English

Problem: Tools fail when using netCDF files that have multibyte characters in the name

Description

Multibyte or Unicode characters, any non-ASCII characters, should not be used when naming netCDF files. Although Unicode characters are not accepted in the naming of netCDF files, they can be used within the attributes of netCDF files.

This is a known limitation that can be observed when running the ‘Input Space Time Cube’ with a multibyte (Unicode) character in the file name:

BUG-000092766 - ERROR 110015: Unable to read the Input Space Time Cube is thrown when Space Time Cube is created in folder that contains multibyte characters.

Multibyte characters: Wikipedia - List of Unicode Characters

Cause

The issue is with netCDF itself which is built by Unidata. We are waiting for a fix from Unidata.

Solution or Workaround

Esri Development has implemented a partial fix for the arcpy class NetCDFFileProperties, and the multidimensional tools.

If the netCDF file name contains Unicode characters, a temporary shortname is assigned and utilized, which is then replaced by the original name at the end.

This partial fix for file names containing Unicode characters lies only in the netCDF datasources, that is, anything that uses DataSourcesNetCDF (libs, dlls, olbs etc.) For the current netcdf version (netcdf-4.3.3.1), the fix does not work for any other combination of binaries. That includes netcdf4-python or anything that links directly into the core netcdf.lib.