Darwin Core Archive QC Report

Tampa Bay Interagency Seagrass Monitoring Program

Published

June 16, 2026

Overview

This report runs quality control checks on the Darwin Core Archive (DwC-A) produced by R/convert_to_dwc.R for the Tampa Bay Interagency Seagrass Monitoring Program. The archive consists of three files in dwc/:

File	Content
`event.csv`	One row per transect visit (parent event) and meter mark observation (child event)
`occurrence.csv`	One row per species per meter mark, plus one absence record per bare point
`emof.csv`	Quantitative measurements (cover, blade length, shoot density, epiphyte density, sediment type) linked to each occurrence

Checks use the obistools R package, which implements OBIS data quality requirements. Run R/convert_to_dwc.R first to generate the files in dwc/ before rendering this report.

Setup

Code

library(dplyr)
library(readr)
library(DT)
library(obistools)
library(here)

event      <- read_csv(here("dwc", "event.csv"),      show_col_types = FALSE)
occurrence <- read_csv(here("dwc", "occurrence.csv"), show_col_types = FALSE)
emof       <- read_csv(here("dwc", "emof.csv"),       show_col_types = FALSE)

Row counts for each output file.
File	Rows
event.csv	47346
occurrence.csv	56643
emof.csv	131524

Taxa inventory

The table below lists all taxa resolved to accepted WoRMS names in the presence occurrence records. The AphiaID is extracted from the scientificNameID field, which is stored as a Life Science Identifier (LSID) of the form urn:lsid:marinespecies.org:taxname:<id>. The count column shows the number of occurrence records referencing each taxon across all transect visits.

Code

occurrence |>
  filter(occurrenceStatus == "present") |>
  mutate(
    AphiaID = as.integer(sub(".*:", "", scientificNameID)),
    WoRMS   = paste0(
      '<a href="https://www.marinespecies.org/aphia.php?p=taxdetails&id=',
      AphiaID, '" target="_blank">', AphiaID, "</a>"
    )
  ) |>
  count(scientificName, taxonRank, WoRMS, name = "occurrences") |>
  arrange(scientificName) |>
  datatable(
    escape   = FALSE,
    rownames = FALSE,
    options  = list(pageLength = 25, dom = "ftp"),
    colnames = c("Scientific name", "Rank", "AphiaID (WoRMS link)", "Occurrences")
  )

Required fields

check_fields() verifies that all Darwin Core fields required by OBIS are present and non-empty. Required fields include eventDate, decimalLatitude, decimalLongitude, scientificName, scientificNameID, occurrenceStatus, and basisOfRecord. Recommended fields (minimumDepthInMeters, maximumDepthInMeters) are also checked at warning level.

Because the event core and occurrence extension split these fields across two files, the check is run on a flat table formed by joining occurrence records with their parent event rows.

Code

occ_flat <- occurrence |>
  inner_join(
    select(event, eventID, eventDate, decimalLatitude, decimalLongitude,
           minimumDepthInMeters, maximumDepthInMeters),
    by = "eventID"
  )

issues <- check_fields(occ_flat)

if (is.null(issues) || nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

No issues found.

Event ID integrity

check_eventids() verifies the internal consistency of the event hierarchy. It checks that eventID is present and unique across all rows, and that every parentEventID value references an eventID that exists in the same table. Orphaned child events — those whose parentEventID does not match any eventID — would prevent OBIS from assembling the transect-to-point hierarchy correctly.

In this archive, child events (meter mark visits, eventType = "Point") link to parent events (transect visits, eventType = "Transect") via parentEventID. Parent events have no parentEventID themselves.

Code

issues <- check_eventids(event)

if (is.null(issues) || nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

No issues found.

Extension linkage

These checks verify that the identifier fields in the extension files all reference valid rows in their respective core tables.

Occurrence to event

check_extension_eventids() confirms that every eventID in occurrence.csv matches an eventID in event.csv. Any unmatched row would indicate an orphaned occurrence record that OBIS cannot link to a location or date.

Code

issues <- check_extension_eventids(event, occurrence)

if (is.null(issues) || nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

No issues found.

eMoF to occurrence

check_extension_eventids() is designed specifically for eventID linkage and cannot be used to verify occurrenceID linkage. Instead, an anti_join() is used to find any occurrenceID values in emof.csv that have no matching row in occurrence.csv. Any unmatched row would produce a measurement record with no associated taxon.

Code

issues <- anti_join(emof, occurrence, by = "occurrenceID")

if (nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

No issues found.

Event dates

check_eventdate() validates that all eventDate values conform to ISO 8601 format. OBIS requires dates as YYYY-MM-DD, YYYY-MM-DDTHH:MM:SS, or date ranges using / as a separator (e.g., 2022-06-01/2022-06-03). Invalid or missing dates prevent records from appearing in OBIS temporal queries and are treated as errors during OBIS ingestion.

Code

issues <- check_eventdate(event)

if (is.null(issues) || nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

No issues found.

Depth QC

check_depth() flags records where reported depths are inconsistent with known bathymetry. It queries the OBIS xylookup service using the record coordinates and compares minimumDepthInMeters / maximumDepthInMeters against the modelled seabed depth at each point. A depthmargin of 20 m accounts for tidal variation, datum offsets, and measurement uncertainty. Records at coordinates landward of the shoreline are also flagged as likely coordinate errors.

Only child events carry depth values; parent events intentionally have NA for depth fields and are excluded before running the check.

Code

child_events <- event |>
  filter(!is.na(minimumDepthInMeters))

issues <- check_depth(child_events, report = TRUE, depthmargin = 20)

if (is.null(issues) || nrow(issues) == 0) {
  cat("No issues found.\n")
} else {
  datatable(issues, rownames = FALSE, options = list(pageLength = 15))
}

Spatial distribution

plot_map() renders a static map of all event coordinates zoomed to the extent of the data. This provides a visual sanity check that points fall within Tampa Bay and that no coordinates are obviously misplaced (e.g., on land, in the wrong bay segment, or at 0, 0).

Code

plot_map(event, zoom = TRUE)

Taxon name matching

match_taxa() cross-checks scientific names against WoRMS using the wm_records_taxamatch API endpoint, which applies fuzzy phonetic matching and returns multiple candidate matches ranked by similarity score. When more than one accepted WoRMS record matches a name, the function prompts interactively for the user to select the correct one.

This check is independent of the WoRMS lookup in convert_to_dwc.R, which uses wm_records_names(). It provides a second opinion that the resolved names are unambiguous in WoRMS. Because it requires interactive input it is not run during rendering — to use it, run the following in an interactive R session after loading the occurrence data:

Code

match_taxa(unique(occurrence$scientificName))

Names flagged as ambiguous by this function should be reviewed against the aphia_overrides vector in convert_to_dwc.R to confirm the correct AphiaID is pinned for each genus.