Date: Fri Apr 04 2014 - 06:45:52 MDT
Hi NCL,
I figured out this problem !! It must be noted that in order to convert the
phase 0-360 to -phi to +phi in NCL, we need to multiply the results by -1,
as follow. This will give a realistic result in term of upward propagation
of quasi-stationary waves.
finfo(1,:,:,:) =
where(finfo(1,:,:,:).gt.180,finfo(1,:,:,:)-360,finfo(1,:,:,:)) *-1
Danke!
Thomas
On Fri, Apr 4, 2014 at 1:28 AM, Dennis Shea <shea@ucar.edu> wrote:
> finfo = fourier_info (....)
>
> Assuming
>
> x(klvl,nlat,mlon) ; where mlon=144 and
>
> dlon= 360.0/mlon ; 2.5
>
> nhx = 3
> finfo = fourier_info(x, nhx, dlon) ;===> finfo(3,klvl,nlat,nhx)
>
> finfo(1,:,:,:) = where(finfo(1,:,:,:).gt.180,
> finfo(1,:,:,:)-360,finfo(1,:,:,:))
>
>
> On 4/3/14, 4:32 PM, Thomas Tobian wrote:
>
>> HI Dennis and NCL developer(s),
>>
>> I am not really sure, since (as you can see) there is a strong scratch
>> from
>> 30 to 80N at lower atmosphere produced by NCL compared to Matlab output.
>> Also, 30N poleward up to 10 hPa, the phases decrease with height, "which
>> are not correct for upward rossby wave propagation (matlab is correct!)! I
>> am wondering. Is there any way around to have a phase output in NCL from
>> -phi to +phi? Thanks
>>
>>
>> On Fri, Apr 4, 2014 at 12:43 AM, Dennis Shea <shea@ucar.edu> wrote:
>>
>> [0] I doubt fourier_info has a bug
>>>
>>> [1] It is difficult to compare visually since NCL returns the
>>> phases as 0 to 360 while Matlab (apparently) uses -180 to 180.
>>> It would be better to match the same ranges.
>>>
>>> [2] Suggestion: Use a 'circular' color map to color plot phases.
>>> That way you do not get abrupt color changes
>>> at the 'ends': eg 359->1 or -179 and 179
>>>
>>> [3] Download the just released 6.2.0. It has few circular color tables
>>> that may be of use.
>>> http://www.ncl.ucar.edu/Download/
>>>
>>>
>>>
>>> On 4/2/14, 2:07 AM, Thomas Tobian wrote:
>>>
>>> Hallo,
>>>>
>>>> I've recently tried to calculate the phase from one season data (DJF)
>>>> with
>>>> dimension [level,lat,lon]. I rescaled the phase in ncl by 2.5 since the
>>>> rightmost dimension size is 144. Unfortunately in NCL the calculated
>>>> phase
>>>> for wave1 (plotting with contour function), produced really different
>>>> patterns compared to Matlab. I'm just really curious, is it a bug in
>>>> phase
>>>> function of NCL? any suggestion would be really appreciated. Please look
>>>> at
>>>> the contour pattern.
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> ncl-talk mailing list
>>>> List instructions, subscriber options, unsubscribe:
>>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>>
>>>>
>>>>
>>
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
