Skip to contents

Why two measurement keys sharing a NERC P01 concept are kept apart

Source: R/catalog_measurements.R
measurement_related_reasons.Rd

P01 identity says two series name the same quantity. It does not say they may be pooled, and related[] is where the record states the difference the crosswalk rule (plan 2026-09-10 § D3) refused to merge across:

Usage

measurement_related_reasons()

Value

A character vector, the related[].why enum.

Details

  • underway_vs_cast — one side is an underway intake, the other a cast (sst_c beside temperature). Different sampling entirely.

  • same_bottles — one side is a btl_* / *_btl series: another dataset's files carrying their own copy of the bottle table. Plausibly the same physical bottles, so merging them would double count.

  • replicate_vs_mean — one side is a …_rep replicate, the other a mean (alkalinity_rep1 beside alkalinity).

  • pre_qc_twin — one side is an r_* series reported before quality control.

  • sensor_vs_mean — the two share a dataset and exactly one side's registry derivation starts with "Mean of": a raw sensor beside the mean of the sensor pair (oxygen_ml_l_1 beside oxygen_ml_l_ave_sta_corr, which is the oxygen_ml_l key). One is an input to the other.

  • paired_sensors — the two sensors of one instrument: the same dataset and the same P01, and neither side is the mean (oxygen_ml_l_1 beside oxygen_ml_l_2). Two readings of one water sample, not two measurements.

  • same_casts — the residual same-quantity case: different datasets, neither underway, and no bottle-table, replicate, pre-QC or sensor marker to name a sharper difference. Plausibly the same water sampled on the same casts, and kept apart for that reason — the DIC package's salinity_pss78, which is the CTD salinity of the DIC casts, beside the unified salinity.

They are tested in that order, because the first that applies is the coarsest difference: an underway intake beside a cast's bottle table is underway_vs_cast, not same_bottles.

A pair sharing a P01 that matches none of the seven gets no related[] entry: the vocabulary states no reason, and the record never invents one.

Examples

measurement_related_reasons()
#> [1] "underway_vs_cast"  "same_bottles"      "replicate_vs_mean"
#> [4] "pre_qc_twin"       "sensor_vs_mean"    "paired_sensors"   
#> [7] "same_casts"