[ncl-talk] Error writing strings to netcdf

Mon May 7 13:37:49 MDT 2018

Andrew,

If wind_dir_top_string is an array of strings, then an attribute is not the
right solution.  To avoid the "ncdimdef" error, try adding explicit
dimension names to the array, before writing to the Netcdf file, like this:

    wnd_dir_top_string2 = tochar(wnd_dir_top_string)
    wnd_dir_top_string2!0 = "dim0"
    wnd_dir_top_string2!1 = "dim1"
    wnd_dir_top_string2!2 = "dim2"
    p->wnd_dir_top_string = wnd_dir_top_string2

Change the names dim0, dim1, dim2 to names that are descriptive for your
application.  Dimension names may be shared between arrays in the Netcdf
file, as long as the dimension size is always the same for each dimension
name.  Please see this reference for the use of dimension names:

http://www.ncl.ucar.edu/Document/Manuals/Ref_Manual/NclVariables.shtml#Dimensions

I can't explain your original "ncdimdef" error.  If you want more help with
that, then send me your complete NCL script and any input files that it
needs, by private e-mail, and I will take a quick look at it.

--Dave

On Sat, May 5, 2018 at 3:30 PM, Andrew Kren - NOAA Affiliate <
andrew.kren at noaa.gov> wrote:

> Dave,
>
> I originally did not have the constructors, and it didn't work when I
> didn't have them. I just tried them to see if it would work but it didn't.
>
> The strings I actually need for reading in a separate program. So the
> wind_dir_top_string is an array of wind directions (N, S, E, W) so if I
> saved it as an attribute would I still be able to access the array?
>
> Here is the output from the ncl_filedump of the file.
>
> Variable: f
>
> Type: file
>
> filename: saved_foothills_variables_nam
>
> path: saved_foothills_variables_nam.nc
>
>    file global attributes:
>
>       title : Saved Foothills Variables for comparison page
>
>       source : /home/arsenic1/scripts/nam/00z/saved_foothills_variables_
> nam.nc
>
>       creation_date : Sat May  5 15:26:37 MDT 2018
>
>    dimensions:
>
>       time = 0  // unlimited
>
>       ncl0 = 26
>
>       ncl1 = 53
>
>       ncl4 = 26
>
>       ncl5 = 53
>
>       ncl17 = 26
>
>       ncl18 = 53
>
>       ncl29 = 26
>
>       ncl30 = 53
>
>       ncl15 = 26
>
>       ncl16 = 53
>
>       ncl13 = 26
>
>       ncl14 = 53
>
>       ncl11 = 26
>
>       ncl12 = 53
>
>    variables:
>
>       float temperature_top ( ncl0, ncl1 )
>
>          _FillValue : 9.96921e+36
>
>
>       float wnd_spd_top ( ncl4, ncl5 )
>
>          _FillValue : 9.96921e+36
>
>
>       float clouds ( ncl17, ncl18 )
>
>          _FillValue : 9.96921e+36
>
>
>       float lightning ( ncl29, ncl30 )
>
>          _FillValue : 9.96921e+36
>
>
>       float precip_chance ( ncl15, ncl16 )
>
>          _FillValue : 9.96921e+36
>
>
>       float precip_type_top ( ncl13, ncl14 )
>
>          _FillValue : 9.96921e+36
>
>
>       float precip_amount ( ncl11, ncl12 )
>
>          _FillValue : 9.96921e+36
>
>
> On Fri, May 4, 2018 at 1:23 PM, Dave Allured - NOAA Affiliate <
> dave.allured at noaa.gov> wrote:
>
>> Andrew,
>>
>> The constructors (/ /) around the tochar function should be unnecessary
>> in that context.  Try removing them.
>>
>> A workaround is to write the string as an attribute instead of a data
>> variable.  This is the more usual way of storing single descriptive
>> strings.  This should completely avoid the mysterious ncdimdef error, and
>> the code is more straightforward.  Will this work for your application?
>>
>>     p at wnd_dir_top_string = wnd_dir_top_string
>>
>> If you need to debug the original statement, then more information is
>> needed.  Insert an exit statement right before line 806, and show us the
>> full output from ncdump -hs on the vestigial output file.
>>
>> --Dave
>>
>>
>> On Fri, May 4, 2018 at 11:19 AM, Andrew Kren - NOAA Affiliate <
>> andrew.kren at noaa.gov> wrote:
>>
>>> Hi,
>>>
>>> In my previous programs I have used the tochar function to convert
>>> strings to characters for writing string data to netcdf using NCL. However,
>>> for some reason, its not working for me this time. It fails with this error:
>>>
>>> ncdimdef: ncid 65536: NetCDF: String match to name in use
>>>
>>> fatal:["Execute.c":8575]:Execute: Error occurred at or near line 806 in
>>> file nam_foothills.ncl
>>>
>>> My section to write the netcdf file is here:
>>>
>>> filo = "/home/arsenic1/scripts/nam/00z/saved_foothills_variables_nam.nc"
>>>
>>> system("rm -rf "+filo) ; remove any pre-existing file
>>>
>>> p = addfile(filo,"c")
>>>
>>> p at title = "Saved Foothills Variables for comparison page"
>>>
>>> p at source = filo
>>>
>>> p at creation_date = systemfunc("date")
>>>
>>> filedimdef(p,"time",-1,True) ; make time unlimited dimension
>>>
>>>
>>> p->temperature_top = temperature_top
>>>
>>> p->wnd_spd_top = wnd_spd_top
>>>
>>> p->clouds = clouds
>>>
>>> p->lightning = lightning
>>>
>>> p->precip_chance = precip_chance
>>>
>>> p->precip_type_top = precip_type_top
>>>
>>> p->precip_amount = precip_6hr_top
>>>
>>> p->wnd_dir_top_string = (/tochar(wnd_dir_top_string)/)
>>>
>>> p->day_of_week_string = (/tochar(day_of_week_string)/)
>>>
>>> p->months = (/tochar(months)/)
>>>
>>> p->valid_day = (/tochar(valid_day)/)
>>>
>>> Line 806 is the first line where it tries to write the wind direction
>>> string:
>>>
>>> p->wnd_dir_top_string = (/tochar(wnd_dir_top_string)/)
>>>
>>> Is there any remedy to this? I've looked online, but have not found much
>>> help.
>>>
>>> Thanks,
>>>
>>> --
>>> Andrew Kren
>>> Assistant Scientist
>>> CIMAS & NOAA/AOML
>>> 325 Broadway, Boulder, CO 80305
>>> <https://maps.google.com/?q=325+Broadway,+Boulder,+CO+80305&entry=gmail&source=g>
>>> (303) 497-5418
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20180507/e4042f39/attachment.html>