Passive Acoustic Sensor Capabilities (M12&15)

TRICUSO achieves two important project Milestones:

M12: European trials of wind and oxygen float off Villefranche
M15: Data pathway from float to data centre operational

The first European sea trials of a profiling float equipped with passive acoustic sensors that are capable of estimating surface wind speed from underwater ambient noise has been successfully achieved within the TRICUSO project. The deployments, conducted in the Ligurian Sea off Villefranche-sur-Mer, demonstrate a scalable pathway for autonomous wind monitoring in regions where direct atmospheric observations remain sparse, with direct implications for improving estimates of air–sea gas exchange and climate-relevant ocean processes.

These milestones, led by Sorbonne Université (LOV), support TRICUSO’s objective of expanding the autonomous biogeochemical observing system through the integration of new sensor capabilities and data pathways, which is part of Work Package 4.

Why are they important milestones?

Wind drives ocean circulation, mediates air–sea gas exchange and shapes climate-relevant biogeochemical processes. Yet, direct wind observations remain sparse across much of the global ocean: satellite scatterometers struggle under extreme weather and at high latitudes, while surface moorings provide only limited spatial coverage. This gap is particularly acute in regions like the Southern Ocean, where storms drive a disproportionate share of total air–sea CO₂ and heat fluxes, and where the very conditions that matter most are the hardest to observe.

Embedding passive acoustic wind sensors on profiling floats offers a way to close this gap. Because wave-breaking and bubble entrainment at the sea surface generate distinctive underwater noise in the 1–20 kHz band, ambient sound recorded at depth can be used to infer surface wind speed, turning every Biogeochemical (BGC) Argo float into a potential, low cost, mobile and autonomous weather station.

How was this achieved?

Phase One: Proof of Concept

The first phase of M12 demonstrated the above capability for the first time on a BGC-Argo platform. Two deployments of an acoustic-equipped PROVOR CTS5 float were carried out near the DYFAMED time-series station in the Ligurian Sea between February and April 2025, with the float drifting at parking depths of 500–1000 metres and recording ambient noise across 23 third-octave bands.

Key outcomes included:

Successful detection of wind-driven acoustic signatures from subsurface depths, with float-derived wind estimates closely matching collocated DYFAMED buoy observations across the full observed wind range (0.5–16.1 m s⁻¹),
Validation of established empirical wind-retrieval models on a profiling-float platform, with the cubic Nystuen formulation achieving the best fit (R² = 0.88),
A scalable calibration framework combining ERA5 reanalysis with sparse in-situ reference data through residual learning, reducing prediction error by 38.6% during high-wind events, a critical regime for air–sea flux estimation,
Demonstration of operational compatibility with existing BGC-Argo missions: the acoustic module adds only ~7–18% to a float’s energy and telemetry budget while operating alongside the standard biogeochemical sensor suite.

The work confirmed that acoustic-equipped profiling floats can deliver high-resolution wind observations from depth, providing a complementary data stream to satellite products and atmospheric reanalyses, particularly during short-lived high-wind events that coarse-resolution products tend to smooth out.

Read the full paper: Delaigue et al. (2026). Passive acoustic monitoring from profiling floats as a pathway to scalable autonomous observations of global surface wind. Ocean Science, 22, 101–117.

Phase Two: A Multi-Buoy Mediterranean Trajectory

Figure 1. Trajectory of the acoustic-equipped profiling float deployed in January 2026 as part of the second phase of TRICUSO’s M4.1 trials. The float was released near ODAS Italia 1 (O) in the Ligurian Sea and is drifting westward through the northwestern Mediterranean, passing in succession by the DYFAMED (D), PACA (P), and LION (L) meteorological buoys. Each buoy provides an independent reference for surface wind, allowing the float’s acoustic wind estimates to be cross-validated across distinct atmospheric and oceanographic regimes. Green markers along the track show successive surface positions; the deployment is ongoing.

The float is following a trajectory that brings it past four meteorological buoys in succession (Figure 1):

ODAS Italia 1 (Ligurian Sea, operated by CNR-IAS) — starting reference station,
DYFAMED (Ligurian Sea, operated by Météo-France / MOOSE) — the original calibration site,
PACA (Provence-Alpes-Côte d’Azur coastal buoy, operated by Météo-France),
LION (Gulf of Lion, exposed to strong Mistral and Tramontane wind regimes; operated by Météo-France).

This multi-station trajectory provides a unique opportunity to evaluate how acoustic wind-retrieval performance evolves as the float moves between distinct atmospheric and oceanographic regimes.

Preliminary Results

Figure 2. Surface wind speeds recorded at the four reference buoys along the float’s trajectory between January and April 2026. Top panel: hourly wind speeds at ODAS Italia 1 (red), DYFAMED (blue), PACA (green), and LION (orange), shown on the same time axis as the float record. LION consistently records the strongest winds during high-wind events, reflecting its exposure to Mistral and Tramontane regimes in the Gulf of Lion, while the Ligurian buoys (ODAS, DYFAMED) experience generally weaker and more variable conditions. Even between buoys located relatively close to one another, wind speeds diverge substantially during storm events, illustrating why a single calibration may not generalise across the trajectory. Bottom panel: distance between the float and each buoy over time. The float starts close to ODAS Italia 1, drifts past DYFAMED, and approaches PACA and LION as the deployment progresses.

Figure 3. Vertical profiles of temperature, salinity, and dissolved oxygen recorded by the acoustic-equipped BGC-Argo float during the ongoing 2026 deployment in the northwestern Mediterranean. Each coloured line represents one profile (one every 5 days), spanning the upper 1000 m. The temperature and salinity profiles capture the seasonal evolution of the surface mixed layer and the underlying Mediterranean water masses, while the oxygen profiles show a pronounced subsurface maximum near 50–100 m and a gradual decline with depth. Together with the passive acoustic record, these profiles confirm that the new sensor suite operates without compromising standard BGC-Argo measurements.

Approximately three months into the deployment, the float has produced a continuous record of acoustic, hydrographic, and biogeochemical observations along its westward path. Several preliminary findings already point to the value of the multi-buoy strategy:

Acoustic wind retrievals have been successfully generated along the full trajectory using the same processing chain established in Delaigue et al. (2026), including transient-noise rejection and depth-dependent attenuation correction. The depth correction is now computed from each individual temperature/salinity profile rather than a deployment mean. This is especially important for trajectories that traverse changing hydrographic conditions.
Wind regimes differ substantially across the four reference buoys (Figure 2). Comparison of collocated buoy winds shows that LION consistently records the strongest winds during high-wind events, consistent with its exposure to Mistral and Tramontane outbreaks, while ODAS Italia 1 and DYFAMED report weaker and more variable conditions. Even between ODAS and DYFAMED (both in the Ligurian Sea) the two buoys diverge during energetic events, illustrating that wind fields decorrelate over relatively short distances. This directly motivates the central question of the deployment: how far can a single calibration travel before it breaks down?
A full suite of biogeochemical profiles (temperature, salinity, dissolved oxygen; Figure 3) has been recorded alongside the acoustic data, with derived carbonate-system variables (total alkalinity, dissolved inorganic carbon, pH, pCO₂; Figure 4) reconstructed using the CANYON-Med neural-network framework. Surface oxygen and pCO₂ both vary systematically along the trajectory, with elevated subsurface pCO₂ values observed closer to LION compared to ODAS Italia 1, a regional signal that will be examined further as the deployment progresses.
The dataset opens new analyses unavailable in the first deployment, including the possibility of correlating in-situ wind variability with surface pCO₂ and oxygen drawdown, and of testing how a residual-learning calibration anchored at one buoy generalises to the next. These analyses will help define the practical “calibration radius” for acoustic wind retrieval, a key parameter for designing future deployments in regions where reference buoys are sparse or absent.

The deployment is expected to continue through spring 2026, with full analysis to follow once the float completes its passage near LION.

Figure 4. Vertical profiles of total alkalinity, dissolved inorganic carbon, pH, nitrate, phosphate, and silicate estimated from the acoustic-equipped BGC-Argo float during the ongoing 2026 deployment in the northwestern Mediterranean using the CANYON-MED algorithm (Fourrier et al., 2020). Each coloured line represents one profile (one every 5 days), spanning the upper 1000 m. These derived carbonate and nutrient fields provide additional insight into the biogeochemical structure of the water column throughout the deployment. Variability is strongest in the upper layers, where biological uptake, remineralisation, and mixing shape the distributions, while deeper waters exhibit more stable vertical gradients associated with the characteristic Mediterranean water masses. By complementing the core float observations, these reconstructed variables help place the passive acoustic measurements in a border environmental context.

Data Pathway to the Argo Data System

An equally important outcome of this deployment is the float’s full integration into the operational Argo data system (Milestone 15). Beyond demonstrating the scientific feasibility of passive acoustic wind retrieval, the deployment also confirms that these observations can be transmitted, processed and distributed through the existing Argo infrastructure, a critical requirement for any future large-scale implementation.

The float’s trajectory and measurements are already accessible through standard Argo monitoring tools, including the Euro-Argo Fleet Monitoring interface and the BGC-Argo data visualisation products. These platforms allow near real-time tracking of the deployment within the broader Argo framework, meaning the float is functioning not as a standalone prototype, but as an operational component of the international observing system.

Dissolved oxygen is being adjusted in real time following established BGC-Argo protocols, and the pH sensor is transmitting successfully, which confirms that the expanded sensor suite is performing reliably throughout the mission. Crucially, the acoustic measurements are also being routed into the Argo data structure, embedded in the float’s auxiliary trajectory files alongside standard metadata and mission variables. This is a significant technical step: it demonstrates that the passive acoustic data stream can be incorporated into the existing Argo architecture without compromising interoperability.

Taken together, these advances show that this work is not only about proving new sensing capabilities in the field, but about demonstrating that those capabilities can enter an operational data system, be made visible through established tools, and contribute in practice to the evolving global biogeochemical observing network.

Looking Ahead

Together, Milestones 12 and 15 establish a clear pathway from single-site validation toward operational, basin-scale deployment of acoustic wind-sensing floats. The lessons learned in the Mediterranean (a region with strong, spatially heterogeneous winds and dense reference infrastructure) are directly informing preparations for TRICUSO’s next ambitious deployment: the first release of profiling floats integrating both wind and oxygen sensors into the Southern Ocean, scheduled for February 2027.

The Southern Ocean is precisely the kind of environment this technology was designed for. Storm-driven winds dominate air–sea CO₂, heat, and momentum exchange there, yet remain among the most poorly observed in the global ocean. This is exactly the gap that acoustic-equipped, oxygen-enabled floats are positioned to fill. By integrating these new sensor capabilities into the BGC-Argo framework, TRICUSO is contributing to a more complete and metrologically anchored ocean observing system, one capable of resolving the air–sea forcing that drives the fluxes between ocean and atmosphere in regions where ships and satellites do not inclusively cover.

This milestone supports TRICUSO’s Key Exploitable Result 4: Description of the capabilities of the new wind sensor relative to the reanalysis wind field and the wave glider mounted wind sensor.

References

Delaigue, L., Cauchy, P., Cazau, D., Bonnel, J., Pensieri, S., Bozzano, R., Gros-Martial, A., Schaeffer, C., David, A., Stil, P., Poteau, A., Schmechtig, C., Leymarie, E., and Claustre, H. (2026). Passive acoustic monitoring from profiling floats as a pathway to scalable autonomous observations of global surface wind.
Ocean Sci., 22, 101–117.

Fourrier M, Coppola L, Claustre H, D’Ortenzio F, Sauzède R and Gattuso J-P (2020). A Regional Neural Network Approach to Estimate Water-Column Nutrient Concentrations and Carbonate System Variables in the Mediterranean Sea: CANYON-MED.
Front. Mar. Sci. 7:620.