Declare category / variable / derivation / is_canonical / NERC ids on measurement types that already exist

Six descriptive columns, none of them an ingest's own definition:

• category — one of the registered categories (metadata/category.csv: Physical Oceanography, Nutrients & Chemistry, Carbonate System, Productivity & Pigments, Meteorology & Sea State, …), read by the CalCOFI Explorer's Browse tab (explorer UI plan D14).

• variable — the crosswalk that says which types measure the same thing comparably across datasets (temperature for the bottle's temperature and the CTD's temperature_ave), which the explorer carried in src/variables.ts as a stopgap.

• derivation — free text saying how a derived type was produced (the _cruise_corr vs _sta_corr distinction, or that a pre-QC r_* type is "interpolated to standard depth and carries no quality code by design").

• is_canonical — whether the type reaches the default obs/ctd_thin selection; a provider-confirmed fact like "the bottle's r_* series are interpolated, so they are not canonical" belongs here, not in an ingest's own literal.

• nerc_p01 — the NERC BODC Parameter Usage Vocabulary (P01) concept URI that a DwC/OBIS eMoF export emits as measurementTypeID.

• units_nerc_p06 — the NERC P06 unit concept URI, emitted as measurementUnitID.

Usage

declare_measurement_fields(
  fields,
  path,
  categories = NULL,
  overwrite = FALSE,
  quiet = FALSE
)

Arguments

fields

data.frame with measurement_type and at least one of category, variable, derivation, is_canonical, nerc_p01, units_nerc_p06 (declarable_measurement_fields()). NA leaves that field as it is.

path

path to metadata/measurement_type.csv

categories

the allowed category values — the category column of metadata/category.csv; NULL skips the check (not recommended)

overwrite

allow replacing a value that is already declared (default FALSE)

quiet

suppress the summary message

Value

The full updated registry, invisibly if nothing changed.

Details

The two vocabulary columns are validated against nerc_uri_prefixes(): a value must be a full concept URI in the right collection (.../collection/P01/current/<CODE>/). They are filled only on an exact vocabulary match — a concept every one of whose stated facets (quantity, matrix, phase, method) the registry or the dataset's documented protocol actually supplies. A generic concept is an exact match at coarser specificity (TEMPPR01, Temperature of the water body); a concept that adds a facet nobody recorded is not. So an empty cell means "no concept says exactly this", never "not looked at", and inventing one to fill the column is the same mistake as inventing a bound to quiet check_measurement_bounds().

Like declare_measurement_bounds() this changes only these columns, only on rows that already exist, refuses an unknown measurement_type, and writes with na = "". A registry predating a column gains it.

See also

build_coverage(), which puts category/variable onto coverage.json's variables[].