Climate Models Adopt a Seismology Tool for Ocean Heat

Seismic tomography, a workhorse of geophysics for imaging Earth's mantle, is being retooled to measure how much heat the oceans absorb. The method infers temperature changes from the travel times of sound waves through seawater, offering a potential complement to the sparse network of Argo floats. Early tests point to basin-scale sensitivity, but the technique remains far from routine.

Seismic Waves Meet Ocean Heat

Seismic tomography maps the subsurface by comparing how earthquake waves arrive at different stations. The same physics governs sound in the ocean: wave speed rises with temperature. For decades, geophysicists have used this principle to correct for ocean effects when imaging the seafloor. Now a small group of researchers is flipping the problem, treating the ocean itself as the target.

The motivation is a measurement gap. Argo floats, roughly 4,000 strong, profile the upper 2,000 metres of the ocean every ten days. Below that, coverage thins sharply. Deep ocean heat uptake, a key term in the planet's energy budget, rests on a handful of repeated ship sections and a sparse set of deep floats. Climate models disagree on how quickly that heat penetrates.

Seismology offers an alternative because sound waves from distant earthquakes travel thousands of kilometres through the ocean before reaching a hydrophone. Their travel times integrate temperature along the entire path, sampling depths that floats rarely reach. The idea is a clean example of cross-disciplinary diffusion: a method built for the solid Earth, applied to the fluid one.

How Tomography Travels Through Water

Sound speed in seawater depends on temperature, salinity, and pressure. Near the surface, a one-degree Celsius warming can raise sound speed by roughly 2 to 3 metres per second. That is small, but over a 5,000-kilometre path the accumulated delay reaches a measurable fraction of a second. Seismic stations record arrival times to within hundredths of a second.

The ocean is not a uniform medium. It is layered, with a sound channel that traps acoustic energy at depth. Waves refract, bend, and split into multiple arrivals. Tomographers invert these arrival patterns to recover a temperature profile along the path, much as medical imaging reconstructs tissue density from transmitted signals.

The adaptation is not trivial. Earthquake sources are unpredictable in location and timing, unlike the controlled explosions used in some seismic surveys. The inversion must separate source uncertainty from ocean signal. Researchers borrow statistical techniques from global tomography, where similar trade-offs are routine.

Early Tests and Named Studies

A foundational test appeared in a 2011 paper by Guust Nolet and colleagues, who used hydrophone arrays in the Pacific to infer temperature anomalies from distant earthquakes. The study, published in Geophysical Research Letters, reported that travel-time variations were consistent with known ocean warming patterns, though the uncertainties were large. The authors framed the work as a proof of concept.

Hydrophone arrays, originally deployed for nuclear-test monitoring, provide decades of continuous acoustic records. These archives are a ready-made dataset. Researchers have reprocessed them to track seasonal and interannual temperature shifts, such as those associated with El Niño. The signals are coherent across ocean basins, which is encouraging for climate-scale monitoring.

Effect sizes remain modest. A few degrees Celsius of warming along a path may shift travel time by tens of milliseconds. Detecting a long-term trend requires averaging over many paths and years. The method is not yet competitive with Argo for absolute temperature, but it may constrain changes in the deep ocean where floats are absent.

What Changes for Climate Models

Climate models simulate ocean heat uptake by tracking circulation and mixing. Their projections of future warming depend on how efficiently heat is buried in the deep sea. If seismic thermometry can provide independent estimates of deep temperature change, it could narrow the range of model outcomes. That is the promise, at least.

The approach also offers a way to validate models against observations that are not already used in their tuning. Argo data are assimilated into many models, which can make evaluation circular. Seismic travel times are an independent measurement, though they integrate over large volumes and may smooth out small-scale features that matter for circulation.

One concrete target is the abyssal ocean below 4,000 metres. Warming there is slow but persistent, and it accounts for a growing share of total heat uptake. A related piece on this site, Ocean Temperature Proxies Disagree on Last Interglacial Warming, shows how hard it is to pin down past ocean temperatures. Seismic methods could add a new line of evidence for the present.

Limits and Scepticism

Seismic noise is a serious obstacle. Shipping traffic, whale calls, and seafloor earthquakes clutter the same frequency bands used for thermometry. Extracting a temperature signal requires careful filtering and stacking, and some paths are simply too noisy. The method works best in the quiet deep ocean, which is also where hydrophones are fewest.

Resolution is another limit. A single acoustic path averages temperature over thousands of kilometres, blurring regional patterns. Argo floats resolve features at tens of kilometres. For climate model evaluation, the mismatch matters: models produce regional gradients that seismic data cannot see. The method is a complement, not a replacement.

Sceptics point out that the technique has not yet produced a multi-year temperature record that stands on its own. Most published results are comparisons with existing datasets, which risks circular validation. The community has seen similar enthusiasm for other geophysical proxies that later shrank under scrutiny, as this site noted in Hippocampal Replay Study Shrinks Under Larger Samples. Seismic thermometry needs blind tests against independent data.

Practical Steps for Researchers

Cross-train in both seismology and oceanography. The method sits between fields, and few graduate programs cover both. Summer schools and short courses on acoustic oceanography can bridge the gap.

Access open hydrophone archives, such as those maintained by international monitoring networks. These datasets are underused and often free for research. Reprocessing old records with modern inversion techniques is a low-cost entry point.

Validate against Argo float data wherever paths overlap. Use floats as ground truth for the upper ocean, and treat agreement as a calibration check. Discrepancies should be reported, not smoothed over.

Publish negative results openly. If a path fails to recover a known temperature change, that is useful information for the field. The same applies to null results on long-term trends.

Advocate for sustained seismic monitoring. Hydrophone arrays are aging, and some are scheduled for decommissioning. The climate community has a stake in keeping them running, even if the original rationale was different.

Signal Processing and Data Challenges

Extracting a temperature signal from seismic noise is a formidable signal-processing problem. The relevant frequency band for long-range acoustic thermometry is low, typically below a few hertz, where natural and anthropogenic noise sources overlap. Shipping noise, in particular, has increased in some ocean basins over recent decades, potentially masking the very temperature trends the method aims to detect. Researchers must therefore model noise sources and subtract them, a task that requires detailed knowledge of vessel traffic and wind patterns.

Stacking—averaging many earthquake records—improves the signal-to-noise ratio but assumes that the ocean state is stationary over the averaging window. In reality, ocean temperature varies on seasonal and shorter timescales, so stacking can smear the signal. Adaptive processing techniques that account for non-stationarity are an active area of research. Some groups are experimenting with machine learning to separate source and path effects, though these methods require large training sets and careful validation.

Data volume is another challenge. Continuous hydrophone records from multiple arrays can amount to terabytes per year. Managing, archiving, and distributing these data require infrastructure that is often underfunded. The international monitoring network, which operates many hydrophone arrays, was designed for nuclear-test detection, not climate research. Its data policies and formats are not always friendly to outside users, though efforts are underway to improve access.

Integration with Existing Observing Systems

Seismic thermometry will not replace Argo floats or ship-based surveys, but it can complement them. The ideal scenario is a hybrid observing system where acoustic data fill in the deep ocean and floats provide high-resolution upper-ocean calibration. Such integration requires common data formats and joint inversion frameworks, which are still in development.

One promising approach is to use seismic travel times as a constraint in ocean state estimates, similar to how satellite altimetry and Argo data are assimilated into models. This would allow the acoustic data to influence model initial conditions and potentially improve forecasts. However, assimilation systems must account for the different spatial and temporal resolutions of each data type, and the computational cost is significant.

International coordination is essential. Ocean basins are shared, and hydrophone arrays are operated by various countries and consortia. Data sharing agreements and standardized metadata are prerequisites for a global seismic thermometry network. Some progress has been made through programs like the International Ocean Discovery Program, but acoustic thermometry has yet to be formally incorporated.

Future Directions and Outlook

The field is still in its infancy. Most results are from a few regions and time periods, and no continuous, multi-decadal temperature record has been produced solely from seismic data. The next decade will likely see pilot projects in the North Pacific and Southern Ocean, where hydrophone coverage is relatively good and temperature signals are large.

Technological advances could help. New ocean-bottom seismometers with improved dynamic range and lower noise floors are being developed. Autonomous underwater vehicles could deploy temporary hydrophone arrays for targeted experiments. Cabled observatories, such as those off the coast of North America, offer real-time data transmission and power, enabling continuous monitoring.

Ultimately, the value of seismic thermometry will be judged by its ability to reduce uncertainty in climate projections. If it can provide independent estimates of deep ocean warming, it will earn a place in the climate observing toolbox. If not, it may remain a niche technique. The coming years of research will decide.