Field notes · deep ocean

The Deep Signal

How a fleet of drifting robots caught a hidden underwater eruption, traced iron from the seafloor to the surface, and revealed a Southern Ocean bloom that returns almost every year — plus what any of this has to do with new islands rising out of the Ring of Fire.

Depth range covered: 0–6,000 m Timespan: 1930–2026 Instrument: Argo float network

A volcano no one saw erupt ~900–2,000 m

In July 2012, the Havre caldera in New Zealand's Kermadec Arc produced one of the largest deep submarine eruptions on record — and almost nobody noticed. Its main visible trace was a floating pumice raft roughly the size of Belgium, drifting across the Pacific.

The eruption itself was caught almost by accident. Argo floats — robotic instruments that drift through the ocean, diving to depth and profiling temperature and salinity every ten days — happened to record a sharp, statistically unmistakable temperature increase and salinity decrease at 1,750–2,000 meters, deeper than the vent itself.

The puzzle: the signal sat below the vent, not above it. That ruled out simple heat conduction. Researchers instead proposed the rising eruptive plume acted like a piston, dragging and mixing deep water layers as it moved — a large-scale mixing event, not direct heating.

The eruption wasn't confirmed by Argo data until a 2019 study — seven years after the fact. That gap says a lot about how much of the ocean's volcanic activity has been running unnoticed beneath a sparse observing network.

The instrument that made this visible 0–6,000 m

Argo is a global array of roughly 4,000 free-drifting floats. Each one sinks to a parking depth, drifts for about nine days, then descends further before rising back to the surface, recording a full profile as it goes. It beams the data home by satellite, then sinks again. The array launched in 1999 and hit its original 3,000-float design target in 2007.

~4,000active floats today

2M+profiles collected since 1999

6,000 mDeep Argo's reach, vs. 2,000 m core

The newer Biogeochemical Argo (BGC-Argo) floats add sensors for chlorophyll, oxygen, nitrate, pH, and suspended particles — turning a physical-oceanography tool into something that can watch phytoplankton blooms and volcanic chemistry directly, not just infer them from temperature.

Why hot water doesn't just float to the surface vent → +200–700 m

Vent fluid does rise on real thermal buoyancy at first — the same principle as smoke off a fire. But that lift is short-lived: as the plume rises it entrains and mixes with surrounding seawater, cooling and diluting until it matches the density of the water around it. That neutral-buoyancy point typically arrives within a few hundred meters of the vent, nowhere near the surface for anything venting below ~1,000 m.

1

Thermal rise. Hot fluid leaves the vent and rises purely on density difference, entraining seawater as it goes.

2

Neutral buoyancy. Diluted roughly 10,000:1, the plume matches ambient density and stops rising — usually a few hundred meters above the vent.

3

Advection takes over. From here, dissolved and colloidal iron simply rides ocean currents — sometimes for thousands of kilometers — with no further thermal push involved.

Iron itself isn't a "heavy mineral" being hauled upward. Nearly all of it leaves the vent already dissolved or as nanometer-scale colloids, stabilized by organic iron-binding ligands that keep it in solution. It has essentially the same density as seawater — it doesn't need lifting so much as carrying.

Where deep iron actually reaches daylight 2,000–2,700 m → surface

Two Southern Ocean ridge systems have supplied the clearest evidence that vent iron can fertilize surface blooms, if currents and topography do the work that thermal buoyancy alone can't.

266,000 km²AAR bloom, Jan. 2014

20 / 22 yrsAAR bloom recurrence

3–4 Gmol/yrrevised global hydrothermal Fe estimate

Southwest Indian Ridge (SWIR): the first documented case (Ardyna et al., 2019). Two BGC-Argo floats tracked hydrothermally-influenced deep water rising downstream of active vents, driven by eddy kinetic energy where currents meet steep ridge topography.

Australian-Antarctic Ridge (AAR): a larger, more persistent bloom near two vent fields, KR1 and KR2 — recurring in 20 of the last 22 satellite-recorded years, and more than three times the peak chlorophyll of the SWIR bloom.

The recharge mechanism: a 2025 study linked year-to-year bloom strength directly to seismicity near the vents in the months before the growing season. Magma chambers refill over years, building pressure that eventually cracks the crust — releasing a fresh pulse of iron each time. More earthquakes, more iron, bigger bloom.

The shallow-vent shortcut 200–450 m

Deep ridge vents mostly feed iron to the ocean's interior. Shallow arc vents are a different story — close enough to the surface that they can fertilize it directly, without needing the elaborate current-and-topography lift that SWIR and AAR require.

Along the Tonga volcanic arc, a shallow vent (roughly 200–450 m deep) sustains a standing chlorophyll patch of about 360,000 km² — larger than the AAR bloom, and continuous rather than seasonal. Dissolved iron there reaches concentrations high enough to fuel one of the ocean's recognized nitrogen-fixation hotspots, contributing an estimated fifth of the world's fixed-nitrogen input.

What a bloom does to the water above it surface

Blooms don't just sit passively in fertilized water — they change its temperature. Phytoplankton pigments absorb sunlight far more efficiently than clear seawater, concentrating heat in a thin surface layer rather than letting it spread through the water column.

+1.5°Cdocumented local SST rise from a bloom

+4.5°Cover just 4 days, one measured case

~0.5°Cmodeled global average bloom-driven SST rise

It's a trade-off, not a simple gain: the same absorption that warms the surface blocks light from reaching deeper water, which cools slightly below the bloom. A shallower, warmer mixed layer can then help sustain the bloom itself — a small feedback loop layered on top of whatever else is warming that patch of ocean.

Following the arc to Indonesia tens of m → 1,800 m

Havre, Tonga, and the Kermadec Arc all sit on the same structure: the Tonga-Kermadec system, the longest single subduction zone on Earth. That arc is part of the wider Ring of Fire, which runs through Indonesia's Sunda and Banda arcs on its way toward Japan.

Indonesia has both ends of the depth spectrum. Kawio Barat, a 3-km-tall submarine volcano off North Sulawesi, vents at 1,800–1,900 m — confirmed active by light-scattering and redox sensors, with ROV footage showing sulfide chimneys and dense shrimp and barnacle communities. That's deep enough that thermal buoyancy alone couldn't reach the surface; any fertilization effect there would need the same current-driven lift as SWIR and AAR, and hasn't yet been measured.

By contrast, two vents off North Sulawesi's coast sit shallow enough to reach by SCUBA — putting them in the same "direct reach" category as Tonga's Volcano 1, though their surface fertilization effect, if any, hasn't been studied the way Tonga's has.

When the plume breaks the surface entirely 0 m

At the shallowest end of this whole story, a plume doesn't just fertilize water — it builds land. Across the Ring of Fire, that's happened repeatedly, though most of what forms erodes away again within months to years.

• 1930Anak Krakatau breaks the surface in Krakatoa's flooded caldera, Sunda Strait — and survives, thanks to durable lava flows rather than loose ash.
• 1973–2013Nishinoshima, Japan, grows a new island through repeated lava-effusion eruptions, later merging with the original islet.
• 2009 / 2015 / 2022Hunga Ha'apai, Tonga, repeatedly forms and loses new land, culminating in the 2022 Hunga Tonga explosion.
• 2021Fukutoku-Okanoba, Japan, forms an island that quickly erodes — its "crumbly" pumice composition can't resist wave action.
• 2022–2023Home Reef, Tonga, builds and reshapes an island across two consecutive eruptions.
• 2026An eruption on Titan Ridge, Bismarck Sea, is caught live by satellite — steam plumes, pumice rafts, and discolored water, with scientists watching to see if it breaches the surface.

What decides permanence: lava-based eruptions (Nishinoshima, Anak Krakatau, Surtsey) tend to survive; ash- and pumice-based ones (Fukutoku-Okanoba, most Home Reef events) almost always lose the race against wave erosion. Either way, the underlying volcano and its plume keep running continuously — the island's presence or absence is just a visible symptom, not the real story.

The observing network, 25 years on 0–6,000 m

None of the above would be visible without a quarter-century of steady buildout in ocean instrumentation. Argo didn't just get bigger — it got categorically more capable.

1999Argo program begins deployment from zero.

2007Reaches its original 3,000-float design target.

2012One-millionth profile collected.

2018Two-millionth profile — the second million took half as long as the first.

2021–2026GO-BGC deploys 500 new biogeochemical floats; Deep Argo pushes coverage to 6,000 m.

2026 →Proposed "OneArgo" expansion targets the Arctic, marginal seas, and denser Southern Ocean BGC coverage.

The distinction matters: Havre's 2012 eruption was caught using data that already existed — plain temperature and salinity. The SWIR and Tonga iron-fertilization blooms couldn't have been seen at all before chlorophyll- and particle-sensing BGC floats existed. Each expansion phase hasn't just added more of the same measurements — it's opened entirely new categories of ocean events to view for the first time.

Sources referenced in researching this piece include Mittal & Delbridge (2019, Earth and Planetary Science Letters) on the Havre Argo detection; Ardyna et al. (2019, Nature Communications) and Schine et al. (2021, Nature Communications) on Southern Ocean hydrothermal blooms; Schine et al. (2025, Nature Geoscience) on seismically modulated iron; and Bonnet et al. (2023, Science) on Tonga arc iron fertilization. Figures on Ring of Fire eruptions and Argo program milestones are drawn from Smithsonian Global Volcanism Program reports and the Argo program's own published statistics.