reading netcdf 4 | Community

Per the documentation FME supports up to version 3.6.2 of netCDF.

I suggest contacting support and requesting it and voting on the idea https://knowledge.safe.com/idea/28450/update-to-netcdf-format.html

@jdh is right, NetCDF 4 isnt't supported by FME and it's a known limitation.

Another, more peculiar limitation, is that FME cannot read NetCDF files >2GB on the Windows edition, although the Mac and Linux editions can, apparently.

It would be great if the NetCDF libraries could be upgraded soon.

@jdhI I am strongly considering making a workaround for NetCDf4 data using either nccopy.exe as a third party tool or this python script to save netCDf4 as netCDF_classic (v3) to then read it through.

I suggested it as my #fmeuc hackathon topic/group at the very least we should be able to wrap up a workaround process to make the data readable. As there is some netCDF4 data i would very much like to tackle with FME.

the python recipe i have found is here :
https://gist.github.com/guziy/8543562#file-converter-py

it will require numpy, netCDF4 dependencies to make it work. And you can expect up to 8x bloat of your netcdf4 data moving to netcdf_classic so i hope you can acheive <2gb limits in your input data. If you have opportunity to come to the User Conference let solve it together, and post something in the FMEHUB!

@jdhI I am strongly considering making a workaround for NetCDf4 data using either nccopy.exe as a third party tool or this python script to save netCDf4 as netCDF_classic (v3) to then read it through.

I suggested it as my #fmeuc hackathon topic/group at the very least we should be able to wrap up a workaround process to make the data readable. As there is some netCDF4 data i would very much like to tackle with FME.

the python recipe i have found is here :
https://gist.github.com/guziy/8543562#file-converter-py

it will require numpy, netCDF4 dependencies to make it work. And you can expect up to 8x bloat of your netcdf4 data moving to netcdf_classic so i hope you can acheive <2gb limits in your input data. If you have opportunity to come to the User Conference let solve it together, and post something in the FMEHUB!

Great idea, I'd be glad to join, just let me know.

A few years ago I already did something similar using the Python GDAL bindings to access the individual cell values in a classic NetCDF without having to coerce it into individual features first. The performance gains were pretty spectacular compared to using the RasterCellCoercer.

Any news regarding support for Netcdf4 format?

NetCDF4 is supported starting with FME 2018.

I have tried Sentinel 5p data with FME(R) 2018.1.0.0 (20180606 - Build 18493 - WIN64) and i get this error:

"NETCDF reader: Opening file 'C:/Temp/data/S5P_NRTI_L2__NO2.nc'

NETCDF reader: C:/Temp/data/S5P_NRTI_L2__NO2.nc is a netCDF file, but without any dimensions >= 2.

NETCDF reader: `C:/Temp/data/S5P_NRTI_L2__NO2.nc' exists but could not be opened. File contents may be invalid, or the file format may not be supported

NETCDF reader: Failed to open the dataset 'C:/Temp/data/S5P_NRTI_L2__NO2.nc'. Please ensure source data is valid and correct reader is selected

A fatal error has occurred. Check the logfile above for details

Failed to obtain any schemas from reader 'NETCDF' from 1 datasets. This may be due to invalid datasets or format accessibility issues due to licensing, dependencies, or module loading. See logfile for more information

FME Session Duration: 0.3 seconds. (CPU: 0.1s user, 0.0s system)

END - ProcessID: 12744, peak process memory usage: 48604 kB, current process memory usage: 43448 kB

Failed to obtain any schemas from reader 'NETCDF' from 1 datasets. This may be due to invalid datasets or format accessibility issues due to licensing, dependencies, or module loading. See logfile for more information

Program Terminating

Translation FAILED."

Any clue about fail?

Thanks in advance