Posts Tagged 'regionalmodeling'

Projected future of African marine ecosystems under climate change and stratospheric aerosol injection

Published 8 May 2026 Science Leave a Comment
Tags: , , , , , , ,

Stratospheric Aerosol Injection (SAI) has been proposed as a potential strategy to cool the planet. The ARISE-SAI-1.5 approach, which employes a moderate emission scenario, is simulated to limit future global warming to 1.5°C by injecting aerosols into the stratosphere in the year 2035. However, the climate response to this SAI scenario, particularly along the African coast, remains unclear. In this study, we investigate the potential impacts of climate change under the SSP2-4.5 scenario and ARISE-SAI-1.5 on regional African marine ecosystems through key biological (chlorophyll), physical (salinity, temperature), and chemical (nitrate, acidification, and dissolved oxygen) parameters. Our results indicate that climate change may reduce productivity in African coastal ecosystems, with chlorophyll concentrations decreasing between 10% and 62%. Sea surface temperatures are projected to rise by 1.5°C along the entire coast by 2069, while surface salinity increases up to 0.3 g/kg, except for a slight decrease of up to 0.1 g/kg along the Congolese-Angolan coast. This salinity dipole in the Gulf of Guinea results from enhanced precipitation and river discharge, reinforced by stratification that traps freshwater at the surface. Additionally, climate change drives ocean acidification and may expand the oxygen minimum zone in the Gulf of Guinea, with oxygen levels decreasing by 10%–30% at depths of 100–200 m. Although ARISE-SAI-1.5 may help reduce surface oxygen depletion, it may not significantly mitigate subsurface oxygen loss or continued acidification. Nevertheless, it may reduce some negative climate change impacts on marine ecosystems by stabilizing chlorophyll levels, sea surface temperatures, and salinity.

Plain Language Summary

Stratospheric Aerosol Injection is being explored as a way to cool the planet and limit future global warming, for instance, to 1.5°C in the scenario we explore here (ARISE-SAI-1.5). However, its effects on the ocean, especially along the African coast, are not fully understood. This study examines key factors such as chlorophyll, water temperature, salinity, and oxygen levels to assess changes in marine ecosystems. Our findings show that climate change could reduce productivity, with chlorophyll levels dropping by 10%–62%. Sea surface temperatures are expected to rise by 1.5°C by 2069, and salinity will increase along most coastal areas. The low-oxygen zone in the Gulf of Guinea may expand, making deep waters less habitable for marine life. While the SAI we study here helps slow oxygen loss near the surface, it does not prevent deeper waters from losing oxygen or the ocean from becoming more acidic. However, it can still reduce some harmful effects of climate change by stabilizing chlorophyll levels, temperatures, and salinity.

Continue reading ‘Projected future of African marine ecosystems under climate change and stratospheric aerosol injection’

Acidification in coastal waters of Adélie Land, Antarctica (1985–2025)

Published 29 April 2026 Science Leave a Comment
Tags: , , , , ,

Ocean acidification is expected to be particularly severe in Antarctic continental shelves due to enhanced anthropogenic carbon uptake in cold waters in response to rising atmospheric CO2, sea-ice retreat, freshening and climate-change feedbacks. Models suggest that undersaturated conditions with respect to aragonite (Ωar), a major form of calcium carbonate formed by marine species, could be reached as soon as 2052 for austral winter.  Here we present new ocean carbonate system observations from cruises conducted since 2010 in the Adélie Land coastal region in East Antarctica, along with data from a BCG-Argo float and results from a neural network model for the period 1985–2025. The region is a permanent CO2 sink and was most pronounced since 2006. The CO2 sink leads to a positive increase of surface water total CO2 concentrations (CT) (+0.44 ± 0.01 µmol.kg-1.yr-1) and to a progressive decrease of pH (-0.013 per decade) and Ωar (-0.035 per decade) for the winter season. The lowest surface Ωar of 1.2 was observed in winter 2024 from the float data, a critical limit for some marine species such as pteropod. A projection of the CT concentrations in the future, based on observed anthropogenic CO2 concentrations and emissions scenarios, suggests that aragonite saturation state (Ωar = 1) will occur in surface waters as soon as 2055 in the Adélie Land region, which is part of a larger area of East Antarctica proposed as a Marine Protected Area by the Commission for the Conservation of Antarctic Marine Living Resources since the early 2010s.

Continue reading ‘Acidification in coastal waters of Adélie Land, Antarctica (1985–2025)’

Future projections of compound events around the Main Hawaiian Islands

Published 15 April 2026 Science Closed
Tags: , ,

The consequences of overlapping environmental stressors — referred to as compound events — may be more harmful to marine ecosystems than as individual stressors. Using recently conducted submesoscale-permitting future projections for the Main Hawaiian Islands, we present the first assessment of future compound events for Hawaiian waters. Our analysis focuses on surface and sub-surface heat-stress, ocean acidification, and low-oxygen events and is based on three different greenhouse gas emission scenarios. We show that a large fraction of ocean around Hawai‘i will be subject to compound events in the near future. However, the projected event characteristics such as duration and intensity vary substantially across the region suggesting that potential ecosystem impacts may differ over short distances. Our results reveal that these spatial differences are mainly driven by considerably different magnitudes of natural variability in ocean physics and chemistry across the domain driven by mesoscale processes, while anthropogenic trends exhibit only minor spatial differences. Our analysis demonstrates that small-scale tidal variability can significantly mitigate compound events in near-shore regions including some designated Marine Protected Areas. Overall, our findings highlight the need for high-resolution numerical models as well as for an extended observation network for robust future projections of local extreme events.

Continue reading ‘Future projections of compound events around the Main Hawaiian Islands’

An interpretable machine learning approach for alkalinity reconstruction in the Mediterranean Sea

Published 9 April 2026 Science Closed
Tags: , , ,

Highlights

  • Genetic Programming provides interpretable alkalinity models for Mediterranean Sea.
  • Genetic Programming models capture typical alkalinity patterns and its finer-scale variability.
  • Genetic Programming matches or exceeds linear models while remaining interpretable.
  • Neural networks yield lowest errors but lack model transparency.

Abstract

Ocean acidification has significant impacts on marine ecosystems and human activities, and its understanding relies on an accurate characterization of the marine carbonate system, in which alkalinity plays a central role.

We propose a Machine Learning (ML) approach based on Genetic Programming (GP) to model alkalinity and apply this framework to the surface layers of the Mediterranean Sea. Our framework produces interpretable equations that capture alkalinity typical patterns and its finer-scale variability by inferring its relation with key physical and biogeochemical variables.

Results, supported by quantitative metrics and visual analyses, demonstrate that our method reliably reproduces the spatio-temporal variability of alkalinity with a high level of predictive accuracy when compared with in situ observations. Moreover, we use the derived alkalinity equations to produce gap-free 2D surface alkalinity maps using satellite data. The maps correctly capture spatial gradients, seasonal patterns, and riverine contributions, reinforcing the robustness of the proposed approach.

Continue reading ‘An interpretable machine learning approach for alkalinity reconstruction in the Mediterranean Sea’

Analysing the distribution and variability of dissolved inorganic carbon and alkalinity over the Bay of Bengal region using the coupled ocean biogeochemical modeling

Published 25 March 2026 Science Closed
Tags: , , ,

Highlights

  • High-resolution regional coupled ocean biogeochemical modeling in the Bay of Bengal.
  • Spatio-temporal variability of Dissolved Inorganic Carbon and Alkalinity is studied.
  • Aragonite (calcite) saturation depth in the Bay of Bengal is estimated.
  • ENSO and IOD events significantly influence surface DIC of the BoB region.

Abstract

A prototype high-resolution regional coupled ocean biogeochemical modeling experiment is carried out in the Bay of Bengal (BoB) region to study the distribution and spatio-temporal variability of Dissolved Inorganic Carbon (DIC) and Alkalinity (Alk) during the period 2000-2021. It is found that in the eastern as well as head BoB, the DIC concentration remains less (1.6-1.7 mol/m3) as compared to the south-west and west-central BoB, where the DIC concentration remains particularly high (>1.9 mol/m3). The highest (lowest) DIC concentration in the BoB remains in the Mar-April (Oct) months. The seasonal variability of the DIC and Alk is studied vis-à-vis seasonal changes in the currents and freshwater flux. The depth profiles of DIC, Alk, and DIC/Alk ratio are also investigated across different sections in the BoB. The DIC remains stratified in the BoB, and the stratification becomes much more pronounced on moving from south to north (and west to east) part of the model domain. The aragonite (calcite) saturation depth ranges between approx. 100-400 m (500-4000 m) in the BoB. The particularly high (>8.1) and low (∼8) pH values are found in the head BoB and southwest BoB, respectively. It is shown that the influence of El Nino – Southern Oscillation (ENSO) event on the surface DIC concentration over the BoB region is much stronger as compared to the Indian Ocean Dipole (IOD) event.

Continue reading ‘Analysing the distribution and variability of dissolved inorganic carbon and alkalinity over the Bay of Bengal region using the coupled ocean biogeochemical modeling’

Contrasting effects of river and erosion-derived inputs on Arctic Ocean acidification

Published 18 March 2026 Science Closed
Tags: , , , ,

Although the Arctic Ocean is relatively small in volume, its extensive coastline delivers large quantities of terrigenous material from rivers and coastal erosion. As a result, the Arctic Ocean is impacted more strongly by terrigenous material than most other parts of the global ocean. Yet the effect of this material on carbon cycling and ocean acidification remains poorly quantified. In this study, we use an ocean biogeochemical model driven by observation-based estimates of terrigenous carbon, alkalinity, and nutrients to evaluate their contribution to the mean state, depth pattern, and seasonal cycle of ocean acidification, as measured by the aragonite saturation state. Riverine alkalinity generally mitigates acidification, whereas organic carbon from coastal erosion intensifies it. Nutrients from both sources mitigate ocean acidification at the surface by stimulating primary production, but intensify it at depth through subsequent remineralisation. Together, riverine and erosion-derived inputs account for about 20–40 % of the seasonal variability in the saturation state of the surface ocean. This amplification of the natural seasonal cycle is primarily caused by an increase in the summertime maximum of the saturation state. Terrigenous inputs also reduce the Arctic Ocean’s capacity to absorb atmospheric CO2 by 17–25 %. Accurately representing carbon and nutrient inputs from rivers and coastal erosion in biogeochemical models is therefore important for reliable assessments of ocean acidification, ecosystem health, and carbon budgets in the Arctic Ocean.

Continue reading ‘Contrasting effects of river and erosion-derived inputs on Arctic Ocean acidification’

Modelling seawater pCO2 and pH in the Canary Islands region based on satellite measurements and machine learning techniques

Published 4 March 2026 Science Closed
Tags: , , ,

Recent advancements in remote sensing systems, combined with new machine-learning model-fitting algorithms, have enabled the estimation of seawater carbon dioxide partial pressure (pCO2,sw) and pH (pHT,is) in the waters around the Canary Islands (13–19° W; 27–30° N). Continuous time-series data collected from moored buoys and Voluntary Observing Ships (VOS) between 2019 and 2024 were used to train and validate the models, providing a robust observational basis for satellite-derived estimates.

Among all models tested, bootstrap aggregation (bagging) performed best, achieving an RMSE of 2.0 µatm (R2>0.99) for pCO2,sw and 0.002 for pHT,is. Multilinear regression (MLR), neural networks (NN) and categorical boosting (CatBoost) also showed good predictive skill, with RMSE values between 5.4 and 10 µatm for pCO2,sw (360–481 µatm) and 0.004–0.008 for pHT,is (7.97–8.07). Using the most reliable model, we identified an increasing trend in pCO2,sw of 3.51±0.31 µatm yr−1, exceeding the atmospheric CO2 growth rate (2.3 µatm yr−1), alongside an acidification trend of −0.003 ± 0.001 yr−1.

Over the 2019–2024 period, rising atmospheric CO2 and increasing sea surface temperatures (reaching up to 0.2 °C yr−1 during the unprecedented 2023 marine heatwave) likely contributed to these trends. The Canary Islands region shifted from a weak CO2 source (0.90 Tg CO2 yr−1) in 2019 to 4.5 Tg CO2 yr−1 in 2024. After 2022, eastern sites that previously acted as annual CO2 sinks became net sources.

Continue reading ‘Modelling seawater pCO2 and pH in the Canary Islands region based on satellite measurements and machine learning techniques’

Physics-guided machine-learning forecasting and analysis of carbonate changes in the surface Western Mediterranean

Published 3 March 2026 Science Closed
Tags: , , ,

Highlights

  • Physics-guided ML forecasts surface pCO2 and pH along a Western Mediterranean VOS line.
  • Day-ahead pCO2 is predicted with μatm-level RMSE; pH behaves nearly deterministically.
  • Boosted trees and sequence models retain skill under strict, deployable forecast conditions.
  • Explainable AI recovers dominant thermal control and air–sea CO2 gradient drivers.
  • Improved pCO2 forecasts directly reduce uncertainty in air–sea CO2 flux estimates.

Abstract

We introduce a hybrid, physics-guided machine-learning system for forecasting and explaining surface marine carbonate changes along a fixed Volunteer Observing Ship route between Gibraltar and Barcelona from 2019 to 2024. The dataset includes more than 90 high-frequency transects collected under ICOS/SOOP standards, containing underway pCO2/fCO2, pH (measured and derived), sea-surface temperature, and salinity. After applying consistent quality control and harmonizing the data in time and space, we combine physics-based carbonate diagnostics—such as the thermal/non-thermal decomposition (FASS) and first-order Taylor attribution of temperature, salinity, total alkalinity, and dissolved inorganic carbon sensitivities—with time-aware models including linear regressions, boosted trees, and sequence networks (1-D CNNs and LSTMs) trained on historical windows. We evaluate generalization and uncertainty through chronological splits, leave-one-year-out tests, and year-wise bootstrap sampling. With all current predictors available, day-ahead pH and pCO2 predictions reach near-optimal skill; pH behaves almost deterministically, while pCO2 achieves RMSE on the order of a few μatm. Even under stricter forecast conditions without real-time carbonate chemistry, boosted trees and sequence models maintain strong performance by exploiting persistence and seasonal timing. Model-explanation tools (SHAP, partial dependence) recover the expected carbonate drivers, highlighting dominant thermal effects and key roles of seawater CO2 state and air–sea gradients. Spatial–temporal diagnostics reveal amplified summer pCO2 peaks in the Alboran/northern Morocco region and out-of-phase pH patterns. Predicted fields are converted to air–sea CO2 flux using standard solubility and gas-transfer formulations, and propagated uncertainties show that improving pCO2 accuracy directly reduces flux uncertainty. The resulting air–sea CO2 fluxes exhibit a pronounced seasonal cycle, with summer outgassing reaching several mmol m-2 d-1 and winter uptake of comparable magnitude along the transect, while interannual variability dominates over 2019–2024 and no statistically robust long-term trend is detected; typical flux uncertainties are on the order of 0.1–0.2 mmol m-2 d-1. Altogether, this delivers an explainable, uncertainty-aware system ready for deployment, linking forecast skill to process understanding and CO2 exchange in a climate-sensitive corridor.

Continue reading ‘Physics-guided machine-learning forecasting and analysis of carbonate changes in the surface Western Mediterranean’

Decadal biogeochemical predictions for the bottom marine environment of the Northeast U.S. Continental Shelf

Published 27 February 2026 Science Closed
Tags: , , ,

The Gulf of Maine and the surrounding Northeast U.S. Continental Shelf are experiencing rapid marine environmental change arising from complex regional dynamics that challenge near-term (1–10 years) predictive capabilities for valuable living marine resources. Here, using a high-resolution regional ocean model, we demonstrate skilful decadal forecasts of ocean bottom habitat characteristics including bottom temperature, dissolved oxygen (O2), pH and aragonite saturation state (Ωar). Bottom temperature and pH predictions show substantial skill driven primarily by radiatively forced warming and carbon uptake trends, while bottom O2 and Ωar predictions benefit more from initialization due to stronger internal variability. Retrospective forecasts successfully predicted observed historical changes in water masses and environmental properties, including recent cooling/freshening transitions driven by replacement of Warm Slope Water with Labrador Slope Water. This water mass variability also modulates biogeochemical conditions and ocean acidification buffering capacity, with our recent forecasts indicating that benefits from the expected respite from rapid warming might be tempered by challenges posed by rapid acidification. The demonstrated predictability of coupled physical-biogeochemical processes supports developing integrated prediction systems for climate-informed marine resource management.

Continue reading ‘Decadal biogeochemical predictions for the bottom marine environment of the Northeast U.S. Continental Shelf’

Understanding the resilient carbon cycle response to the 2014–2015 Blob event in the Gulf of Alaska using a regional ocean biogeochemical model

Published 24 February 2026 Science Closed
Tags: , , ,

Marine heatwaves (MHWs), characterized by anomalously high sea surface temperatures, are occurring with increasing frequency and intensity, profoundly impacting ocean circulation, biogeochemistry, and marine ecosystems. The MHW known as the Blob, which persisted in the subarctic NE Pacific from 2014 to 2015, significantly affected surrounding ecosystems. Warming-induced solubility reduction is expected to raise the partial pressure of carbon dioxide (pCO2) in the surface water, causing outgassing of CO2 to the atmosphere. Outgassing of CO2 is another source of atmospheric CO2 in addition to anthropogenic fossil fuel burning. However, moored observations at Ocean Station Papa (OSP; 145° W, 50° N) shows a moderate decrease in oceanic pCO2 during the Blob, resisting the warming-induced outgassing of CO2. This response is opposite of what is expected from warming alone, and instead has been attributed to reductions in dissolved inorganic carbon (DIC), although the mechanisms driving this reduction have remained unclear. We employed a regional model that accurately reproduces the temporal variability of oceanic pCO2 at OSP to investigate the cause of decrease pCO2 during the Blob. The analysis of model outputs indicates that the observed oceanic pCO2 decline resulted from the offset between warming-induced solubility reduction (increasing pCO2) and weakened physical transport of DIC (decreasing pCO2), with the latter dominating. Both horizontal and vertical transports played important roles. The near-surface carbon budget over the broad region was primarily driven by changes in the vertical transport. The decrease in DIC during the Blob resulted from the suppression of upwelling of DIC-rich subsurface waters in the winter of 2013. In this period, the horizontal transport also contributed substantially to DIC reduction. In particular, at OSP, the effect of the horizontal transport was comparable to that of the vertical transport, reflecting the northward advection of low-DIC water masses. These findings indicate that changes in physical circulation were the primary driver of the moderately enhanced CO2 uptake observed during the Blob. This study provides a critical insight into the complexity of biogeochemical response to extreme warming events and underscores the importance of resolving physical transport processes in assessing oceanic carbon uptake during MHWs.

Continue reading ‘Understanding the resilient carbon cycle response to the 2014–2015 Blob event in the Gulf of Alaska using a regional ocean biogeochemical model’

Wind control of the interannual ocean‐biogeochemical variability in the South Atlantic Bight

Published 23 February 2026 Science Closed
Tags: , , , ,

Abstract

In the South Atlantic Bight (SAB), changes in the Gulf Stream (GS), particularly its strength and proximity to the coast, are thought to be primary factors determining the shelf-break upwelling rate. However, it is still not entirely clear if and to what extent those factors influence cross-shelf nutrient fluxes and shape the ocean biogeochemistry at interannual and longer timescales. Here, we use a high-resolution regional ocean-biogeochemical model and an ocean reanalysis product (1993–2022), along with a satellite-derived chlorophyll data set (1997–2022), to investigate the interannual ocean-biogeochemical variability in the SAB. Regional model outputs suggest that year-to-year changes in phytoplankton production are indeed largely driven by upwelling of cold and nutrient-rich water to the shelf-break. The upwelling variability, reflected in bottom temperature and vertically integrated production patterns, is strongly linked to surface velocity changes in the GS near the shelf break, but weakly related to the depth-integrated GS transport. The GS’s velocity changes, and the temperature and production anomalies, are well correlated to the alongshore wind stress, suggesting that local wind is the leading driver of the shelf-break upwelling variability at interannual timescales. Those relationships are also supported by circulation patterns from ocean reanalysis and satellite chlorophyll anomalies. Finally, examining the simulated shelf-slope interchanges in the carbonate system, we find that shelf-break upwelling significantly increases bottom acidification, a pattern linked to the low carbonate concentration in the slope waters. This study thus provides new insight for understanding and predicting GS and winds impacts on biogeochemical patterns from the SAB.

Plain Language Summary

The ocean current known as the Gulf Stream (GS) can induce upwelling of subsurface cold and nutrient-rich waters into the coastal margin of the South Atlantic Bight, influencing coastal temperature and phytoplankton growth. Previous studies suggested that the GS strength and its proximity to the coast are key factors determining the intensity of upwelling events. However, the degree to which these factors impact the year-to-year changes in phytoplankton production and other ocean properties remains unclear. Here we use numerical models of ocean currents and seawater biogeochemistry, as well as chlorophyll records derived from satellite measurements, to investigate that impact. The patterns showed that interannual changes in coastal temperature, phytoplankton production, water acidity, and dissolved oxygen are strongly modulated by upwelling changes in the outer edge of the continental margin (about 70 m depth in this region). This interannual upwelling variability is tightly coupled to variations in the surface alongshore GS velocity close to that outer edge, which is modulated by alongshore wind variability. Our study characterizes GS patterns associated with high and low productivity years, and highlights the role of surface wind as ultimate driver of the interannual upwelling variability in the South Atlantic Bight.

Key Points

  • A regional ocean model is used to investigate interannual variability of ocean-biogeochemistry in the South Atlantic Bight
  • Year-to-year changes in primary production, chlorophyll, and carbonate system patterns respond to shelf-break upwelling anomalies
  • Shelf-break upwelling is closely linked to the Gulf Stream velocity near the shelf break, modulated by alongshore wind variability
Continue reading ‘Wind control of the interannual ocean‐biogeochemical variability in the South Atlantic Bight’

Ocean acidification in Canada: the current state of knowledge and pathways for action

Published 23 February 2026 Science Closed
Tags: , , , , , ,

Ocean acidification (OA) generally receives far less consideration than other climate stressors and related hazards, such as global warming and extreme weather events. Canada is uniquely vulnerable to OA given its extensive coastal oceans, the oceanographic processes in its three basins, accelerated warming and sea-ice melt, and extensive coastal communities and maritime economic sectors. Canada’s coastline is also home to extensive and diverse First Nations peoples with distinct histories, rights, title, laws, governance and whose traditions and cultures are extrinsically linked to the sea. However, there are currently very limited pathways to support OA action, mitigation, and/or adaptation in Canada, particularly at the policy level. Here, we present a first synthesis of the current state of OA knowledge across Canada’s Pacific, Arctic, and Atlantic regions, including monitoring, modelling, biological responses, socioeconomic and policy perspectives, and examples of existing OA actions and efforts at local and provincial levels. We also suggest a step-wise pathway for actions to enhance the coordinated filling of OA knowledge gaps and integration of OA knowledge into decision-making frameworks. The goals of these recommendations are to improve our ability to respond to OA in Canada, and minimize risks to coastal marine environments and ecosystems, vulnerable sectors, and communities.

Continue reading ‘Ocean acidification in Canada: the current state of knowledge and pathways for action’

Nonlinear interactions of timing and amplitude biases in modeled Southern Ocean pCO2: the roles of dissolved inorganic carbon, total alkalinity, and sea surface temperature

Published 21 January 2026 Science Closed
Tags: , , ,

The Southern Ocean is a major sink for atmospheric carbon dioxide and critical to the current and future carbon cycle. This net annual CO2 flux reflects the balance between strong seasonal variability characterized by opposing periods of winter outgassing and summer uptake. Using a simple framework, we evaluate how model biases in both the amplitude and timing of dissolved inorganic carbon (DIC) and total alkalinity (TA) and in the amplitude of sea surface temperature (SST) impact simulated pCO2. We examine seasonal CO2 fluxes and pCO2 south of the Subantarctic Front in 42 Earth System Model and three state estimate simulations. Only 11 of the 45 simulations have a seasonal pCO2 cycle with a correlation of ≥0.7 to observed pCO2, while 26 have a correlation of <0. Four of the well-correlated models accurately represent the seasonality of SST, DIC, and TA, while TA biases compensate for DIC or SST biases in the other seven. DIC and SST amplitude biases are related to mixed layer (MLD) biases, with shallow MLDs, especially in the summer, correlated with larger amplitude DIC and SST cycles than observed. The amplitude of seasonal Net Primary Production is correlated to DIC and TA timing. We provide input on the main adjustments needed to correct the simulated pCO2 seasonality in each of the evaluated models. These findings highlight the difficulty and importance of capturing the seasonal processes influencing the carbonate system to correctly model and predict the Southern Ocean carbon sink and its response to a changing climate.

Continue reading ‘Nonlinear interactions of timing and amplitude biases in modeled Southern Ocean pCO2: the roles of dissolved inorganic carbon, total alkalinity, and sea surface temperature’

Explainable machine learning models for coastal pH forecasting at aquaculture-relevant thresholds in Eastern Canada

Published 19 January 2026 Science Closed
Tags: , , , ,

Highlights

  • Benchmark of ML models for coastal pHSWS forecasting.
  • Models trained on rare high-frequency data from Eastern Canada.
  • XGBoost balances sensitivity and precision at pHSWS < 7.75
  • SHAP shows Julian day dominance as composite environmental driver.
  • Promising low-cost framework for aquaculture acidification early warning.

Abstract

Ocean acidification poses a growing threat to marine ecosystems and aquaculture productivity, particularly in under-monitored coastal regions such as Eastern Canada. Existing pH prediction frameworks typically rely on multi-year records combining extensive carbonate chemistry, physical, and biological parameters. While these models can achieve high accuracy, their data requirements make them costly, complex, and challenging to implement for local, site-specific acidification forecasting in aquaculture contexts. To address this limitation, this study benchmarks several machine learning models for coastal pHSWS prediction using only three routinely measured environmental variables (temperature, salinity, sea level), from which we derived moving-average descriptors, local gradients, and two temporal indicators, resulting in a compact set of 11 input features. Six different models and a multivariate linear regression baseline were trained on one of the most complete and extended high-frequency datasets available (BSSS2018) and evaluated across four independent datasets: one from the same site but six months earlier (BSSS2017), and three from nearby bays in northeastern New Brunswick collected between 2017 and 2019. Among all tested models, XGBoost emerged as the most reliable and interpretable, achieving the best trade-off between sensitivity and precision at the operational acidification threshold (pHSWS < 7.75). Its performance remained acceptable within-site but declined across bays due to environmental and seasonal discrepancies, underscoring the importance of training data representativeness. SHAP-based explainability confirmed that Julian day was the dominant predictor, integrating the composite effects of seasonal environmental variability. Overall, this study demonstrates that using only low-cost, routinely measured features provides a promising foundation for short-term coastal pH forecasting, particularly for aquaculture monitoring needs. Despite limited inter-bay generalization, the proposed framework shows that interpretable machine learning models can deliver actionable early-warning insights under realistic data constraints. It constitutes one of the first data-driven benchmarks explicitly tested at aquaculture-relevant thresholds, highlighting a scalable and transparent approach toward operational acidification forecasting.

Continue reading ‘Explainable machine learning models for coastal pH forecasting at aquaculture-relevant thresholds in Eastern Canada’

Ocean acidification in Massachusetts bay and Boston harbor: insights from a 1-D modeling approach

Published 8 January 2026 Science Closed
Tags: , , , ,

Highlights

  • NeBEM, an ERSEM-based biogeochemical and ecosystem model, is established for the U.S. Northeast.
  • NeBEM provides process-based insights into carbonate system variability beyond the capability of empirical data-fitting methods.
  • Biological processes strongly influence TA and DIC variability in outer Massachusetts Bay.

Abstract

Massachusetts Bay (MB)/Boston Harbor (BH) in the northeastern United States has reduced buffering capability, making it highly vulnerable to ocean acidification (OA). We applied the U.S. Northeast Biogeochemistry and Ecosystem Model (NeBEM), integrating the unstructured grid, Finite Volume Community Ocean Model with a modified European Regional Seas Ecosystem Model (ERSEM), to investigate seasonal and interannual OA variability through one-dimensional (1-D) experiments. Objectives were to (a) evaluate model skill in reproducing observed seasonal cycles of OA-related variables, particularly pCO2 and pH, in shallow and deep regions, and (b) assess sensitivity to parameterizations and algorithms for calculating dissolved inorganic carbon (DIC), total alkalinity (TA), pCO2, and pH. The 1-D NeBEM reproduced variability of nutrients, dissolved oxygen, chlorophyll-a, pCO2, and pH at the deep outer bay site, where air-sea interactions dominate, but failed at the shallow inner bay site due to the absence of river discharge-driven advection. Of TA algorithms tested, the semi-diagnostic method best captured observed seasonal pCO2 variation, achieving the highest correlation and lowest root mean square error, although all methods performed similarly for pH. Comparisons with multi-linear regression methods showed that empirical models are highly sensitive to calibration set. Mechanistic analysis indicated that TA variability is mainly regulated by nitrification and net community production (NCP), while DIC variability is driven primarily by NCP. Atmospheric CO₂ loading was the first-order contributor to DIC change in magnitude. However, it has decreased in MB over the past two decades, in contrast to regional and global trends. Therefore, it is not a major driver of OA progression in this system.

Continue reading ‘Ocean acidification in Massachusetts bay and Boston harbor: insights from a 1-D modeling approach’

Water property variability into a semi-enclosed sea dominated by dynamics, modulated by properties

Published 15 December 2025 Science Closed
Tags: , , , ,

The biogeochemistry of the Salish Sea is strongly connected to its Pacific Ocean inflow through Juan de Fuca Strait (JdF), which varies seasonally and interannually in both volume and property flux. Long-term trends in warming, acidification, and deoxygenation are a concern in the region, and inflow variability influences the flux of tracers potentially contributing to these threats in the Salish Sea. Using ten years (2014–2023, inclusive) of Lagrangian particle tracking from JdF, we quantified the contributions of distinct Pacific source waters to interannual variability in JdF inflow and its biogeochemical properties. We decompose variability in salinity, temperature, dissolved oxygen, nitrate, and carbonate system tracers into components arising from changes in water source transport (dynamical variability) and changes in source properties (property variability). Observations in the region provide insight into source water processes not resolvable in the Lagrangian simulations, including denitrification and trace metal supply. Deep source waters dominate total inflow volume and drive variability in nitrate flux through changes in transport. Shallow source waters, particularly south shelf water, exhibit greater interannual variability and disproportionately affect temperature, oxygen, and [TA–DIC], driving change through both dynamical and property variability. This study highlights the combined roles of circulation and source water properties in shaping biogeochemical variability in a semi-enclosed sea, and how these roles differ between biogeochemical tracers. It provides a framework for attributing flux changes to specific source waters and physical and biogeochemical drivers, with implications for forecasting coastal ocean change under future climate scenarios.

Continue reading ‘Water property variability into a semi-enclosed sea dominated by dynamics, modulated by properties’

Compound marine heatwaves and acidity extremes in the Southern Ocean

Published 21 November 2025 Science Closed
Tags: , , ,

Abstract

Compound extremes of temperature and acidity that extend over substantial fractions of the water column can be particularly damaging to marine organisms, as they experience not only additional stress by the potentially synergistic effects of these two stressors, but also a reduction in habitable vertical space. Here, we detect and analyze such column-compound extremes (CCX) in the Southern Ocean between 1980 and 2019, and characterize their duration, intensity, and spatial extent. To this end, we use daily output from a hindcast simulation of the Regional Ocean Modeling System (ROMS), coupled with the Biological Elemental Cycling (BEC) model. We first detect extremes in temperature and acidity ([]) within the top 300 m using a relative threshold of 95% and then identify CCX where conditions are extreme for both stressors for at least 50 m of the water column. When analyzed on a fixed baseline, positive trends in ocean warming and acidification caused CCX to last longer, intensify, and expand throughout the Southern Ocean. In the Antarctic zone, CCX expanded between 1980 and 2019 more than ten times in volume, lasted up to 120 days longer, and doubled in anomaly. Some of the largest and longest events occurred in Antarctic Marine Protected Areas (MPAs), covering more than 200,000 km2 and persisting for over 500 days. CCX in the Subantarctic and Northern zones quadrupled in volume and increased by more than 30% in anomaly. Across the Southern Ocean, the increasing occurrence of CCX exacerbates the risks to marine ecosystems from warming and acidification.

Plain Language Summary

Extreme heat events in the ocean, known as Marine HeatWaves (MHW), are becoming more common due to climate change. These events can be even more harmful when they occur at the same time as Ocean Acidity eXtreme (OAX) events, synergistically causing stress for marine life. In this study, we looked at how often these combined events in the upper ocean, called Column-Compound eXtremes (CCX), occurred in the Southern Ocean between 1980 and 2019. We used a numerical model simulation to investigate changes in CCX during the study period. Compared to conditions in 1980, we find that CCX in the Antarctic zone have expanded more than 10 times in volume and lasted up to 120 days longer. In addition, expansive and intense CCX are found in Antarctic Marine Protected Areas (MPAs), posing a threat to vulnerable ecosystems. These events covered more than 200,000 km2 and lasted more than 500 days. The increasing occurrence of CCX across the Southern Ocean exacerbates the risks to marine ecosystems arising from ocean warming and acidification.

Key Points

  • In the Antarctic zone, Column-Compound eXtremes (CCX) occupied in 2019 relative to 1980 ten times more volume and doubled in anomaly
  • Marine Protected Areas in the Ross Sea and Antarctic Peninsula are disproportionately affected by the largest, longest, most intense CCX
  • More than 70% of surface marine heatwaves contain CCX in 2019, although up to 60% of CCX occur without any surface expression
Continue reading ‘Compound marine heatwaves and acidity extremes in the Southern Ocean’

A century of change in the California Current: upwelling system amplifies acidification

Published 19 November 2025 Science Closed
Tags: , , , , , , ,

Predicting the pace of acidification in the California Current System (CCS), a productive upwelling system that borders the west coast of North America, is complex because the anthropogenic contribution is intertwined with other natural sources. A central question is whether acidification in the CCS will follow the pace of increasing atmospheric CO2, or if climate effects and other biogeochemical processes will either amplify or attenuate acidification. Here, we apply the boron isotope pH proxy to cold-water orange cup corals to establish a historic level of acidification in the CCS and the Salish Sea, an associated marginal sea. Through a combination of complementary modeling and geochemical approaches, we show that the CCS and Salish Sea have experienced amplified acidification over the industrial era, driven by the interaction between anthropogenic CO2 and a thermodynamic buffering effect. From this foundation, we project future acidification in the CCS under elevated CO2 emissions. The projected change in pCO2 over the 21st century will continue to outpace atmospheric CO2, posing challenges to marine ecosystems of biological, cultural, and economic importance.

Continue reading ‘A century of change in the California Current: upwelling system amplifies acidification’

Climate refugia could disappear from Australia’s marine protected areas by 2040

Published 3 November 2025 Science Closed
Tags: , , , ,

Abstract

Climate change manifests in the ocean as chronic stressors, including warming, acidification and deoxygenation, and as acute stressors such as marine heatwaves. While marine protected areas (MPAs) are often designed to mitigate local stressors such as fishing and mining, their design seldom considers climate change. Using the Australian marine estate as a case study, we use projections from 11 CMIP6 Earth System Models to assess the climate exposure of Australian waters, and implications for the MPA network. We find that, under scenarios that exceed 1.8°C of global surface warming this century, ocean climate is projected to surpass recent variability (1995–2014) from mid-century. This results in the disappearance of climate analogs—where future ocean conditions remain within recent variability—and of climate refugia—regions with slowest rates of environmental change, most likely to retain biodiversity—by 2040. Australian MPAs and unprotected areas exhibit similar patterns of exposure to warming, acidification, deoxygenation, and marine heatwaves, suggesting that MPA placement with respect to future climate is no better than random. Despite potential re-emergence of climate refugia after 2060 under lower-emissions scenarios, continued emissions under current Nationally Determined Contributions (SSP2–4.5) risk ecosystem collapse from chronic and acute thermal stress across protected and unprotected waters. While cutting emissions can partially cap or delay climate impacts, even under lower-emissions scenarios, effective conservation requires adaptive strategies that protect biodiversity in place and on the move.

Plain Language Summary

Marine protected areas (MPAs) are designed to safeguard ocean biodiversity from threats like fishing, but their design rarely considers climate change impacts. We assessed the future exposure of Australia’s MPAs to climate change using projections of ocean climate. Our findings reveal that if global surface warming exceeds 1.8°C this century, Australian marine ecosystems will face entirely novel ocean conditions beyond recent historical variability (1995–2014) by mid-century. This results in the Australia-wide disappearance of regions with slowest rates of climate change—climate refugia—representing a substantial threat to marine biodiversity. Our results suggest that MPAs are no better off than unprotected areas, facing the same risks from warming, acidification, deoxygenation, and marine heatwaves as unprotected waters. We found that reducing emissions could facilitate the reappearance of some climate refugia after 2060, but continuing along current emissions trends risks ecosystem collapse from warming throughout Australia’s protected and unprotected waters. Effective marine conservation requires both emissions reductions and adaptive strategies to protect biodiversity as species respond to a changing ocean climate.

Key Points

  • Ocean climate in Australia will reach a climate horizon by mid-century, representing novel conditions beyond recent variability (1995–2014)
  • Under global warming scenarios exceeding 1.8°C this century, climate refugia are projected to disappear from Australian waters by 2040
  • Existing MPAs and unprotected areas exhibit equivalent patterns of exposure to multiple ocean climate metrics, suggesting a lack of climate-smart design
Continue reading ‘Climate refugia could disappear from Australia’s marine protected areas by 2040′

Unprecedented carbon accumulation in the Indian Ocean during 2016–2017

Published 29 October 2025 Science Closed
Tags: , , ,

Abstract

During 2016–2017, the Indian Ocean experienced a pronounced increase in dissolved inorganic carbon (∼0.39 PgC/yr), approximately four times greater than the annual mean air–sea CO2 flux. Using a reconstructed data product and a state-of-the-art ocean biogeochemical model, we attribute this anomaly to an enhanced Southern Ocean inflow and a weakened Indonesian Throughflow associated with an El Niño event accompanied by a positive Indian Ocean Dipole (IOD), and followed by a negative IOD during the El Niño-to-La Niña transition. The resulting carbon accumulation leads to a decline in aragonite saturation and a shoaling of the aragonite saturation horizon in the southeastern Indian Ocean. This subsurface acidification may pose risks to deep-water calcifying organisms. Our findings demonstrate that ocean carbon storage and acidification are strongly modulated by circulation-driven transport processes, highlighting the need for improved subsurface observations and model capabilities to better capture the interior carbon response to climate variability.

Plain Language Summary

Between 2016 and 2017, the Indian Ocean stored a much larger amount of carbon than usual—about four times more than the typical annual exchange of carbon between the ocean and atmosphere. Using reconstructed observations and an advanced ocean model, we show that this unusual carbon buildup was caused by stronger inflow from the Southern Ocean and a weaker Indonesian Throughflow, driven by El Niño and negative Indian Ocean Dipole events. This extra carbon made the water more acidic and caused the depth at which aragonite (a mineral important for shell-building organisms) remains stable to rise by nearly 20 m in the southeastern Indian Ocean. These chemical changes could threaten deep-water organisms that rely on stable chemical conditions. Our results highlight how ocean currents can strongly affect carbon storage and acidification, and point to the need for better subsurface measurements and models to understand how climate variability impacts the ocean interior.

Key Points

  • Indian Ocean carbon storage varied unprecedentedly in 2016–2017, driven by circulation anomalies linked to climate variability
  • Anomalous dissolved inorganic carbon inventory was mainly due to increased Southern Ocean inflow and weakened Indonesian Throughflow
  • Anomalous carbon redistribution caused subsurface acidification, shoaling aragonite saturation depth by ∼20 m in the southeast Indian Ocean
Continue reading ‘Unprecedented carbon accumulation in the Indian Ocean during 2016–2017’

Subscribe

Search

  • Reset

OA-ICC Highlights

Resources