The study, published in PLOS Ecosystems by Emmanuel Boss, Charlotte Begouen Demeaux, and Michael J. Behrenfeld, revisits how the euphotic depth should be computed. The authors revive an argument made two decades ago by the influential biological oceanographer Karl Banse, who died in 2025: phytoplankton, the microscopic algae that anchor the marine food web, do not sense light as a percentage of surface illumination. They experience only the absolute flux of photons arriving at whatever depth they happen to occupy. A definition based on relative light levels, the authors contend, is therefore physiologically meaningless.

The one-percent rule has undeniable practical appeal. It can be estimated with an uncalibrated radiometer, a simple Secchi disk, or satellite ocean-color products, and it appears in studies of phytoplankton seasonal cycles, carbon flux calculations, and remote-sensing data distributed worldwide. But the new analysis, built on nearly 21,000 profiles from autonomous BGC-Argo floats spanning October 2012 to December 2023, shows that the absolute amount of light found at the one-percent horizon varies enormously with latitude, season, and cloud cover. At any given moment, roughly seventy percent of the ocean lies beneath clouds, and seasonal swings in solar angle halve surface irradiance by thirty degrees latitude and extinguish it entirely near the poles in winter.

Instead of a percentage, the authors recommend anchoring the euphotic depth to a fixed absolute light level called the compensation irradiance: the minimum daily light at which photosynthesis can outpace algal respiration, allowing net growth. Drawing on recent Arctic field campaigns and laboratory work, they propose a value of 0.0035 plus or minus 0.002 mol photons per square meter per day, a figure orders of magnitude lower than many earlier estimates and close to the theoretical minimum of about 0.0009 mol photons per square meter per day derived from the bioenergetics of photosynthesis.

The evidence for such extraordinarily low-light growth comes from converging sources. In Baffin Bay, researchers using chlorophyll fluorescence and particle backscattering sensors below sea ice detected the onset of phytoplankton accumulation at an average light level of 0.0043 mol photons per square meter per day. A separate Arctic study coupling biomass measurements with radiocarbon-based production assays found a compensation irradiance of 0.0035 mol photons per square meter per day, essentially identical to the value now proposed. In the Southern Ocean near the Kerguelen Islands, incubation experiments previously documented positive net primary production at the 0.01 percent light level, corresponding to roughly 0.006 mol photons per square meter per day.

Critically, the new work extends the argument beyond polar waters. Analyzing cytometric counts of Prochlorococcus, the abundant photosynthetic bacterium that dominates nutrient-poor oceans, from the Atlantic Meridional Transect cruise 13, the authors found significant cell concentrations down to depths where daily light falls to the proposed 0.0035 threshold, even in subtropical latitudes. Oxygen-based measurements of gross primary production from other Atlantic transects likewise show detectable production at the 0.1 percent light level. Because the ocean is well connected and microbial strains disperse globally, the authors reason that an algal lineage capable of near-theoretical-minimum growth in the Arctic almost certainly has counterparts elsewhere.

The quantitative consequences are striking. Across the BGC-Argo dataset, the one-percent light horizon sits, on average, about 80 meters, or 48 percent, shallower than the depth of the proposed compensation isolume. When the authors integrated chlorophyll and phytoplankton carbon from the surface down to each horizon, the conventional one-percent cutoff missed more than sixty percent of the biomass inventory captured by the deeper boundary: average integrated chlorophyll was 40.9 versus 65.3 milligrams per square meter, and phytoplankton carbon 1.1 versus 1.8 grams per square meter. A substantial reservoir of phytoplankton, in other words, lives below the depth where most studies assume production stops.

That oversight matters for the biological pump, the suite of processes that transports carbon from the surface ocean into the deep sea, where it can remain sequestered for decades to centuries. Many pump estimates treat everything below the one-percent horizon as a realm of pure respiration, assuming any phytoplankton found there will simply sink. If production actually continues deeper, those assumptions inflate the apparent strength of the pump. Similar biases arise in estimates of the mixed-layer pump, the seasonal export of carbon when springtime shallowing of the mixed layer strands biomass in darkening water; using the shallower horizon overstates how much carbon that mechanism delivers to depth.

Numerical ecosystem and biogeochemical models, notably, already sidestep the percentage convention. They instead assign phytoplankton a constant basal respiration rate, typically 0.02 to 0.03 per day, and growth occurs only where photosynthesis exceeds that cost. The authors show, through a standard photosynthesis formulation, that such respiration rates imply a compensation irradiance of roughly 0.036 mol photons per square meter per day, still an order of magnitude above the lowest field estimates. Even models, it seems, do not yet allow production at the dimmest levels where real algae demonstrably persist.

The authors acknowledge open questions. Temperature might, in principle, raise respiration and thus the required light at lower latitudes, but meta-analyses suggest the temperature dependence of algal net growth weakens and effectively vanishes as light becomes limiting, lending support to a single global threshold. Deep mixed layers and internal waves complicate the light environment individual cells experience, and the appropriate isolume for a given question may differ: migrating zooplankton follow far dimmer isolumes, while nutrient and particle-attenuation horizons align with much brighter ones. Still, the recommendation is unambiguous: after nearly a century, oceanographers should retire the one-percent rule for defining the euphotic depth and adopt a physiologically grounded, absolute light threshold, a shift the authors suggest could be implemented with existing satellite products and may ultimately revise how the living ocean is measured from top to bottom.

Subject of Research: Redefining the ocean's euphotic depth using a physiologically based compensation irradiance instead of the one-percent surface light convention

Article Title: Revisiting how the euphotic depth is computed

Article References: Revisiting how the euphotic depth is computed. (n.d.). https://doi.org/10.1371/journal.pesy.0000032

Image Credits: AI Generated

DOI: 10.1371/journal.pesy.0000032

Keywords: euphotic depth, phytoplankton, compensation irradiance, oceanography, primary production, biological pump, BGC-Argo, Prochlorococcus, light attenuation, carbon export, Arctic Ocean, marine ecosystems

The core insight of the method, published in the journal Nonlinear Processes in Geophysics, is deceptively simple. Instead of manipulating the inputs or outputs of a trained network, as most previous approaches to the out-of-distribution problem have done, the Potsdam team manipulates the operator itself, meaning the network’s weights and biases. Their technique unfolds in three steps. First, a trained network, which the authors call the parent model, is fine-tuned on individual data points or subsets of its own training data, and the resulting deviations in every weight and bias are carefully recorded. Second, a regression is established between these weight anomalies and suitable predictors drawn from the data. Third, that regression is extrapolated to the application data, generating a new network, the child model, whose weights are tailored to the conditions of the target domain.

The reasoning behind the approach rests on a subtle property of neural networks. When a trained model is fine-tuned on a single training example, its weights shift only slightly, but those shifts are meaningful: they encode the network’s sensitivity to that particular sample. Ideally, retraining on many individual samples produces a cloud of related models scattered around the parent in parameter space, and that cloud encodes how the network’s internal parameters respond to changes in the data. By fitting a regression to this cloud and extrapolating it beyond the training distribution, the researchers can generate networks whose weights are adapted to regimes that were never part of the original training. In effect, the method turns the network’s own training sensitivity into a compass pointing toward uncharted territory.

To demonstrate the technique, the authors designed three demanding test cases drawn from climate and Earth sciences. The first tackles one of the most consequential problems in modern climate science: the potential tipping of the Atlantic Meridional Overturning Circulation, the vast system of ocean currents that includes the Gulf Stream. Using simulation data from a complex climate model in which the circulation collapses under freshwater forcing, the researchers trained a convolutional neural network to predict the current’s strength from two-dimensional maps of ocean velocity. Crucially, the network was trained only on data from before the tipping point, which occurs around year 1800 of the simulation, yet it had to produce accurate predictions during and after the collapse, a genuinely severe out-of-distribution challenge.

The results were striking. The parent model, trained conventionally, remained stubbornly biased toward the conditions it had learned, producing outputs that lagged far behind the actual collapse of the circulation. The child model, whose weights had been extrapolated using predictors based on the leading empirical orthogonal functions of the velocity fields, tracked the tipping and its aftermath far more faithfully. Because the experiment involves an element of randomness inherent to neural network training, the authors repeated it several hundred times with varying training windows. Across the ensemble, the child models consistently showed lower root mean squared errors than their parents, and the advantage grew larger as conditions diverged further from the training data. Notably, simple linear regressions of the weights performed as well as or better than higher-order polynomials, which tended to assign noisy, spurious slopes to terms that barely varied within the training range.

The second experiment addressed a spatial rather than temporal challenge: estimating the density of seawater, a highly nonlinear function of salinity, temperature, and pressure governed by an empirical international standard known as TEOS-10. The researchers trained a small feed-forward network on gridded observations from the World Ocean Atlas, but only above 2000 meters depth, the limit of the Argo float array that dominates modern ocean monitoring. Below that boundary, the network had to operate entirely on its own. Here the weight-prediction method proved remarkably stable: in the deep ocean, the child models reduced the error of the parent models by 50 to 100 percent, an average improvement of about 2 kilograms per cubic meter that the authors describe as substantial by oceanographic standards. The improvement was largest precisely where the parent model performed worst, in the cold, high-pressure depths far removed from the training data.

The third experiment pushed the method into cross-regime territory. The researchers trained a network to estimate uncertainties in global wind velocity reanalyses, using two widely used atmospheric products, ERA5 from the European Centre for Medium-Range Weather Forecasts and CFSv2 from the United States National Centers for Environmental Prediction. The training data came from a single oceanic grid point in the Southern Ocean, but the child models were asked to predict wind uncertainty over the continents, where boundary-layer processes, surface roughness, and convection differ fundamentally from conditions over water. Even in this extreme scenario, the approach delivered: the global mean error over land dropped by roughly 10.7 percent in the headline experiment, with repeated trials showing improvements between minus 2 and 14 percent and an average of 5.25 percent. The largest gains appeared over southeastern Australia, eastern Brazil, the East African Highlands, Indonesia, and southeastern Russia.

Perhaps the most philosophically intriguing result came from the third experiment, where the researchers used a neural network itself to generate the child models, rather than a conventional regression. Using one network to fix the out-of-distribution failures of another might seem circular, since the terrestrial data is equally foreign to both. The authors argue that the trick works because it splits the problem into two parts: solving the task and adapting the task to a new domain. That separation, they suggest, amounts to a structural implementation of prior knowledge that a single network cannot yet develop on its own. They also note that nonlinear regime shifts in a network’s input and output space can translate into simpler, even linear, changes in weight space, because nonlinear activation functions allow complex output changes to be realized through modest parameter adjustments distributed across layers.

The method is not without caveats. Because it inherits the notorious stochasticity of neural network training, not every parent model produces a better-behaved offspring; some parents sit in loss landscapes so flat or so narrow that fine-tuning barely moves their weights, and occasionally a child model deteriorates on data resembling the training conditions. The authors recommend ensemble approaches for now and identify parent-model selection, predictor pruning, and regression validation as priorities for follow-up research. They also caution that the choice of predictors and regression methods must be guided by domain knowledge, and that for very large networks, low-rank adaptation techniques may be needed to keep the per-weight regressions computationally feasible. In their appendix comparisons, none of the standard extrapolation methods they tested, including Gaussian processes, random forests, and support vector machines, outperformed the child models on the out-of-distribution tasks.

Even so, the implications reach well beyond oceanography and climate science. The authors suggest the approach could benefit any field where training data are spatiotemporally limited and conditions at the application site differ from those at the training site, from astrophysics and biology to medicine, where learned diagnostic skills might be transferred to patient groups underrepresented in the training data. They add that the technique may even improve networks that are undertrained, overfitted, or deliberately simple, even when no distribution shift is present. In an era when artificial intelligence is being asked to forecast climates it has never observed, monitor oceans it has never sampled, and anticipate tipping points that have not yet arrived, methods that let neural networks rewrite their own internal parameters for the world they actually face may prove essential.

Subject of Research: A weight-prediction method for improving neural network performance on out-of-distribution data in the geosciences

Article Title: Conditional updates of neural network weights for increased out of training performance

Article References: Saynisch-Wagner, J., & Sari, S. R. (2026). Conditional updates of neural network weights for increased out of training performance. Nonlinear Processes in Geophysics, 33(4), 503-520. https://doi.org/10.5194/npg-33-503-2026

Image Credits: AI Generated

DOI: 10.5194/npg-33-503-2026

Keywords: neural networks, out-of-distribution, machine learning, climate science, AMOC tipping, oceanography, weight prediction, seawater density, wind reanalysis uncertainty, GFZ Potsdam, extrapolation, geosciences

Melissa Omand, a research professor at the URI Graduate School of Oceanography, is leading the project, which is funded by a grant of nearly 700,000 dollars from the U.S. National Science Foundation. The three-year effort is scheduled to begin in January 2027 and run through December 2029. Working with collaborators Meg Estapa and Chaofan Chen at the University of Maine, Omand and her team will investigate whether machine learning models can predict the chemical composition of marine snow from characteristics that are visible in underwater images, such as particle size, shape, and transparency. If the approach succeeds, it could transform the way oceanographers interpret the enormous volumes of imagery their instruments already collect, turning pictures of drifting particles into quantitative estimates of what those particles are made of.

The stakes are considerable. Marine snow sits at the heart of the biological carbon pump, the set of processes by which the ocean sequesters carbon dioxide from the atmosphere. When phytoplankton die or are consumed, their remains aggregate into particles dense enough to sink. As Omand explains, these particles are rich in carbon and other nutrients, provide a key food source for organisms living in the deep ocean, and play an important role in marine carbon sequestration. The rate at which marine snow sinks, and the rate at which it is decomposed or consumed on the way down, determines how much carbon is locked away in the deep ocean and for how long. Better measurements of particle composition could therefore sharpen estimates of how the ocean will respond to a warming climate.

Underwater cameras have been part of the oceanographer’s toolkit for decades, and modern instruments can capture images of thousands of individual particles in a single deployment. Cameras mounted on autonomous floats and profilers can now document particle distributions across entire ocean basins. What the images cannot generally reveal, however, is chemistry. A particle that appears as a translucent blob in a photograph might be a nutrient-rich aggregate of organic matter, a mineral ballast fragment, or a piece of microplastic. Distinguishing among these possibilities has traditionally required collecting physical samples, filtering them, and analyzing them in laboratories, a slow and expensive process that limits how much of the ocean can be characterized.

The labor involved in manual image analysis is not trivial. Estapa noted that one of her graduate students spent months classifying particles and identifying what appeared in underwater images. That kind of effort, repeated across the field, represents a bottleneck at precisely the moment when imaging technology is generating more data than human analysts can process. The researchers hope that AI can automate much of this classification work, allowing scientists to collect and interpret more information while spending more of their time on scientific analysis and discovery. Crucially, the team also plans to develop models in ways that help scientists understand how the models reach their conclusions, addressing a common criticism of machine learning approaches in the sciences.

The foundation of the project is a carefully constructed database that pairs images with ground truth. Omand’s role in the first year will center on collecting marine snow images and matching physical samples across a broad range of ocean environments, spanning coastal waters off Ghana, the equatorial Atlantic, the New England shelf, and the California current system. Through collaborations with the Monterey Bay Aquarium Research Institute and the University of Ghana, the team already collected samples from two of these sites during the summer of 2026. The resulting database will link each image to information about the geochemical properties of the corresponding particles, the amount of microplastic present, and the location where the samples were gathered.

Once that database is assembled, the researchers will use data from six major oceanographic field campaigns to test whether groups of particles visible in images can accurately predict their chemical composition. The logic is that particle appearance is not random. Aggregates formed from diatom blooms look different from fecal pellets produced by zooplankton, and mineral-rich particles scatter light differently from soft organic ones. If those visual signatures correlate reliably with chemistry, a trained model could estimate carbon content, nutrient loading, or plastic contamination directly from photographs, without waiting for laboratory analysis. Validating such predictions across diverse ocean regions, from productive coastal shelves to the oligotrophic open ocean, will be the critical test of whether the method generalizes.

One of the most socially resonant applications involves plastic pollution. By identifying and quantifying plastic particles alongside naturally occurring marine snow, the researchers hope to better understand how plastic debris moves from surface waters into the deep ocean and through marine food webs. Microplastics have been found in the deepest ocean trenches and in the tissues of animals at every level of the marine food chain, yet the pathways by which they travel downward remain poorly quantified. If sinking aggregates are a major transport mechanism, as several studies suggest, then an AI system capable of flagging plastic-containing particles in routine imagery could produce basin-scale maps of plastic export that no sampling program could match.

The same tools could illuminate the natural cycling of carbon and nutrients through the ocean. As Omand puts it, gaining detailed insights into marine snow particles, their associated communities, and the links to their environment will allow scientists to better predict the movement of carbon and nutrients in the ocean and the impact that changes may have on marine life. That predictive capacity matters at a time when ocean warming, acidification, and shifting plankton communities are altering the efficiency of the biological pump in ways that global climate models struggle to capture. Particle-level observations, scaled up through automated analysis, could provide the empirical grounding those models lack.

For Omand, the project also represents a broader shift in how ocean science handles data. Research vessels, autonomous floats, and seafloor observatories now generate imagery faster than any laboratory can interpret it, and the techniques developed for marine snow could extend well beyond oceanography, to any scientific field that relies on complex images that are difficult and time-consuming to interpret. In that sense, the humble flecks drifting through the deep ocean may end up teaching researchers far from the sea a lesson about extracting knowledge from pictures. The snowfall that carries the ocean’s carbon downward may, with the help of machines that learn to read it, carry scientific understanding upward.

Subject of Research: Using artificial intelligence to determine the chemical composition of sinking marine snow particles from underwater imagery

Article Title: Reading the ocean’s snowflakes: How AI decodes marine snow

Article References: Reading the ocean’s snowflakes: How AI decodes marine snow. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: marine snow, artificial intelligence, carbon sequestration, microplastics, oceanography, biological carbon pump, underwater imaging, machine learning, National Science Foundation, University of Rhode Island, deep ocean, nutrient cycling

The finding matters because the Mediterranean is not behaving like a passive body of water that warms slowly and uniformly. Instead, the sea is undergoing a basin-wide transformation whose tempo is increasing decade by decade, a pattern that challenges the assumption of steady, linear change. The upper 100 meters of the Mediterranean have in recent years been roughly two degrees Celsius warmer than their 1950 to 1999 average. Earlier studies had detected hints that surface warming was speeding up, but no one had previously quantified whether that acceleration extended beneath the surface across the entire sea.

The numbers behind the acceleration are striking. In some regions, warming is speeding up by as much as 0.3 degrees Celsius per decade with each passing decade, and each kilogram of seawater is gaining about a tenth of a gram of salt per decade, every decade. Since 2000, Mediterranean surface waters have warmed at about half a degree Celsius per decade, two to three times faster than the global ocean surface. That represents a dramatic shift from the second half of the twentieth century, when most of the basin warmed by less than a tenth of a degree per decade. In other words, the pace of surface warming has multiplied several times over within a single generation of measurements.

What makes the study unusual is the depth to which the acceleration penetrates. The authors report that warming and salinification are statistically significant down to three or four thousand meters, and that the speed-up itself reaches to about 2,500 meters. Deep waters in most seas change slowly, insulated from atmospheric variability by layers of stable density stratification. Detecting a statistically robust acceleration at these depths means the signal of rapid surface change is being actively transported downward, most likely through the formation and sinking of dense water, rather than emerging from slow, gradual diffusion alone.

The mechanism hinges on seawater density, which governs the sea’s deep circulation. Warming makes seawater lighter, while added salt makes it heavier. Because both temperature and salinity are rising simultaneously, the density of Mediterranean water has changed far less than either property individually. Over most of the basin, warming is prevailing, so surface waters are becoming lighter and mixing less readily with the water beneath them. But in the places where dense water forms, above all the Adriatic, the added salt is still keeping surface water heavy enough to sink during cold winter conditions, carrying the new warmth and salinity into the deep interior.

The Adriatic is one of only a few sites in the Mediterranean where cold winter winds cool the surface enough to trigger deep convection. The dense water generated there spreads southward into the central and eastern Mediterranean, and it transports dissolved oxygen that sustains marine life in the deep sea. Since 2000, the deep waters of the southern Adriatic have warmed about six times faster than those of a typical Mediterranean basin. Dense water is still forming, but the water that sinks is now warmer and saltier than it used to be, which means the entire deep reservoir of the eastern Mediterranean is being progressively reset toward a new, hotter, saltier state.

Whether the Adriatic deep-water formation will continue indefinitely is one of the open questions the study raises. If warming eventually outpaces the salinity gain, surface waters could become too buoyant to sink, weakening or shutting down the ventilation that supplies oxygen to the deep sea. The new study did not measure oxygen directly, but warmer water holds less dissolved gas, and oxygen declines have already been documented in parts of the Mediterranean. The ecological implications are compounded by geography: species attempting to escape the warming have nowhere to go, because the sea is closed to the north and the deep water is warming too, leaving few thermal refuges in an increasingly enclosed basin.

The Mediterranean offers an unusually clear preview of how quickly a sea can respond to the heat trapped by greenhouse gases. The world’s oceans have absorbed more than a quarter of humanity’s carbon dioxide emissions and over 90 percent of the excess heat, buffering the climate system from far more severe atmospheric warming. Because Mediterranean waters circulate between the surface and the depths roughly ten times faster than in the open ocean, the basin compresses processes that would take centuries elsewhere into decades. The authors argue that this makes the sea an early indicator of how rapidly larger ocean basins can transform, and a demonstration that predictions assuming steady, linear change systematically underestimate the true speed of ocean response.

Terzić and Vilibić had previously documented unprecedented warming and salinification in the deep southern Adriatic, along with a telling basin-wide signature: pools of unusually salty surface water appearing across the whole Mediterranean, whereas historically such extreme salinity was confined to the sea’s far eastern end. That observation motivated the present study, which set out to quantify the changes across the rest of the basin. The consistent acceleration found everywhere they looked, including at great depth, convinced the authors that they were observing a fundamental reorganization of the sea’s heat and salt budget rather than a regional anomaly.

Several questions now drive the next phase of the research. The authors want to determine what is driving the acceleration, and how much of it comes from the atmosphere, from shifting patterns of evaporation, rainfall and river discharge, and from the Atlantic inflow through the Strait of Gibraltar, which is the Mediterranean’s principal source of replenishing water. They also intend to test whether today’s climate models capture the observed speed-up, a question that determines how much confidence can be placed in the models’ projections for the region. As Terzić notes, the Mediterranean is small enough that scientists can watch a sea respond on timescales they can witness themselves, and what they are seeing is unprecedented change, a warning of how fast the ocean can transform that will persist for as long as economies continue to rely on fossil fuels.

Subject of Research: Accelerated warming and salinification of the Mediterranean Sea from the surface to the deep ocean

Article Title: The Mediterranean is getting warmer and saltier at an ever-faster pace — even thousands of meters below the surface

Article References: The Mediterranean is getting warmer and saltier at an ever-faster pace — even thousands of meters below the surface. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: Mediterranean Sea, ocean warming, salinification, deep circulation, Adriatic Sea, dense water formation, oceanography, climate change, Geophysical Research Letters, dissolved oxygen, thermohaline circulation, marine ecosystems

The paper grew out of The Fifth Forum on Artificial Intelligence Oceanography, held earlier this year in Jinan, in China’s Shandong Province. The forum convened more than 300 experts, scholars, and students drawn from more than 80 institutions, making it one of the most comprehensive gatherings to date at the intersection of marine science and machine learning. Across talks and panel discussions, a clear consensus crystallized among participants: the next stage of AI oceanography will increasingly depend on models that fuse the flexibility of learning algorithms with the governing principles of ocean physics. The authors of the paper distilled those discussions into a forward-looking assessment of where the field stands and where it must go.

At the heart of that assessment lies the concept of physics-constrained artificial intelligence. In conventional machine learning, algorithms sift through enormous volumes of historical observations and model simulations, extracting statistical patterns without any explicit understanding of the dynamical laws that produce them. That approach can yield impressive results when training data are abundant and conditions resemble the past. But the ocean is an unforgiving test bed. Observations beneath the surface are sparse and expensive to collect, and environmental conditions can shift in ways that leave a purely statistical model extrapolating blindly. Physics-constrained approaches counter this weakness by embedding physical knowledge, dynamical equations, and process-based relationships directly into AI learning and evaluation, so that every prediction the system makes must remain compatible with the known behavior of rotating, stratified, wind-driven fluids.

The practical payoff, the authors argue, is threefold: improved reliability, improved interpretability, and improved performance under precisely the conditions where conventional AI falters. When observations are scarce, physical constraints act as a form of regularization, narrowing the space of plausible solutions to those the real ocean could actually occupy. When environmental conditions diverge from the training distribution, embedded dynamics give the model a scaffold to lean on, reducing the risk of physically nonsensical outputs. And when scientists interrogate a prediction, a model organized around meaningful physical relationships is far easier to diagnose and trust than an opaque network whose internal representations may correspond to nothing in the real ocean.

Drawing on the forum’s presentations, the authors organized the field’s most promising applications into three complementary classes of problems. The first is reconstructing subsurface ocean states from surface information. Satellites can observe sea surface temperature, sea surface height, and surface salinity with global coverage, but the three-dimensional structure of the ocean beneath remains largely hidden. AI systems trained on historical observations and simulations can infer subsurface temperature and salinity fields from what is visible at the surface, and studies presented at the forum showed that incorporating physical knowledge into these reconstructions makes them more reliable and more useful for research and operations alike.

The second class is long-range climate prediction, exemplified by forecasting the El Niño Southern Oscillation. ENSO emerges from coupled interactions between the tropical Pacific Ocean and the atmosphere, and its swings reshape rainfall, temperature, and storm patterns across the globe. Machine learning models have recently matched or exceeded traditional dynamical forecast systems in some ENSO prediction tasks, but the forum’s participants emphasized that learning physically meaningful relationships among ocean and atmosphere processes is what elevates these forecasts from statistical tricks to genuine scientific tools. When an AI system captures the underlying coupled dynamics, its predictions become interpretable, its errors diagnosable, and its skill more likely to persist as the climate changes.

The third class is operational wave forecasting, a domain with immediate consequences for shipping, offshore engineering, coastal management, and marine hazard warning. Waves respond rapidly to winds and currents, and forecasting them demands models that are both fast and physically sound. AI-based wave prediction systems presented at the forum demonstrated that embedding wave dynamics and process relationships into learning frameworks can deliver the speed of machine learning without sacrificing the physical consistency that operational users require. In each of the three problem classes, the pattern was the same: physical knowledge was not a constraint on AI’s power but the source of its trustworthiness.

The authors are explicit about what comes next. In their view, the field must move beyond isolated demonstrations toward AI systems that systematically combine physical constraints, uncertainty estimation, and real ocean observations. Uncertainty estimation deserves particular emphasis, because a forecast that comes with an honest measure of its own confidence is vastly more valuable to decision-makers than one that does not. A shipping company rerouting around a storm, a coastal community preparing for a marine hazard, and a climate negotiator weighing long-term risks all need to know not just what a model predicts but how much that prediction can be trusted. Building calibrated uncertainty into physics-constrained AI is therefore a central challenge for the coming years.

The longer-term ambition articulated in the paper is equally clear. The authors state that the ultimate goal is to make AI a trustworthy complement to numerical ocean models, one capable of supporting faster and more reliable ocean forecasting, climate prediction, marine hazard warning, and scientific understanding. This framing matters. Rather than positioning machine learning as a replacement for the general circulation models and dynamical simulation systems that oceanographers have refined over decades, the authors envision a hybrid landscape in which AI accelerates computation, fills observational gaps, and extracts structure from data, while numerical models and physical theory supply the dynamical foundation that keeps learning algorithms anchored to reality.

The implications extend well beyond oceanography. The tension the paper addresses, between the pattern-finding power of data-driven models and the explanatory rigor of physics-based ones, is one of the defining questions of modern Earth system science. The ocean, with its sparse observations, slow memory, and outsized influence on climate, is arguably the harshest environment in which to resolve that tension. If physics-constrained AI proves itself in the deep blue, the lessons will carry directly into atmospheric science, hydrology, cryosphere research, and the broader effort to model a planet in flux. For now, the message from more than 300 researchers gathered in Jinan is unambiguous: the future of AI in ocean science belongs not to algorithms that merely memorize the sea, but to those that learn the motion of the ocean itself.

Subject of Research: Physics-constrained artificial intelligence for ocean forecasting and climate prediction

Article Title: AI needs to learn the motion of the ocean, researchers report

Article References: AI needs to learn the motion of the ocean, researchers report. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: artificial intelligence, oceanography, physics-constrained machine learning, El Nino Southern Oscillation, wave forecasting, ocean modeling, climate prediction, subsurface ocean reconstruction, uncertainty estimation, numerical ocean models, marine hazards, AI oceanography forum

The physical setting explains why this region is so difficult to simulate. The Bay of Bengal is the largest semi-enclosed bay in the global ocean, and its northern boundary is closed, so the entire basin responds intensely to the twice-yearly reversal of the monsoon winds. From May to September, strong southwesterly winds dominate; from November to February, weaker northeasterlies take over. During the transitional months, westerly winds over the equatorial Indian Ocean generate disturbances that propagate toward the bay as equatorial and coastal Kelvin waves and westward-moving Rossby waves, reshaping the thermocline and upper-ocean stratification on seasonal to interannual timescales. On top of this remote forcing sits an extraordinary local input: the Ganges–Brahmaputra–Meghna river system discharges more than 4.4 × 10¹¹ cubic meters of freshwater each year, spreading a buoyant, sediment-laden lens across the shelf that dominates the hydrography up to roughly 21.5°N. This freshwater cap produces intense haline stratification, a barrier layer between the halocline and the isothermal layer, and, in winter, temperature inversions in which warm subsurface water becomes trapped between cooler surface and deeper layers.

Those inversions are a signature feature of wintertime hydrography in the northern bay, and the CTD profiles captured them in remarkable detail. Nine of the twelve coastal stations showed temperature inversions, and several stations, including stations 4, 5, 6, 9, and 12, displayed multiple inversion layers stacked at different depths. At stations 5 and 6, a sharp thermal inversion appeared near 15 meters in the in situ data, yet the model completely missed the feature. At stations 9 and 10, small but distinct mid-depth temperature drops were present in the observations but absent from the model output. Station 3 revealed a mixed layer only 4 to 5 meters deep, while the model showed no clear mixed-layer signature at all. The largest discrepancy occurred at station 12, where the observed mixed layer reached nearly 19 meters but the model profile ended around 11 meters. Only a handful of stations, such as stations 2 and 8, showed broadly similar profile shapes between observations and model, and even there the agreement was only approximate.

The coastal observations also revealed how shallow the winter mixed layer really is on this shelf. Mixed-layer depths from the CTD casts ranged from just 4 to 12 meters, with a mean of 5.18 meters at the northeastern stations and 10.31 meters at the northwestern stations. The authors attribute the shallower northeastern mixed layers to stronger near-surface stratification, while the relatively deeper northwestern values suggest somewhat greater vertical mixing or weaker surface stabilization near the river mouths. A time series of mixed-layer depth averaged over the northern bay showed the layer deepening slightly to a peak of about 11.4 meters around 8 February before shoaling again through late February and March. The model, by contrast, barely varied at all, holding steady between 9.8 and 9.9 meters throughout the entire three-month period and capturing none of the observed fluctuation. In station-by-station comparisons, the model generally overestimated mixed-layer depth, exceeding observations by roughly 3 to 5 meters at stations 1, 5, and 6, while underestimating at stations 11 and 12.

Offshore, the picture was somewhat better but still flawed. The team compared three Argo float profiles, located at 17.50°N 88.50°E, 18.50°N 90.50°E, and 19.50°N 92.50°E, against the model for January through March. Below about 80 to 100 meters, the model performed reasonably well, reproducing the depth of the well-developed thermocline and the general stratification, and both datasets showed a gradual deepening of the thermocline from January to March. In the upper 50 meters, however, the model diverged sharply from the floats. Argo revealed strong salinity-driven stratification and wintertime temperature inversions that the model did not fully reproduce, and collocated comparisons at 5 meters depth showed a consistent warm near-surface temperature bias alongside a tendency to underestimate salinity. Density differences followed the same pattern, with negative density biases in freshwater-dominated regions. The authors caution that because only three Argo sites were available over a short seasonal window, these bias patterns should be read as preliminary and location-specific rather than a complete regional assessment.

The broader oceanographic context of the 2020 winter helps explain the observed structures. Sea-level anomaly remained negative across the entire northern bay, ranging from about −0.32 to −0.36 meters in January and February and deepening to roughly −0.44 meters by March, with the strongest negative anomalies concentrated near the mouths of the Ganges–Brahmaputra system. Surface currents showed only weak variability, but subsurface currents told a different story: at 55 meters depth, currents shifted direction considerably and strengthened through the season, while at 21 meters they remained weak and variable. These depth-dependent current changes align closely with the thermal features seen in vertical sections, including warm subsurface layers of 25.4 to 26.6 °C trapped between cooler 23 to 24 °C waters in January and February, a cold intrusion between 90.4°E and 91.6°E that strengthened into March, and a downward penetration of warm water between 89.4°E and 90.2°E in February. Hovmöller diagrams confirmed a sudden cooling of roughly 2 to 2.5 °C around 21.3 to 21.5°N in February, followed by broad warming to about 26 °C in March as the winter monsoon weakened and wind stress declined from peaks of 0.04 to 0.054 pascals in the northeastern bay.

Why does a global model with a roughly 9-kilometer horizontal resolution and 5-meter vertical spacing near the surface struggle so much here? The authors point to a combination of factors rather than a single culprit. The model’s discrete vertical levels inherently smooth thin mixed layers, sharp thermohaline gradients, and narrow inversion layers that the CTD, sampling at roughly 0.2-meter intervals, resolves easily. Its global configuration and tripolar ORCA12 grid may not adequately represent regional subsurface processes, and the satellite datasets used to set initial conditions are often poorly suited to coastal and estuarine zones, forcing the use of pseudo-observation methods that can miss delicate features. Complex deltaic bathymetry, tide–river interactions, and the sheer magnitude of freshwater discharge add further difficulty. Importantly, the authors note that the region’s observational scarcity itself weakens the data assimilation that models depend on, although their analysis does not establish a direct causal link between sparse data and model bias. The result is a model that treats the northern shelf as an averaged, interpolated approximation of a far more intricate reality.

The observational gap itself is sobering. Despite the Argo program’s more than two million temperature–salinity profiles since 1999, fewer than 50 collocated Argo observations were reported for the extreme northern bay between 2015 and 2019, because the floats simply cannot sample the shallow Bangladesh shelf where freshwater input, sediment loading, and boundary currents are strongest. The in situ record consists almost entirely of short campaign-based surveys: a 2018 Nansen cruise with 38 CTD stations, a winter 2016 campaign with 15 casts, the 2020 dataset of 12 profiles used here, and roughly 60 nearshore stations near Cox’s Bazar collected in 2021–2022. To the authors’ knowledge, none of these CTD data are openly accessible, making their twelve profiles, to their knowledge, the only open-source coastal CTD data available for the region. Satellites offer no easy substitute, since they measure mainly the surface and often lack reliable algorithms for optically complex coastal waters, and validating satellite products requires exactly the kind of in situ data that is missing.

The stakes extend well beyond academic model evaluation. Two of the four major fishing grounds of the Bay of Bengal lie along the northeastern Bangladesh coast, suggesting a possible connection between episodic upwelling, nutrient supply, and fisheries distribution, yet the relationships among monsoon forcing, stratification, and nutrient delivery remain poorly understood precisely because strong stratification may limit vertical nutrient transfer. The bay’s freshwater-dominated upper ocean also shapes air–sea heat fluxes, cyclone response, and the monsoon system itself, so errors in simulating its mixed layer and barrier layer can propagate into regional climate predictions. The authors argue that continuous, large-scale, integrated monitoring of the northern bay is essential, not only for characterizing its subsurface structure but for building a regional model capable of accurate forecasting. Until such sustained observations exist, they conclude, the northern Bay of Bengal will remain what their title calls it: a blind spot, where even the best global models cannot see the delicate, fast-changing structures that define this extraordinary shelf sea.

Subject of Research: Winter upper-ocean hydrography and model performance in the data-scarce northern Bay of Bengal

Article Title: Inadequate in situ observations make the northern Bay of Bengal a blind spot for models

Article References: Inadequate in situ observations make the northern Bay of Bengal a blind spot for models. (n.d.). https://doi.org/10.1007/s44289-026-00148-y

Image Credits: AI Generated

DOI: 10.1007/s44289-026-00148-y

Keywords: Bay of Bengal, oceanography, CTD profiles, Argo floats, mixed layer depth, thermal inversion, barrier layer, monsoon, ocean modeling, CMEMS reanalysis, freshwater stratification, in situ observations

Tracking glacial melt in the ocean is deceptively difficult. When ice shelves and glacier termini discharge freshwater, whether as basal melt from floating ice or as subglacial outflow at grounding lines, that water mixes almost immediately with ambient seawater. Oceanographers traditionally quantify the meltwater fraction using tracer-based calculations that require predefined source water types, known as end members. In the classic approach, an analyst assumes the ocean can be described as a mixture of a small number of pure inputs, for example warm deep water, winter-modified shelf water, and pure glacial melt, each with known temperature and salinity. The meltwater fraction is then inferred from the leftover properties that cannot be explained by the mixing of those assumed sources.

The problem is that the answers depend heavily on the choices made. Pick a different deep-water definition, adjust the salinity of the meltwater end member, or allow a glacial ice end member in addition to liquid melt, and the estimated freshwater fractions can shift substantially. In regions with complex hydrography, where Antarctic Bottom Water formation, modified Circumpolar Deep Water intrusions, and seasonal sea-ice processes all compete to shape water properties, the ambiguity grows worse. East Antarctica, with its thousands of kilometers of ice front and sparse observations, has been especially vulnerable to these methodological uncertainties, leaving the meltwater budget of the region poorly constrained.

The new study sidesteps the end-member problem entirely. Rather than prescribing source water types and solving for their proportions, the researchers developed a parameterization that identifies glacier-derived freshwater directly from the structure of the hydrographic data itself. The technique exploits the fact that glacial melt alters temperature and salinity along characteristic lines in property space: because meltwater enters the ocean at the freezing point and carries negligible salt, its addition moves water masses in predictable directions in temperature-salinity coordinates. By parameterizing these trajectories without fixing the end points, the method estimates the freshwater contribution at every measured depth, producing not just a surface map but a three-dimensional reconstruction of the meltwater field.

The reconstruction is built from the vast archive of hydrographic observations collected across the East Antarctic shelf and slope, including conductivity-temperature-depth profiles, seal-mounted sensor data, and ship-based measurements gathered over multiple decades. Each profile is processed to separate the meltwater signal from other processes that also modify salinity, such as sea-ice formation and melting, precipitation, and the intrusion of off-shelf water masses. The end-member-independent framework then assembles these individual column estimates into a continuous three-dimensional field, resolved in longitude, latitude, and depth, that captures the horizontal pathways and vertical distribution of glacier-derived freshwater around the continent’s eastern half.

The resulting picture is striking. Meltwater is not distributed uniformly along the coast. Instead, the reconstruction shows concentrated lenses and layers of freshwater that accumulate at intermediate depths, typically well below the surface, where melt-laden water spreads neutrally according to its density. Along several major glacier systems, plumes of meltwater extend tens to hundreds of kilometers offshore, following the contours of shelf banks and canyon systems that steer the flow. In some locations the freshwater signal reaches the upper slope, hinting that glacial melt from East Antarctica may be exported into the broader Southern Ocean circulation rather than being trapped locally over the shelf, as older, two-dimensional assessments often implied.

These vertical details carry significant implications for ocean physics and climate. Freshwater stabilizes the water column by reducing surface density, which suppresses vertical mixing and can alter the formation of dense shelf waters that ultimately feed Antarctic Bottom Water, a key component of the global overturning circulation. By quantifying where melt accumulates at depth, the reconstruction allows scientists to test whether meltwater is interfering with bottom-water formation sites, potentially weakening the engine that ventilates the deep ocean and stores carbon and heat on centennial timescales. The three-dimensional view also provides essential validation data for ocean and coupled climate models, which historically have struggled to represent meltwater pathways realistically and often rely on crude runoff schemes at the ice-ocean boundary.

The methodological advance is as important as the observational findings. Because the parameterization does not require users to specify source water properties, it can be applied consistently across regions and through time, enabling fair comparisons between sectors of Antarctica and between different observational eras. Consistency is precisely what large-scale budget studies need: aggregating meltwater estimates produced with different end-member choices has been a persistent obstacle to constructing a continent-wide picture. An end-member-independent approach also reduces the risk of circular reasoning, in which assumptions about meltwater properties determine the meltwater fraction that is then used to infer melt rates. The authors show that their framework yields robust meltwater distributions under a range of environmental conditions, offering a template that can be transferred to other glacier-influenced seas.

For East Antarctica specifically, the study arrives at a pivotal moment. Long considered more stable than the marine-terminating glaciers of West Antarctica, the eastern ice sheet is increasingly showing signs of change, with warm modified deep water reaching the flanks of some major ice shelves and several basins identified as potential candidates for future accelerated retreat. A reliable reconstruction of where glacier-derived freshwater already enters the ocean provides both a baseline against which future change can be measured and a diagnostic of which systems are presently discharging melt at elevated rates. If meltwater export from the region strengthens, the three-dimensional fields produced by this method will help determine how quickly that signal propagates into the abyssal circulation.

The work also demonstrates how reanalysis of existing observations can yield new science without new expeditions. Decades of shipboard hydrography and the growing record of instrumented seals have created an underexploited treasure trove for the Southern Ocean; the challenge has been extracting subtle signals, like glacial freshwater, from noisy, unevenly sampled data. By turning a long-standing methodological weakness, the dependence on assumed source waters, into a solved problem, the researchers have converted scattered profiles into a coherent, multidimensional dataset of one of climate science’s most consequential tracers. As observations accumulate and parameterization techniques mature, the approach promises continuously updated maps of Antarctic meltwater, giving scientists and policymakers a clearer view of how the ice sheet, the ocean, and the global climate system are entangling beneath the surface of the far South.

Subject of Research: Three-dimensional mapping of glacier-derived freshwater in East Antarctic coastal waters using an end-member-independent hydrographic method

Article Title: Three-dimensional reconstruction of glacier-derived freshwater in East Antarctica using an end-member-independent hydrographic parameterization

Article References: Watanabe, Y. W., Hirano, D., Ohashi, Y., Sugita, M., Nakano, Y., Makabe, R., & Mizobata, K. (2026). Three-dimensional reconstruction of glacier-derived freshwater in East Antarctica using an end-member-independent hydrographic parameterization. Nature Communications, 17(1), Article 9498. https://doi.org/10.1038/s41467-026-77441-z

Image Credits: AI Generated

DOI: 10.1038/s41467-026-77441-z

Keywords: East Antarctica, glacial meltwater, hydrography, ice shelf melt, Southern Ocean, Antarctic Bottom Water, temperature-salinity analysis, sea-level rise, ocean circulation, water mass analysis, climate change, oceanography

The research, published in the Journal of Geophysical Research: Machine Learning and Computation, introduces a technique called Sparse Discrete Empirical Interpolation Method, or S-DEIM. It was developed by Mohammad Farazmand, associate professor of mathematics at NC State, together with graduate student Louisa Ebby and a team of undergraduate researchers from institutions across the United States. According to the authors, the method reconstructs high-resolution global sea surface temperature fields from as few as 100 in situ observations, a sampling density that amounts to just 0.2 percent of the full spatial grid. Even with so little direct information, more than 90 percent of the S-DEIM estimates landed within one degree Celsius of the true values in the team’s benchmark tests.

The problem the researchers set out to solve is one that oceanographers and climate scientists have wrestled with for decades. Federal agencies such as the National Oceanic and Atmospheric Administration have long relied on combinations of complicated differential equations to estimate temperatures across the vast unmonitored stretches of ocean. These model-based approaches are rigorous, but they are computationally demanding and depend on physical assumptions that may not hold perfectly across every ocean basin and season. Meanwhile, the explosive growth of machine learning has produced an alternative family of tools that can learn patterns directly from data, but at a steep price: deep neural networks often require hours of training on powerful hardware and enormous quantities of data before they can make a single useful prediction.

Somewhere in between sits the Discrete Empirical Interpolation Method, an established technique that the new work builds upon. Rather than modeling the ocean purely from physical first principles, DEIM specifies a basis, essentially a compact library of spatial patterns that jointly encode the structure of the temperature field being estimated. Given a handful of actual measurements, the method selects which of these patterns to activate and with what weights, producing a full-field estimate from sparse data. The approach is elegant and efficient, but it has a well-known weakness. When the available observations are truly sparse, as they typically are in the open ocean, the estimates it produces degrade considerably, because the method struggles to determine which patterns best explain a scattering of disconnected data points.

The NC State team’s insight was to bring historical information to bear on precisely this weakness. S-DEIM augments the classical framework with a so-called kernel vector, a quantity for which no closed-form mathematical formula exists, estimated instead from the long historical record of observations. In practice, the reconstruction produced by S-DEIM consists of two complementary terms. The first is computed from instantaneous in situ measurements using empirical interpolation, anchoring the estimate to what sensors are actually reporting right now. The second is learned from the historical time series using recurrent neural networks, which are particularly well suited to capturing how patterns in the data evolve over time. The marriage of the two allows the method to lean on decades of accumulated knowledge about ocean behavior while still respecting the fresh, if sparse, observations streaming in.

To train and test the method, the researchers used NOAA’s weekly high-resolution sea surface temperature dataset spanning 1989 through 2021, a record covering more than three decades of global ocean variability. The final year of the record, from January 2022 through January 2023, was withheld from the models entirely and reserved as a blind test. The team then asked S-DEIM, the classical DEIM method and a high-performing convolutional neural network to predict the sea surface temperatures for that unseen year, and compared their outputs against the actual historical data. This head-to-head design provided a rigorous measure of how each technique would perform under realistic conditions, where the future is genuinely unknown and the data available is sparse.

The results were striking. S-DEIM proved roughly 40 percent more accurate than DEIM, a substantial leap over the method it directly extends. More surprisingly, it also edged out the convolutional neural network, delivering estimates about 2 percent more accurate than the best AI model in the comparison. The efficiency gap was even more dramatic. Training the recurrent neural network at the heart of S-DEIM took approximately one minute, a one-time offline step, whereas the convolutional neural network required an hour and a half to train. Once trained, S-DEIM generates its full reconstructions in less than a second, making the approach practical for operational settings where forecasts must be produced continuously and quickly.

The method also displayed a robustness that matters greatly for real-world deployment. Sensor networks in the ocean are rarely arranged optimally; instruments drift, fail and are deployed wherever ships happen to travel. When the researchers distributed the sensors randomly rather than in favorable positions, the reconstruction error deteriorated by only 1 to 2 percent, suggesting that S-DEIM does not depend on carefully engineered measurement placements to deliver its accuracy. That resilience, combined with its computational thrift, makes the method attractive for agencies monitoring the ocean with limited and unevenly distributed instrumentation, and it opens the door to assimilating streaming observations in near real time.

The implications extend in two directions at once. In the short term, accurate and rapidly computed sea surface temperature fields feed directly into weather forecasting, where ocean conditions influence storm tracks, intensity and precipitation patterns on timescales of days to weeks. In the longer term, the same fields underpin climate models that track how the ocean absorbs and redistributes heat over decades. A tool that can deliver high-resolution temperature reconstructions from a sliver of the usual data, at a fraction of the computational cost, could meaningfully lower the barrier to both endeavors. The work also grew out of a National Science Foundation supported Research Experience for Undergraduates, with co-authors Cassidy All of the University of Colorado Boulder, Kevin Ho of Mississippi State University, Maya Magnuski of Bard College and Christopher Nicolaides of Indiana University contributing to the study alongside the NC State team.

Farazmand and his colleagues emphasize that this is not the end of the road. The team hopes to continue improving the accuracy of S-DEIM, and the framework’s flexibility suggests room for refinement, from richer historical models to better handling of measurement noise. For now, the study makes a compelling case that when data is scarce, a thoughtfully designed hybrid of classical interpolation and lightweight learning can beat brute-force deep learning on its own terms. In a field where every degree matters and every observation counts, S-DEIM offers a reminder that sometimes the smartest algorithm is not the biggest one, but the one that knows how to make the most of very little.

Subject of Research: A sparse-data interpolation method for rapidly reconstructing global sea surface temperatures from limited in situ observations.

Article Title: New method estimates sea surface temps quickly and accurately

Article References: New method estimates sea surface temps quickly and accurately. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: sea surface temperature, S-DEIM, data assimilation, machine learning, recurrent neural networks, NOAA, climate modeling, oceanography, sparse data, empirical interpolation, weather forecasting, North Carolina State University

The event grew from Argentina’s long-running Oceanography Week, established in the late 1970s, and became the National Marine Sciences Conferences in 1989 as researchers sought a broader forum spanning physical oceanography, marine biology and related disciplines. Since 2003, the triennial meeting has rotated among Argentine coastal cities; in 2025, it returned to Puerto Madryn after nearly two decades. The organizing effort involved researchers from several CONICET institutes and three higher-education institutions, creating a national network that linked oceanographers with biologists, technologists, social scientists, managers and representatives of economic sectors. For the first time, the scientific community was invited to propose thematic sessions, allowing emerging priorities to help shape the program rather than relying solely on a fixed institutional structure.

The resulting program included 37 thematic scientific sessions, 12 keynote lectures, 10 workshops, eight roundtables, a discussion panel and six training courses. In total, participants delivered 592 presentations: 294 ten-minute oral talks on site and 298 three-minute virtual speed talks. Replacing conventional printed posters with online presentations was intended to reduce material waste and the meeting’s carbon footprint while broadening participation. About 88 percent of attendees participated in person despite difficult economic conditions, and students made up more than half of the audience. Researchers came from across Argentina and from Uruguay, Chile, the United States, Mexico, Spain, the United Kingdom, Poland and Australia, giving the meeting a regional base with international reach.

Many of the scientific themes converged on the idea that ocean ecosystems cannot be understood through isolated disciplines. Sessions on physical, chemical and biological oceanography combined satellite observations, numerical models and measurements collected in the sea to investigate ocean structure, metabolism and variability. Marine microbiology and plankton research focused on organisms that drive food webs and regulate the movement of carbon and nutrients. One keynote examined the “viral engine” concept, in which viruses infecting marine phytoplankton influence microbial mortality and the recycling of matter. Another described the nitroplast, a nitrogen-fixing organelle associated with the marine microorganism UCYN-A and the alga Braarudosphaera bigelowii, highlighting an evolutionary development with implications for understanding nitrogen cycling in the ocean.

Climate change emerged as a force operating across scales, from the physiology of individual organisms to the circulation of the continental shelf. Presentations considered how phytoplankton, invertebrates and vertebrates respond biochemically and physiologically to environmental stress, and how those responses may affect ecosystem health, fisheries and aquaculture. Research on biodiversity addressed intertidal habitats, deep-sea ecosystems, ecological networks, trophic relationships, functional traits and biological invasions. A keynote drawing on the BioTIME database discussed rapid compositional turnover in marine communities linked to climate change, even where overall species richness appears comparatively stable. That distinction matters: an ecosystem can retain a similar number of species while the identities and ecological roles of those species change, potentially altering resilience and ecosystem functioning.

Regional circulation was another central concern. A keynote on the Southwest Atlantic shelf used observations and high-resolution climate modelling to examine how changes associated with the Southern Annular Mode and future emissions scenarios could modify circulation and exchanges between the deep ocean and the Patagonian continental shelf. Storm waves and surges on the Argentine shelf and in the Río de la Plata were studied through numerical simulations combined with observations, improving understanding of how extreme events are generated, propagated and connected across oceanic and coastal environments. Such physical processes affect the transport of heat, sediments, nutrients and pollutants, and they help determine where organisms can live and how human activities are exposed to marine hazards.

Human pressures formed a second major thread. Marine pollution sessions examined biological indicators, anthropogenic particles, persistent organic pollutants and the ecological consequences of contamination. Roundtables on microplastics considered evidence from multiple coastal and marine environmental matrices, as well as possible ecological, economic, health and cultural effects. A workshop explored phycoremediation, using algae or other photosynthetic organisms as a nature-based approach for treating nutrient- and organic-rich wastewater from urban, industrial and fisheries activities. Other discussions addressed marine biological invasions, with emphasis on shipping as a vector, early detection and coordinated prevention between Argentina and Chile. These topics point toward management strategies that combine monitoring, ecological research and action before damage becomes difficult to reverse.

Fisheries, aquaculture and the blue economy were discussed as socio-ecological systems rather than merely sources of production. Contributions examined sustainability and governance in industrial fisheries, as well as the social and regulatory challenges facing artisanal and recreational fisheries in coastal communities. Sessions on San Jorge Gulf and Península Valdés considered pathways toward formalization, while a roundtable on the South Atlantic’s adjacent area linked fisheries and conservation with geopolitics and international relations. Marine spatial planning, ecosystem-based management and coastal governance were also examined through case studies including “Blue Holes,” water-filled vertical openings in carbonate rock with distinctive morphologies, ecologies and water chemistry. These discussions emphasized that scientific evidence must be connected with institutions, local knowledge and decision-making if ocean policies are to work in practice.

Technology and capacity building rounded out the meeting’s forward-looking agenda. Researchers presented work involving marine genomics, biotechnology, hydroacoustics, scientific diving, remote sensing, spatial analysis and numerical modelling. Workshops addressed sustained marine observation in the Argentine Sea and Antarctica, identifying scientific, technological and institutional gaps that limit knowledge of ocean change. Training courses covered aquatic sampling, ultrasound techniques in octopus and flounder, QGIS and R for spatial data analysis, scientific illustration and academic English. A new code of conduct, developed by a working group on inclusion, diversity, equity, accessibility and language, established standards for a safer and more collaborative environment. The next National Marine Sciences Conference and Oceanography Colloquium is scheduled for December 2027 in Mar del Plata, where organizers plan to continue building the regional networks needed to study and protect a changing ocean.

The meeting report is valuable as a map of research capacity as well as a record of presentations. Its breadth shows that Southwest Atlantic marine science is increasingly organized around linked systems: circulation influences the delivery and retention of nutrients; nutrient availability shapes plankton communities; plankton supports food webs; and biological activity feeds back into carbon and nutrient transformations. Connecting these processes requires observations collected at different temporal and spatial scales, together with models and laboratory measurements that can be compared rather than developed in isolation.

This integration is particularly important on continental shelves, where land, atmosphere, open ocean and seabed interact over relatively short distances. Estuaries and coastal waters receive material from rivers and human activities, while tides, storms and shelf circulation redistribute it. The same transport pathways can move nutrients that sustain productivity, sediments that alter habitats, and contaminants or introduced organisms that create ecological risks. Treating these as separate issues can obscure their common physical drivers. The conference’s combination of coastal science, oceanography, pollution research and management therefore provides a framework for asking how one intervention or environmental change may produce several consequences at once.

Biological measurements add another layer of interpretation. Species counts alone may not reveal whether ecosystem functions are being maintained, because organisms with different traits can replace one another while total richness changes little. Studies of physiology, trophic relationships, ecological networks and genomics can help identify which changes affect energy transfer, reproductive success, stress tolerance or vulnerability to disturbance. These approaches also make it possible to connect individual responses with consequences for fisheries, aquaculture and conservation. In this context, biodiversity monitoring is not simply an inventory exercise; it can serve as an early indication of altered ecosystem processes.

The emphasis on observation infrastructure has practical significance because many marine questions cannot be answered by occasional expeditions. Sustained measurements allow researchers to distinguish long-term trends from seasonal cycles, unusual storms or short-lived biological events. Combining ship-based sampling with remote sensing, hydroacoustics, autonomous or fixed observations, and numerical analysis can extend coverage across places that are difficult or expensive to visit regularly. The report’s attention to scientific, technological and institutional gaps suggests that continuity, data comparability and coordination are as important as acquiring individual instruments. Without those foundations, evidence about change may remain fragmented even when many studies are being conducted.

Knowledge production was also presented as a social process. The inclusion of local and traditional knowledge, participatory research and co-production can help identify questions that matter to coastal communities and reveal changes that are not captured by standardized surveys. It can also improve the feasibility and legitimacy of management measures, especially where conservation objectives intersect with fishing, tourism, shipping or other uses. The code of conduct and training activities complement this scientific agenda by supporting the conditions needed for collaboration across career stages, institutions and national boundaries. Taken together, the meeting portrays regional ocean science as both an analytical enterprise and a long-term public infrastructure for responding to environmental change.

Subject of Research: Marine science research and collaboration in the Southwest Atlantic Ocean

Article Title: Recent advances in Southwest Atlantic Ocean Marine Sciences: outcomes from the XII National Marine Sciences Conferences and XX Oceanography Colloquium

Article References: Barbieri, E. S., Argüelles, M. B., Torres, A. I., & Giarratano, E. (2026). Recent advances in Southwest Atlantic Ocean Marine Sciences: outcomes from the XII National Marine Sciences Conferences and XX Oceanography Colloquium. Ocean Microbiology, 2(1), Article 4. https://doi.org/10.1186/s44375-026-00010-8

Image Credits: AI Generated

DOI: 10.1186/s44375-026-00010-8

Keywords: Southwest Atlantic, marine science, oceanography, climate change, marine biodiversity, fisheries, marine pollution, ocean governance, Recent, advances, Southwest, Atlantic