You are using the variable "message_type" to separate between observations that are either GPS/position (G) or waves (W), thus acting as a mask if one wants e.g. only GPS positions.
char message_kind(trajectory, observation) ;
message_kind:long_name = "whether the current [trajectory, observation] contains GPS [kind G], or Waves [kind W], or None [kind N, in case of failed transmission] data." ;
This seems fine and understandable for a human reader, but I guess this is an "own invention", and am thus not sure if the file then strictly complies with the CF specifications for trajectories(?)
I.e. can the file be automatically parsed by a reader only knowing the CF specifications?
Yes, this may well be suboptimal... I get if I wanted to avoid this method that I would need several different "obs" time bases, i.e. an "obs_GPS", "obs_waves", etc? Do you think this would be better? If so, I am of course happy to change to a "several obs method".
@knutfrode
Yes, this may well be suboptimal... I get if I wanted to avoid this method that I would need several different "obs" time bases, i.e. an "obs_GPS", "obs_waves", etc? Do you think this would be better? If so, I am of course happy to change to a "several obs method".