[ncl-talk] spectrum significant level

Dennis Shea shea at ucar.edu
Sat Jul 25 10:09:50 MDT 2020

```White Noise represents the same variance at all frequencies. The spectrum
is flat.
----
sdof = *specx_anal(*x,d,sm,pct)
print(sdof)
nspc = dimsizes(sdof at spcx)
;************************************************
; calculate average spectrum value
;************************************************
white  = *sum*(sdof at spcx)/nspc     ; average; same power at all
frequencies   ; scalar
;************************************************
; upper and  lower confidence bounds
; *** I can not remember if *cdfchi_x i*s the correct function
*  ****
;************************************************
w025  = white**cdfchi_x* (0.025,sdof)/sdof                      ; scalar
w975  = white**cdfchi_x *(0.975,sdof)/sdof                      ; scalar
print(nspc+"  "+ssum+"  "+w025+"  "+white+"  "+w975)

WHITE = *conform*(sdof at spcx, white, 0)      ; WHITE[*] ==> WHITE(nspc)
W025   = *conform*(sdof at spcx, w025, 0)      ; W025[*]
W975   = *conform*(sdof at spcx, w975, 0)      ; W975[*]

Good Luck

On Fri, Jul 24, 2020 at 1:51 PM Jing Sun via ncl-talk <
ncl-talk at mailman.ucar.edu> wrote:

> Dear NCL-users,
> Has anybody ever calculated the significant level of the spectrum
> estimated from the "white-noise"?
> The specx_ci only offer the Markov "Red Noise" spectrum significant level.
> Does anyone have experience on this?
> Thanks.
> Best regards,
> Jing
>
>
>
> _______________________________________________
> 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...