I think this question would be best directed to GFDL, who made the file.
You can't always trust files to follow their declared missing_value and
FillValue. If you know that -999000000.000000 means a value is missing,
you should treat it as such, regardless of what is declared in the file.
That would be another bug to report to GFDL.
-- Elizabeth
On Thu, Jan 21, 2016 at 1:25 AM, Sujata Mandke <sujatamandke@xxxxxxxxx>
wrote:
> Dear Group
> I have following problem in reading
> netcdf file. This data file is downloaded
> from CMIP5 project. It is daily precipitation
> of GFDL HIRAM180 model simulation
> for the period 1979-2008. The problem is
> as follows
> ----------------------------------------------
> 1. File-"pr_day_GFDL-HIRAM-C180_amip_r1i1p1_19790101-20081231.nc"
> is opened in grads using "sfdopen"
> 2. When "pr" (which is in units
> of (kg m2 s-2 ) is plotted, it is found that range is incorrect.
>
> 3. Range of "pr" after multiplying by 86400.0
> (to convert to units of mm/day)
> is one oreder of magnitude higher than expected.
> as shown below.
> ga-> d pr
> Contouring: 0 to 0.0035 interval 0.0005
> ga-> d pr*86400.0
> Contouring: 0 to 330 interval 30
>
> 4. in grads >q undef gives
> Output undef value is set to -999000000.000000
>
> Whereas header of netcdf file obtained using "ncdump"-
> "pr_day_GFDL-HIRAM-C180_amip_r1i1p1_19790101-20081231.nc"
> shows undefine as
> --------------------------------
> pr:missing_value = 1.e+20f ;
> pr:_FillValue = 1.e+20f ;
> --------------------------------
> 5. Also shaded plot shows value of
> variable "pr" less than zero over
> some regions. (precipitation can
> not be less than zero).
> --------------------------------------
> --
> Dr. Sujata Mandke
> Scientist, Indian Institute of Tropical Metetorology
> PUNE-INDIA.
>
> _______________________________________________
> netcdfgroup mailing list
> netcdfgroup@xxxxxxxxxxxxxxxx
> For list information or to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
