Projection failing for NetCDF file using NetCDF4 in python

Question

I am trying to create a netCDF file from an excel sheet that contains precipitation data. The excel file is organized with latitude and longitude in columns: a column with latitudes, a column with longitudes both in EPSG:3006 - SWEREF99 TM. The subsequent columns contain the data for each day and I am reconstructing this into a netCDF file. The code works, but when opening the netCDF file the projection is completely wrong. I have followed the answer given in this post, but without getting the projection right (both when opening it in QGIS and running gdalinfo on the file).

nc_file = os.path.join(outpath, 'gridded_precip.nc')
nc = Dataset(nc_file, 'w', format='NETCDF4')
calendar = 'gregorian'
units = 'days since '+ str(start_date)
tim_dim = nc.createDimension('time', len(time))
lat_dim = nc.createDimension('lat', 9)
long_dim = nc.createDimension('lon', 11) 
time_var = nc.createVariable(varname='time', dimensions=('time',), datatype='float64')
time_var[:] = netCDF4.date2num(times, units=units, calendar=calendar)
latitude_var = nc.createVariable(varname='lat', dimensions=('lat',),datatype='f4')
latitude_var.units = 'meters'
latitude_var.axis = 'Y'
latitude_var[:] = lat.mean(axis=1)
longitude_var = nc.createVariable(varname='lon', dimensions=('lon',), datatype='f4')
longitude_var.units = 'meters'
longitude_var.axis = 'X'
longitude_var[:] = long.mean(axis=0)
precip_var = nc.createVariable('precipitation_mm_day','f4', ('time', 'lat', 'lon'),)
precip_var[:] = data
precip_var.units = 'mm/day'
precip_var.setncattr('grid_mapping', 'spatial_ref')
# Add crs to dataset
crs = nc.createVariable('spatial_ref', 'i4')
project_crs = 'PROJCS["SWEREF99 TM",GEOGCS["SWEREF99",DATUM["SWEREF99",SPHEROID["GRS 1980",6378137,298.257222101],TOWGS84[0,0,0,0,0,0,0]],PRIMEM["Greenwich",0,AUTHORITY["EPSG","8901"]],UNIT["degree",0.0174532925199433,AUTHORITY["EPSG","9122"]],AUTHORITY["EPSG","4619"]],PROJECTION["Transverse_Mercator"],PARAMETER["latitude_of_origin",0],PARAMETER["central_meridian",15],PARAMETER["scale_factor",0.9996],PARAMETER["false_easting",500000],PARAMETER["false_northing",0],UNIT["metre",1,AUTHORITY["EPSG","9001"]],AUTHORITY["EPSG","3006"]]'
crs.spatial_ref = project_crs
nc.close() 

Any ideas of what might be wrong?
A second but less important question: Both the longitude and latitude data are in 2D matrices, but I could only add the longitude and latitude data as vectors (the reason why a I am taking the average in the code)). Is there a way to have 2D latitude and longitude data in a netCDF?

• rschmunk: Create your data array on a grid with dimensions (time,y,x) and the lon and lat arrays on grids with dimensions (y,x). Add a coordinates attribute to the data array naming the lon and lat arrays. See section 5.2 of the CF Convention for an example (which has a level dimension rather than a time dim), cfconventions.org/Data/cf-conventions/cf-conventions-1.11/…
• Idalh: Thank you for the link. I have created my data on a grid of (time,y,x) and my long and lat values in a matrix with dimension (y,x). I do not understand though how you add a coordinate attribute, is this using nc.createVariable or nc.createDimension?
• rschmunk: I'm not a big Python user, but based on your code above, it should just be a matter of using the setncattr method to set attribute with name "coordinates" and value "lon lat" on the precipice variable.