Closed karlmsmith closed 6 years ago
Modified by @AnsleyManke on 25 Mar 2014 16:44 UTC
Comment by @AnsleyManke on 25 Mar 2014 16:44 UTC Should we really refuse to make a plot when coordinate data is missing? The CF standard says there cannot be missing coordinates, but there can be missing data in auxiliary coordinate variables.
Should we not just skip any missing coordinate data and go on with the plot, as if there were missing data in the variable being plotted?
Comment by @AnsleyManke on 26 Mar 2014 20:39 UTC See also ticket #2151, missing data in a Geopotential Height variable, and using a 3-argument shade command to view x-vs-height.
For 3-argument graphics commands, we can just skip coordinate data that is missing. Regridding ops may have other requirements.
Reported by @AnsleyManke on 28 Feb 2014 19:05 UTC This question, http://www.pmel.noaa.gov/maillists/tmap/ferret_users/fu_2014/msg00061.html
has an hdf5 dataset, readable through the NetCDF4 api. There is missing data in the curvilinear coordinate variables, so Ferret should just issue an error message. Instead it makes a garbled plot.
The test is done in curv_coord_range_sub.F. The bad-value flags for the coordinate fields have been saved as single precision values. For this data, the single-precision value of the missing-data flag does not match the missing double-precision coordinate data.
Migrated-From: http://dunkel.pmel.noaa.gov/trac/ferret/ticket/2147