Hi Peter, I looked through our test files for Mercator-projected datasets and found that they all name the x and y projection variables in their "coordinate" attributes, not lat and lon. So, I created an NcML file that changed all variables' "coordinate" attributes to "time level y x" and ran your program. It successfully identified the Mercator projection! The NcML file is attached. Make sure to change the netcdf@location attribute. But why can't NJ build the correct coordinate system from your file as it's written? As far as I can tell, it is CF-compliant, so it certainly *should* be able to. Looks like a bug. Do you mind if I create a (public) issue on GitHub for this, including your Java program and example file? Cheers, Christian Ticket Details =================== Ticket ID: YDE-831191 Department: Support netCDF Java Priority: Normal Status: Closed
Attachment:
2015_341_2208_n19_wj.ncml
Description: Binary data