latitude and latCell in latlon.nc (an output file from convert_mpas)

[jjang]

Hello.

I am seeking clarification regarding the difference between "latitude" (with latitude dimension) and "latCell" (with 'latitude, longitude' dimension) in the latlon.nc fie, generated by 'convert_mpas'.

I am currently remapping my 60-3 km variable mesh output to a 0.1-degree lat-lon grid using convert_mpas. I have plotted the difference between these two variables, calculated as latitude ( : ) - latCell ( : , 0 ) *180.0/(atan(1)*4.0), and the difference is approximately 0.02 degrees near the poles. (Please see the attached plot.) While this might not have a significant impact in most cases, could you please let me know why they differ and which variable would be more suitable?

Thank you,
Jihyeon

 

[mgduda]

The 'latitude' field in the latlon.nc file provides the target latitudes to which fields from the MPAS output file are interpolated, and the 'latCell' field in the latlon.nc file gives the cell latitudes from the Voronoi mesh interpolated to the target lat/lon locations. The difference you're seeing between 'latitude' and 'latCell' represent interpolation errors.

As to which field is more suitable, I think that would depend on what you're computing. Generally, if you're able to compute diagnostic quantities directly on the MPAS unstructured mesh, that may be preferable to computing those diagnostics using interpolated output.

 

[jjang]

Thank you so much for the clarification, Michael!