Re: FillValue attributes

The fatal error you are getting is with the 'levs' variable.
It does not have a _FillValue attribute. A solution is

   levs = f4->lv_ISBL1
   levs@_FillValue = -999

For several grib files I have available for testing,
'relhum' works correctly. I did find a grib file where
it fails as you reported.

However, even with the above 'solution', I got values ranging from
0 to 189 which are bad.

There may be an interface error between the NCL and underlying
fortran code.

More testing will be done today.

On 11/11/13 7:59 AM, Kumar, Anil (GSFC-617.0)[UNIV OF MARYLAND] wrote:
> Phani and Denis,
> Yes I also found similar problem recently when same script used with two
> different ncl versions.
> Thanks to point out this bug and fix.
> Anil
>
>
>
> On 11/11/13 9:33 AM, "R Phani" <rphani@tropmet.res.in> wrote:
>
>> Dear Dennis,
>>
>> Thank you for finding out the bug. Initially, I thought my data was
>> wrong.
>>
>> Regards,
>> Phani.
>>
>>
>> ----- Original Message -----
>> From: "Dennis Shea" <shea@ucar.edu>
>> To: "R Phani" <rphani@tropmet.res.in>
>> Cc: "NCL" <ncl-talk@ucar.edu>
>> Sent: Monday, November 11, 2013 9:35:02 AM
>> Subject: Re: FillValue attributes
>>
>> After trying several local files, I have managed to reproduce
>> the error. A fix will likely appear in the 6.2.0 release.
>>
>> On 11/9/13 7:59 PM, R Phani wrote:
>>> Dear Dennis,
>>>
>>> "The error message is likely bogus."
>>> I don't think so, I have solved the error by making the integer to
>>> float with 'round' and it worked.
>>>
>>> Anyway, my concern was that it gives different values (max value) for
>>> different versions! (5.1.1 is realistic) with the script being same.
>>>
>>> Phani
>>> -------------------------------------
>>> bash-3.2$ ncl rel_humid_denis.ncl
>>> Copyright (C) 1995-2009 - All Rights Reserved
>>> University Corporation for Atmospheric Research
>>> NCAR Command Language Version 5.1.1
>>> The use of this software is governed by a License Agreement.
>>> See http://www.ncl.ucar.edu/ for more details.
>>>
>>>
>>> Variable: rh
>>> Type: float
>>> Total Size: 1227264 bytes
>>> 306816 values
>>> Number of Dimensions: 3
>>> Dimensions and sizes: [lv_ISBL1 | 17] x [g0_lat_2 | 94] x [g0_lon_3 |
>>> 192]
>>> Coordinates:
>>> lv_ISBL1: [10..1000]
>>> g0_lat_2: [90..-90]
>>> g0_lon_3: [ 0..358.125]
>>> Number Of Attributes: 3
>>> units : %
>>> long_name : relative humidity
>>> _FillValue : 1e+20
>>> (0) rh: min=0 max=100
>>> bash-3.2$ /gpfs1/scientific_appln/ncl-6.0.0/bin/ncl rel_humid_denis.ncl
>>> Copyright (C) 1995-2011 - All Rights Reserved
>>> University Corporation for Atmospheric Research
>>> NCAR Command Language Version 6.0.0-beta
>>> The use of this software is governed by a License Agreement.
>>> See http://www.ncl.ucar.edu/ for more details.
>>>
>>> Variable: rh
>>> Type: float
>>> Total Size: 1227264 bytes
>>> 306816 values
>>> Number of Dimensions: 3
>>> Dimensions and sizes: [lv_ISBL1 | 17] x [g0_lat_2 | 94] x [g0_lon_3 |
>>> 192]
>>> Coordinates:
>>> lv_ISBL1: [10..1000]
>>> g0_lat_2: [90..-90]
>>> g0_lon_3: [ 0..358.125]
>>> Number Of Attributes: 3
>>> units : %
>>> long_name : relative humidity
>>> _FillValue : 1e+20
>>> (0) rh: min=0 max=1996.69
>>> ==============================================
>>>
>>> ----- Original Message -----
>>> From: "Dennis Shea" <shea@ucar.edu>
>>> To: "R Phani" <rphani@tropmet.res.in>
>>> Cc: "NCL" <ncl-talk@ucar.edu>
>>> Sent: Saturday, November 9, 2013 7:10:25 PM
>>> Subject: Re: FillValue attributes
>>>
>>> None of your GRIB files were complete.
>>> --------------------------------------
>>> The error message is likely bogus.
>>>
>>> If the variables are 4D, the following is *incorrect*
>>>
>>> rh = relhum (t_spcfs, mr_spcfs, conform(t_spcfs, levs*100, 0))
>>>
>>> Please read the Mini-Language Manual about array numbers.
>>>
>>> If 4D, it should be
>>>
>>> rh = relhum (t_spcfs, mr_spcfs, conform(t_spcfs, levs*100, 1))
>>>
>>> The ordering is
>>>
>>> (time,lev,lat,lon) => (0,1,2,3)
>>>
>>> ====
>>> Please try the attached.
>>>
>>> Also, please do not send data unless we ask for it.
>>>
>>>
>>>
>>> On 11/9/13 9:18 AM, R Phani wrote:
>>>> Dear Dennis,
>>>>
>>>> Thank you for your help and making my code efficient.
>>>>
>>>> 1) I have tried on both 5.1.1 and 6.0.0-beta versions
>>>> 2) Now, I have allocated the memory and modified following your
>>>> suggestions.
>>>>
>>>> Still I am getting the same error.
>>>>
>>>> fatal:relhum: If one _FillValue attribute is set, then all of them
>>>> must be set.
>>>> fatal:Execute: Error occurred at or near line 45 in file rel_humid.ncl
>>>>
>>>> Is it like, when I use 'conform', the attributes also should be set? I
>>>> have uploaded the data and the script.
>>>>
>>>> Phani
>>>>
>>>>
>>>> ----- Original Message -----
>>>> From: "Dennis Shea" <shea@ucar.edu>
>>>> To: "R Phani" <rphani@tropmet.res.in>, "NCL" <ncl-talk@ucar.edu>
>>>> Sent: Saturday, November 9, 2013 9:47:47 AM
>>>> Subject: Re: FillValue attributes
>>>>
>>>> [1] What version of NCL are you using?
>>>>
>>>> %> ncl -V
>>>>
>>>> [2] Where have you allocated memory for 'rh'. In your code,
>>>> I would have expected to see:
>>>>
>>>> rh = new (dimsizes(temp_cfs), typeof(temp_cfs))
>>>>
>>>> [3] Your code could be made more efficient.
>>>>
>>>> [a] Whenever an entire variable is being read don't use (:,:,:,:).
>>>> NCL accesses the data more efficiently without the (:,:,:,:).
>>>> In practice, you will likely not notice any speed difference.
>>>>
>>>> temp_spcfs = f1->TMP_GDS0_ISBL_10 ; (:,:,:,:)
>>>> sp_spcfs = f2->SPF_H_GDS0_ISBL_10
>>>> temp_cfs = f3->TMP_GDS0_ISBL_10
>>>> sp_cfs = f4->SPF_H_GDS0_ISBL_10
>>>>
>>>> [b] NCL is an interpreted language. Generally, it is better to
>>>> avoid 'do' loops. NCL has 2 functions which can be used to
>>>> replace your do loop
>>>>
>>>> http://www.ncl.ucar.edu/Document/Functions/Built-in/conform.shtml
>>>> http://www.ncl.ucar.edu/Document/Functions/Built-in/conform_dims.shtml
>>>>
>>>> Replace
>>>> rh = new(dimsizes(temp_cfs),typeof(temp_cfs)) ; added for
>>>> correctness
>>>>
>>>> do k = 1,nlevs
>>>> do j = 1, nlon
>>>> do i = 1,nlat
>>>> pPa = levs(k)*100
>>>> rh(i, j, k) = relhum (t_spcfs(i,j,k), sph_spcfs(i,j,k),
>>>> pPa)
>>>> end do
>>>> end do
>>>> end do
>>>>
>>>> With
>>>>
>>>> rh = relhum (t_spcfs, sph_spcfs, conform(t_spcfs, levs*100, 1)
>>>>
>>>> There is no need to preallocate space for 'rh'. rh will be created
>>>> automatically.
>>>>
>>>> ------------
>>>>
>>>> If you are new to NCL, please read the Mini-Language Manual at:
>>>>
>>>> http://www.ncl.ucar.edu/Document/Manuals/
>>>>
>>>>
>>>>
>>>> On 11/9/13 6:59 AM, R Phani wrote:
>>>>> Hi NCL users,
>>>>>
>>>>> I am getting the following error.
>>>>>
>>>>> fatal:relhum: If one _FillValue attribute is set, then all of them
>>>>> must be set.
>>>>> fatal:Execute: Error occurred at or near line 45 in file rel_humid.ncl
>>>>>
>>>>> What could be the problem in the script?
>>>>>
>>>>> Phani
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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 Mon Nov 11 08:05:56 2013