Skip to contents

The flags[] a measurements.json series can carry

Source: R/catalog_measurements.R
measurement_series_flags.Rd

A series[] entry is one measurement_type × dataset of obs_env — what a dataset calls a measurement. A flag says something about that series a reader of the page needs in order to read its numbers correctly, so a measurement page can show a quiet pill rather than a footnote nobody writes:

Usage

measurement_series_flags()

Value

A character vector, the series[].flags enum of measurements.schema.json.

Details

  • sensor_mean — the registry's derivation says the series is a mean of a sensor pair (temperature_ave, "Mean of the two temperature sensors"). The per-sensor series ride the full-resolution product.

  • replicate — a …_rep1 / …_rep2 replicate beside a mean of the same quantity (alkalinity_rep1), never itself the mean.

  • reported_pre_qc — an r_* series: the value the source reported before quality control, kept beside the QC'd one.

  • no_bound — the registry declares neither valid_min nor valid_max, so nothing in the pipeline can call one of its values impossible.

  • sentinel_suspected — a value the series' own registry row says is impossible, or one that looks like a fill where nothing is declared. Precisely: out_of_bounds$n > 0 (at least one value outside the declared valid_min / valid_max), or — with no bound declared at all — an observed maximum both at or above CC_MEASUREMENT_SENTINEL (99, the shape a 9-fill takes) and more than CC_MEASUREMENT_SENTINEL_RATIO (100) times the series' own 95th percentile. Both halves of the second test are needed: PAR reads 14,187 uE/m2/s and short-wave radiation 1,456 W/m2 legitimately, while the METS sst_c of v2026.09.06 reads 9,895 degC against a 95th percentile of 20.4. The minimum is tested the same way — at or below -99 and more than 100x the 5th percentile in magnitude — which is what catches the CTD spar of -3.07e17.

  • no_flag_at_grain — the registry names no _qual_column, so the series reaches the release carrying no quality code. The CTD's temperature_ave is the headline case: the sensor flags ride temperature_1 / temperature_2 in obs_ctd_full.

  • no_p01 — no nerc_p01 concept. Under the registry's exact-match rule that means "no concept says exactly this", never "not looked at".

Examples

measurement_series_flags()
#> [1] "sensor_mean"        "replicate"          "reported_pre_qc"   
#> [4] "no_bound"           "sentinel_suspected" "no_flag_at_grain"  
#> [7] "no_p01"