Write the CF Discrete-Sampling-Geometry attributes — nc_profile

These attributes are what make the file a CF profile, trajectory or point dataset rather than a table that happens to be stored in netCDF: cf_role marks the instance identifier and sample_dimension on rowSize declares the contiguous ragged array. Without them a CF-aware reader sees unrelated dimensions.

Usage

nc_profile_atts(
  nc,
  obs_types,
  var_meta = list(),
  profile_vars = character(),
  profile_id_var = "profile_id",
  feature_type = c("profile", "trajectory", "point"),
  obs_cols = "depth"
)

Arguments

nc: Open ncdf4 handle.
obs_types: Obs-level measurement variable names.
var_meta: Named list from measurement_var_meta().
profile_vars: Instance-level variable names present in the file; the coordinate ones among them get their standard_name/axis.
profile_id_var: The instance-identifier variable, or NULL for a point collection, which has no instances.
feature_type: "profile", "trajectory" or "point". Sets the cf_role of profile_id_var (profile_id vs trajectory_id) and, for "point", skips the ragged-array attributes entirely.
obs_cols: Coordinate columns on the observation dimension.

Value

TRUE, invisibly.