[ncl-talk] Writing attributes to netCDF file

[Dave Allured - NOAA Affiliate]

Laura, it looks like ESMF_regrid helpfully attached lat2d and lon2d
attributes directly to your data variable.  This is useful inside a single
NCL script for generating plots.  However, this is not suitable for output
to a Netcdf file that follows normal CF Conventions.  It also looks like
the same 2-D coordinates are duplicated in independent variables g5_lat_0
and g5_lon_1.  You can make these two the primary 2-D coordinate variables
within the Netcdf file, and remove unwanted lat2d and lon2d attributes,
like this.  Presumably you already have lines 4, 5, and 6 or their
equivalent, so just add these first three lines.

    delete (diag_r at lat2d)
    delete (diag_r at lon2d)
    diag_r at coordinates = "g5_lat_0 g5_lon_1"

    fout->g5_lat_0 = g5_lat_0
    fout->g5_lon_1 = g5_lon_1
    fout->precip = diag_r


On Mon, Aug 30, 2021 at 2:36 PM Laura Fowler via ncl-talk <
ncl-talk at mailman.ucar.edu> wrote:

> Hello:
>
> I am trying to write a variable to a netCDF file. This variable was
> initially re-gridded using ESMF_regrid_with_weights so that a
> printVarSummary of that variable is:
>
> Variable: diag_r
>
> Type: float
>
> Total Size: 3950404 bytes
>
>             987601 values
>
> Number of Dimensions: 2
>
> Dimensions and sizes: [g5_x_0 | 881] x [g5_y_1 | 1121]
>
> Coordinates:
>
> Number Of Attributes: 7
>
>   _FillValue : 9.96921e+36
>
>   units : mm
>
>   long_name : time-step convective precipitation
>
>   lat2d : <ARRAY of 987601 elements>
>
>   lon2d : <ARRAY of 987601 elements>
>
>   remap : remapped via ESMF_regrid_with_weights: Bilinear
>
>   missing_value : 9.96921e+36
>
>
>
> where lat2d and lon2d are the latitudes and longitudes of the NCEP Stage
> IV precipitation data. When I write the re-gridded variable to a netCDF
> file, then I get the following output of ncdump :
>
>
> [laura at casper-login2 6488066.CONUS]$ ncdump -h
> CONUS_mpas.precipitation.init_20120611_06.nc | more
>
> netcdf CONUS_mpas.precipitation.init_20120611_06 {
>
> dimensions:
>
> StrLen = 19 ;
>
> Time = 49 ;
>
> g5_x_0 = 881 ;
>
> g5_y_1 = 1121 ;
>
> variables:
>
> char xtime(Time, StrLen) ;
>
> string xtime:units = "YYYY-MM-DD_hh:mm:ss" ;
>
> string xtime:long_name = "Model valid time" ;
>
> float g5_lat_0(g5_x_0, g5_y_1) ;
>
> g5_lat_0:La1 = 23.117f ;
>
> g5_lat_0:Lo1 = -119.023f ;
>
> g5_lat_0:Lov = -105.f ;
>
> g5_lat_0:Dx = 4763.f ;
>
> g5_lat_0:Dy = 4763.f ;
>
> string g5_lat_0:ProjectionCenter = "north" ;
>
> string g5_lat_0:units = "degrees_north" ;
>
> string g5_lat_0:GridType = "Polar Stereographic Projection Grid" ;
>
> string g5_lat_0:long_name = "latitude" ;
>
> g5_lat_0:corners = 23.117f, 19.80426f, 45.61898f, 53.51062f ;
>
> float g5_lon_1(g5_x_0, g5_y_1) ;
>
> g5_lon_1:La1 = 23.117f ;
>
> g5_lon_1:Lo1 = -119.023f ;
>
> g5_lon_1:Lov = -105.f ;
>
> g5_lon_1:Dx = 4763.f ;
>
> g5_lon_1:Dy = 4763.f ;
>
> string g5_lon_1:ProjectionCenter = "north" ;
>
> string g5_lon_1:units = "degrees_east" ;
>
> string g5_lon_1:GridType = "Polar Stereographic Projection Grid" ;
>
> string g5_lon_1:long_name = "longitude" ;
>
> g5_lon_1:corners = -119.023f, -80.74809f, -59.95461f, -134.0418f ;
>
> float precip(g5_x_0, g5_y_1) ;
>
> precip:_FillValue = 9.96921e+36f ;
>
> string precip:units = "mm" ;
>
> string precip:long_name = "time-step convective precipitation" ;
>
> precip:lat2d = 23.1169986724854, 23.12473487854, 23.132453918457,
> 23.1401557922363, 23.1478404998779, 23.1555061340332, 23.1
>
> 631546020508, 23.1707859039307, 23.1784000396729, 23.1859951019287,
> 23.1935729980469, 23.2011337280273, 23.2086753845215, 23.2162017822266,
>
> 23.2237091064453, 23.2311973571777, 23.2386703491211, 23.2461242675781,
> 23.2535591125488, 23.2609786987305, 23.2683792114258, 23.27576255798
>
>
>
> which shows that, by default, it explicitly outputs the lat2d and lon2d
> attributes. Is there a way that ncdump only outputs the actual values of
> the re-gridded variables without info about the lat2d and lon2d attributes,
> like a "normal" netCDF variable?
>
>
>
> Thanks,
>
> Laura
>
> !-------------------------------------------------------------------------------------------------------------
> Laura D. Fowler
>
> Mesoscale and Microscale Meteorology Division (MMM)
> National Center for Atmospheric Research
> P.O. Box 3000, Boulder CO 80307-3000
>
> e-mail: laura at ucar.edu
> phone: 303-497-1628
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20210830/9d139ebb/attachment.html>