Find samples deeper than the seafloor at their position

For every root sample (no parent) takes the deepest depth attributed to it — its own depth_max_m/depth_min_m, its descendants' and its observations' — and compares it with the deepest GEBCO cell in the 3x3 neighbourhood of its position plus tolerance_m. Positions outside the raster are unknown, not violations.

Usage

check_depth_vs_seafloor(
  con,
  seafloor,
  sample_tbl = "sample",
  obs_tbl = "obs",
  tolerance_m = 10,
  nominal_datasets = character()
)

Arguments

con

DBI connection holding sample_tbl (with parent_sample_key, root_sample_key) and optionally obs_tbl.

seafloor

Result of sample_seafloor() (or a GEBCO tif path, in which case it is computed).

sample_tbl, obs_tbl

Table names; obs_tbl may be absent.

tolerance_m

Metres a sample may exceed the neighbourhood-deepest cell before it is a finding (default 10).

nominal_datasets

dataset_keys whose depths are nominal maxima, measured apart from the rest (default none: every depth is in the main measure).

Value

A tibble of violators — sample_key, dataset_key, sample_type, cruise_key, longitude, latitude, depth_m, seafloor_depth_m, seafloor_max3x3_m, excess_m, on_land — worst first, with attributes summary (per-dataset n_root, n_unknown, n_over, max_excess_m, n_over_nominal) and nominal (the violators by a nominal depth alone, same columns).

Details

A dataset named in nominal_datasets publishes a NOMINAL depth: the maximum a gear could reach, not where it went (SWFSC ichthyoplankton Net.NetDepth, the maximum possible depth of the net). Where the sea floor is shallower the net simply did not get there, so that is not a finding of the same kind. Those datasets' sample and observation depths are left out of the main measure and measured on their own: attribute nominal (same columns as the result) and n_over_nominal in summary, so a release can ratchet each separately.