Re: Fwd: uv2dvF and uv2vrF NaN values

Arrrgh, sorry, there was a typo in the message below. It should be:

   x = where(abs(x).gt.1e20, x@_FillValue, x)

If "x" doesn't have a _FillValue, then use "default_fillvalue" to assign one:

if(.not.isatt(x,"_FillValue")) then
  x@_FillValue = default_fillvalue(typeof(x))
end if
x = where(abs(x).gt.1e20, x@_FillValue, x)

You may need to replace the "1e20" value as appropriate, but it should be good for most cases.

--Mary

On Feb 21, 2014, at 2:23 PM, Mary Haley <haley@ucar.edu> wrote:

> Correction on my last email on "Inf" versus "NaN":
>
> Begin forwarded message:
>
>> From: Dennis Shea <shea@ucar.edu>
>> Subject: Re: uv2dvF and uv2vrF NaN values
>> Date: February 21, 2014 2:22:45 PM MST
>> To: Mary Haley <haley@ucar.edu>
>>
>> offline ....
>>
>>
>> re: I'm assuming that "Inf" values will be seen as "NaN"
>>
>> I don't think this is true
>>
>> I think you need something like
>>
>> x = where(abs(x).gt.120, x@_FillValue, x)
>>
>> D
>>
>> On 2/21/14, 2:17 PM, Mary Haley wrote:
>>> Jon,
>>>
>>> It would be an issue if you are passing in "Inf" or "NaN" values to *any* function in NCL, except for two special functions we have for checking this kind of thing.
>>>
>>> You can use the "isieee_nan" and "replace_ieeenan" functions to "fix" any data that has bad values in it.
>>>
>>> See the FAQ:
>>>
>>> http://www.ncl.ucar.edu/FAQ/#data_hand_004
>>>
>>> I'm assuming that "Inf" values will be seen as "NaN" values by these functions. Let me know if you try this and it doesn't work.
>>>
>>> --Mary
>>>
>>> On Feb 21, 2014, at 12:28 PM, jonathan meyer <tallj03@hotmail.com> wrote:
>>>
>>>> Hi Mary,
>>>>
>>>> I've been digging into this and believe I have found the issue with returns of NaNs.
>>>>
>>>> It seems that the input arrays that were being fed to these functions have some 'inf and -inf' values in a few grid cells.
>>>>
>>>> I re-created the input files so no 'inf' values were included and the functions seem to be working normally (although plotting even a single panel is painstakingly long which I'm guessing is a product of the very small values of vorticity and divergence and the precision of the contour levels needed????).
>>>>
>>>> It seems the functions saw some of these 'inf' values and choked?
>>>>
>>>> I can send along both versions of the input files and my working .ncl code if you feel it is necessary to look into the issue further.
>>>>
>>>>
>>>> Thanks,
>>>> Jon
>>>>
>>>>
>>>>
>>>>> Subject: Re: uv2dvF and uv2vrF NaN values
>>>>> From: haley@ucar.edu
>>>>> Date: Thu, 20 Feb 2014 08:53:37 -0700
>>>>> CC: ncl-talk@ucar.edu
>>>>> To: tallj03@hotmail.com
>>>>>
>>>>> Hi Jon,
>>>>>
>>>>> If *any* NCL function is returning NaN, this is a problem. They should be returning missing values or else issuing some kind of error.
>>>>>
>>>>> Can you provide the two NetCDF files online? You can use our ftp:
>>>>>
>>>>> http://www.ncl.ucar.edu/report_bug.shtml#HowToFTP
>>>>>
>>>>> --Mary
>>>>>
>>>>>
>>>>> On Feb 19, 2014, at 3:50 PM, jonathan meyer <tallj03@hotmail.com> wrote:
>>>>>
>>>>>> Good afternoon NCL-ers,
>>>>>>
>>>>>> I am computing ageostrophic winds using actual winds and geostrophic winds. In order to convince myself the ageostrophic wind fields satisfy the mathematics of the atmosphere I am computing the divergence and vorticity of the ageostrophic winds which should equal zero everywhere.
>>>>>>
>>>>>> Using the uv2dvF and uv2vrF functions to compute these two fields returns NaN values for the entire array. I've confirmed there are no missing data, and the dataset is on a global fixed grid with the latitude monotonically increasing (-90 to 90).
>>>>>>
>>>>>>
>>>>>>
>>>>>> My question:
>>>>>>
>>>>>> Is the return of NaN an artifact of the functions where divergence/vorticity is actually zero?
>>>>>>
>>>>>> I've tested these functions using actual U and V winds and they return appropriate ranges of values, so something is unique to the ageostrophic winds.
>>>>>>
>>>>>> Thanks in advance,
>>>>>> Jon
>>>>>>
>>>>>>
>>>>>> My Code Snippet
>>>>>>
>>>>>> begin
>>>>>> ;==================================================================
>>>>>> print("ESTABLISH CCSM FILES AND RETRIEVE DIMENSION INFO")
>>>>>> ; =================================================================
>>>>>>
>>>>>> f1_in = addfile(dir1 + "CCSM_REG_U_test1979.nc", "r")
>>>>>> f2_in = addfile(dir2 + "CCSM_REG_V_test1979.nc", "r")
>>>>>>
>>>>>> lev = f1_in->lev
>>>>>> lat = f1_in->lat2(::-1)
>>>>>> lon = f1_in->lon2
>>>>>>
>>>>>> n_lev = dimsizes(lev)
>>>>>>
>>>>>> ; =================================================================
>>>>>> print("READ IN CCSM DATA")
>>>>>> ; =================================================================
>>>>>>
>>>>>> Ug = f1_in->Ug(:,:,::-1,:)
>>>>>> U = f1_in->U( :,:,::-1,:)
>>>>>>
>>>>>> Vg = f2_in->Vg(:,:,::-1,:)
>>>>>> V = f2_in->V( :,:,::-1,:)
>>>>>>
>>>>>> ; =================================================================
>>>>>> print("COMPUTE AGEOSTROPHIC WINDS")
>>>>>> ; =================================================================
>>>>>>
>>>>>> Ua = U-Ug
>>>>>> Va = V-Vg
>>>>>>
>>>>>> ; =================================================================
>>>>>> print("COMPUTE DIVERGENCE AND VORTICITY OF AGEOSTROPHIC WINDS")
>>>>>> ; =================================================================
>>>>>>
>>>>>> print(num(ismissing(Ua)))
>>>>>> print(num(ismissing(Va)))
>>>>>>
>>>>>> div = uv2dvF(Ua,Va)
>>>>>> vor = uv2vrF(Ua,Va)
>>>>>>
>>>>>> ****RESOURCE SETTING BLAH BLAH*******
>>>>>>
>>>>>> do z = 0, n_lev-1
>>>>>> print("Plotting level " + lev(z))
>>>>>>
>>>>>> resdiv@gsnCenterString = "Divergence"
>>>>>> plot(0) = gsn_csm_contour_map_ce(wks, div(0,z,:,:), resdiv)
>>>>>>
>>>>>> resvor@gsnCenterString = "Vorticity"
>>>>>> plot(1) = gsn_csm_contour_map_ce(wks, vor(0,z,:,:), resvor)
>>>>>>
>>>>>> resP@txString = "00Z 1 July 1979 "+ lev(z) + "-mb"
>>>>>> gsn_panel(wks,plot, (/2,1/), resP)
>>>>>>
>>>>>> end do
>>>>>>
>>>>>>
>>>>>> end
>>>>>>
>>>>>>
>>>>>>
>>>>>> When Running the code:
>>>>>>
>>>>>> See http://www.ncl.ucar.edu/ for more details.
>>>>>> (0) ESTABLISH CCSM FILES AND RETRIEVE DIMENSION INFO
>>>>>> (0) READ IN CCSM DATA
>>>>>> (0) COMPUTE AGEOSTROPHIC WINDS
>>>>>> (0) COMPUTE DIVERGENCE AND VORTICITY OF AGEOSTROPHIC WINDS
>>>>>> (0) 0
>>>>>> (0) 0
>>>>>> (0) Plotting level 30
>>>>>> warning:ContourPlotInitialize: no valid values in scalar field; ContourPlot not possible:[errno=1101]
>>>>>> warning:ContourPlotInitialize: no valid values in scalar field; ContourPlot not possible:[errno=1101]
>>>>>> (0) Plotting level 50
>>>>>> warning:ContourPlotInitialize: no valid values in scalar field; ContourPlot not possible:[errno=1101]
>>>>>> warning:ContourPlotInitialize: no valid values in scalar field; ContourPlot not possible:[errno=1101]
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> 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
>>>
>>> _______________________________________________
>>> 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

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Fri Feb 21 14:32:23 2014