Skip to content

Argo float profiles around Hawaii

A real research prompt — and a valuable lesson in catalog interpretation.

What this teaches

  • The GADR collection indexes Argo floats at the float level (one item per float, lifetime bbox), not per-profile
  • Why a Hawaii bbox query can return Southern Ocean floats — and how to spot the mismatch using column_stats_summary
  • When to escalate from AQUAVIEW's discovery layer to the underlying GDAC's spatial index

Source used: GADR — Global Argo Data Repository, NOAA NCEI

Prompt

Find Argo float profiles within 200 km of Hawaii from March 2026. I want
profiles where the float reached at least 1500 dbar pressure (deep dives).
Summarize the temperature range observed and give me a few download links.

Transcript

1. Scope to GADR + Hawaii bbox + March 2026 + deep dives

search_datasets(
  collections = "GADR",
  bbox = "-161,18,-154,23",
  datetime = "2026-03-01T00:00:00Z/2026-03-31T23:59:59Z",
  filter = {
    "op": ">=",
    "args": [
      {"property": "aquaview:column_stats_summary.variables.Pressure.max"},
      1500
    ]
  }
)

Returns 6 floats. Sample of the top 5:

ID Deployer Pressure max (dbar) Temp range (°C) Salinity range
argo_jma_7900868 JMA 3783.0 -1.84 to 3.02 33.72–34.75
argo_coriolis_6902814 French Coriolis Centre 3551.1 -1.87 to 2.35 33.79–34.74
argo_coriolis_3902129 French Coriolis Centre 3366.0 -1.87 to 1.93 33.62–34.74
argo_aoml_7900678 NOAA AOML 4605.3 -1.86 to 3.51 33.06–34.75
argo_aoml_5905269 NOAA AOML 4789.4 -1.86 to 4.49 33.74–34.75

2. Sanity check — these don't look like Hawaiian water

The temperature ranges (-1.87 to 2.35 °C etc.) are oceanographically inconsistent with Hawaii. Hawaiian surface waters run 23–27 °C; profiles in the central N. Pacific don't get colder than ~2 °C until well below 1500 m, and even then the surface min is far above zero. These are Southern Ocean / Antarctic floats.

Inspecting the bbox of argo_jma_7900868: 116.11°E, -63.76°S, -81.0°W, 90.0°N. That's a polygon spanning from the western Pacific across the antimeridian to South America, latitudes from -64°S to 90°N — basically half the globe. That's how a Southern Ocean float ended up matching a Hawaii spatial filter.

3. Why this happens

AQUAVIEW's GADR collection stores one STAC item per Argo float, with bbox being the rectangular envelope of all profile locations across the float's lifetime. Highly-mobile floats (and floats with antimeridian-crossing trajectories) can have bboxes that encompass huge regions they never actually visited. Six items match the Hawaii bbox; none of them are oceanographically near Hawaii.

4. The right tool for this prompt

AQUAVIEW is a discovery layer for Argo at the float level. To find profiles geographically near Hawaii in March 2026, the right path is:

  1. Use AQUAVIEW to discover candidate floats and grab their GDAC URLs (we did that — got the JMA / Coriolis / AOML floats).
  2. Pull the per-profile NetCDF from GDAC and filter on actual latitude / longitude of each profile cycle. The IFREMER GDAC at https://data-argo.ifremer.fr/dac/ provides per-cycle profile files.
  3. Or use the Argo project's spatial selection tool at https://erddap.ifremer.fr/erddap/ which queries on per-profile coordinates rather than lifetime bboxes.

Result the agent gave

"Important caveat first: the prompt asks for Argo profiles 'within 200 km of Hawaii' but AQUAVIEW's GADR collection indexes floats at the float level, with bbox set to each float's lifetime trajectory envelope. Six floats match the Hawaii spatial + March 2026 temporal filter (and the deep-dive pressure constraint), but inspecting their column_stats_summary reveals temperature ranges of -1.87 to 4.49 °C — these are Southern Ocean / Antarctic floats whose bboxes happen to cross the antimeridian and encompass Hawaii. None of them are oceanographically near Hawaii.

The catalog returns: - argo_jma_7900868 (JMA), argo_aoml_7900678, argo_aoml_5905269 (NOAA AOML), argo_coriolis_6902814, argo_coriolis_3902129 (French Coriolis), and one more. - All have polar lifetime bboxes (e.g. 116°E to -81°W, -64°S to 90°N) and tropical-incompatible temperature ranges.

To answer the original question — Argo profiles actually near Hawaii in March 2026 — pull profile-level data from the IFREMER GDAC (https://data-argo.ifremer.fr/dac/) and filter on per-cycle latitude/longitude. AQUAVIEW's value here is identifying the float catalog and providing per-float download URLs; profile-level spatial filtering is the GDAC's job.

Lesson: when a column_stats_summary field looks oceanographically wrong for the region you queried, you've probably hit the lifetime-bbox-envelope artifact. Use temperature as a sanity check and escalate to the source's per-profile index."

Variations

  • Filter on temperature to enforce tropical: add a clause aquaview:column_stats_summary.variables.Temperature.min >= 15.
  • Different region: California Current -127,32,-117,42 — narrow, less polar-bbox contamination.
  • Specific float by WMO ID: q = "argo 7900869", no bbox.
  • All deep dives globally last year: drop bbox, set datetime = "2025-01-01T00:00:00Z/2025-12-31T23:59:59Z", filter Pressure.max >= 4000. Useful for finding abyssal-depth profilers without spatial scoping.