>
> My approach:
> firstly I have plotted the spectrum of the unfiltered data set (
> rainfall_pc_00000.nc) using attached script
> (plot_spectrum_unfiltered_data.ncl).
>
> Secondly, I decided to filter my dataset at 20-60 days.
>
> Thirdly; Using the Butterworth filtering technique, filtered the dataset (
> rainfall_pc_00000.nc) at 20-60 days, and plot the spectrum of the
> filtered dataset.
>
> Problem: After plotting the spectrum of the filtered dataset, I intended
> to see all peaks out of 60-20days bandpass to disappear however it came the
> opposite way.
>
> I have attached scripts, plots, and datasets for easy follow-up.
>
> Please help me know if my approach to proof whether I have correctly
> filtered the datasets is correctly and then where I am getting wrong and
> plot opposite results, contrary to the theory.
>
> Thanks
>
> Edwin
>
>
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at mailman.ucar.edu
> List instructions, subscriber options, unsubscribe:
> https://mailman.ucar.edu/mailman/listinfo/ncl-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20200915/9a74aadd/attachment.html>