Build the Darwin Core Event core for one dataset

From sample's adjacency list — the same rows every consumer reads — plus the cruise reference as the root event, plus the effort denominator from sample_measurement:

Usage

dwc_event(
  con,
  dataset_key,
  gear = NULL,
  measurement_type = NULL,
  cruises = TRUE,
  close_tree = TRUE
)

Arguments

con

a DBI connection to the release

dataset_key

the dataset

gear

the registry from read_gear_registry(), or NULL (no samplingProtocol)

measurement_type

the registry from read_measurement_type(), or NULL (sampleSizeUnit is then empty)

cruises

emit a cruise root event per cruise_key and parent the roots to it

close_tree

pull in an ancestor event that belongs to ANOTHER dataset, so the archive's parentEventIDs all resolve. sample_key is globally unique and a parent_sample_key may point across datasets — cdfw_dungeness-crab parents 306 examined subsamples onto swfsc_ichthyo site occupations, and calcofi_dic parents 6 bottles onto calcofi_bottle casts (that is how the DIC/bottle dedup works). In the release those are not orphans; in a single-dataset archive they would be, so the ancestors travel with their own datasetID rather than the pointer being dropped.

Value

A data frame of Event-core rows, all-NA columns dropped.

Details

Darwin Core	core
eventID	sample.sample_key
parentEventID	sample.parent_sample_key, or the row's cruise_key for a root
eventType	sample.sample_type ("cruise" for a cruise event)
eventDate	sample.datetime as ISO 8601 UTC; a cruise's date_min/date_max span
decimalLatitude / decimalLongitude	sample.latitude / sample.longitude
minimumDepthInMeters / maximumDepthInMeters	sample.depth_min_m / depth_max_m
locationID	sample.site_key
samplingProtocol	gear.csv dwc_samplingProtocol for sample.tow_type
sampleSizeValue / sampleSizeUnit	sample_measurement volume_sampled + its registry unit
geodeticDatum	"WGS84" — the release's own CRS
datasetID	dataset_key

Cruise events. A dataset's root samples carry cruise_key but no parent, so cruises = TRUE (the default) emits one event per cruise the dataset touches and parents the roots to it. That is a derivation from a column already on the row, not an invention, and it is what makes an archive's events group the way a reader expects. cruises = FALSE leaves the roots parentless.

Nothing here asserts countryCode, waterBody or coordinateUncertaintyInMeters: the release measures none of them, and the ichthyo notebook's hand-typed values were dataset metadata living where no provider could edit them (D-8).

See also

dwc_occurrence(), dwc_emof(), dwc_archive()