Search

Abstract: Recent advances in explainable artificial intelligence (XAI) methods show promise for understanding predictions made by machine learning (ML) models. XAI explains how the input features are relevant or important for the model predictions. We train linear regression (LR) and convolutional neural network (CNN) models to make one-day predictions of sea-ice velocity in the Arctic from inputs of present-day wind velocity and previous-day ice velocity and concentration. We apply XAI methods to the CNN and compare explanations to variance explained by LR. We confirm the feasibility of using a novel XAI method (i.e. global layerwise relevance propagation (LRP)) to understand ML model predictions of sea-ice motion by comparing to established techniques. We investigate a suite of linear, perturbation-based, and propagation-based XAI methods in both local and global forms. Outputs from different explainability methods are generally consistent in showing that wind speed is the input feature with the highest contribution to ML predictions of ice motion, and we discuss inconsistencies in spatial variability of the explanations. Additionally, we show that CNN relies on both linear and non-linear relationships between the inputs and uses non-local information to make predictions. LRP shows that wind speed over land is highly relevant for predicting ice motion offshore. This provides a framework to show how knowledge of environmental variables (i.e. wind) on land could be useful for predicting other properties (i.e. sea-ice velocity) elsewhere.

The Southern Ocean modulates global biogeochemical (BGC) cycles substantially, affecting biological production and the global air-sea balance of carbon dioxide and interior dissolved oxygen content. Concurrently, the Southern Ocean is rich in highly dynamic mesoscale eddies. These eddies have the potential to alter local carbon, nutrient, and oxygen distributions through eddy pumping, stirring, and trapping. Additionally, the strong westerly winds could result in significant eddy-induced Ekman pumping counteracting the eddy pumping effects. However, the impact of mesoscale eddies on upper-ocean Southern Ocean biogeochemistry has not been quantified observationally at a regional scale.We now have nearly a decade of BGC observations from Argo floats deployed as part of the Southern Ocean Carbon and Climate Observations and Modeling project (SOCCOM). In addition, the Mesoscale Eddy Trajectory Atlas, version 3.2, delayed time (Meta3.2DT) database provides us with a robust assessment of eddies as detected by satellite altimeter measurements. Together, the two datasets allow us to investigate the three-dimensional structure of the biogeochemistry in Southern Ocean eddies. Here, we co-locate Southern Ocean eddies with BGC Argo floats to characterize composite vertical and horizontal structures of dissolved inorganic carbon (DIC), oxygen, and nitrate inside anticyclonic and cyclonic eddies compared to the mean climatological fields. We conduct this analysis in several subregions with different dominant processes. We find positive DIC and nitrate anomalies in cyclonic eddies, which we attribute to upward eddy pumping. We also find positive oxygen anomalies near the surface, which we attribute to upwelled nutrients that enhance biological production, leading to enhanced photosynthesis. At depth, we find negative oxygen anomalies in cyclonic eddies, which may be driven both by enhanced respiration due to increased biological production as well as the heaving of isopycnals via eddy pumping. The opposite is true for anticyclonic eddies due to downward eddy pumping (negative DIC and nitrate anomalies; negative oxygen anomalies near the surface and positive oxygen anomalies at depth). The magnitudes of the eddy imprints on biogeochemistry vary by region, indicating that stratification and other background signals influence the magnitude of the effect of eddies in a region. Our findings can help us to interpret the influence of mesoscale eddies on the Southern Ocean carbon fluxes and biogeochemistry, including assessing the relative dominance of eddy pumping and eddy-induced Ekman pumping in different subregions of the Southern Ocean.

Abstract: The Surface Water Ocean Topography (SWOT) satellite mission provides high-resolution two-dimensional sea surface height (SSH) data with swath coverage. However, spatially correlated errors affect these SSH measurements, particularly in the cross-track direction. The scales of errors can be similar to the scales of ocean features. Conventionally, instrumental errors and ocean signals have been solved for independently in two stages. Here, we have developed a one-stage procedure that solves for the correlated error at the same time that data are assimilated into a dynamical ocean model. This uses the ocean dynamics to distinguish ocean signals from observation errors. We test its performance relative to the two-stage method using simplified dynamics and a data set consisting of westward propagating Rossby waves, along with correlated instrumental errors of varying magnitudes. In a series of tests, we found that the one-stage approach consistently outperforms the two-stage approach when estimating SSH signal and correlated errors. The one-stage approach can recover over 95% of the SSH signal, while skill for the two-stage approach drops significantly as error increases. Our findings suggest that solving for the correlated errors within the assimilation framework can provide an effective analysis approach, reducing the risks of confounding signal and instrument noise.

Abstract: The Southern Ocean is a region of intense air–sea exchange that plays a critical role for ocean circulation, global carbon cycling, and climate. Subsurface chlorophyll‐a maxima, annually recurrent features throughout the Southern Ocean, may increase the energy flux to higher trophic levels and facilitate downward carbon export. It is important that model parameterizations appropriately represent the chlorophyll vertical structure in the Southern Ocean. Using BGC‐Argo chlorophyll profiles and the Biogeochemical Southern Ocean State Estimate (B‐SOSE), we investigate the sensitivity of chlorophyll vertical structure to model parameters. Based on the sensitivity analysis results, we estimate optimized parameters, which efficiently improve the model consistency with observations. We characterize chlorophyll vertical structure in terms of Empirical Orthogonal Functions and define metrics to compare model results and observations in a series of parameter perturbation experiments. We show that chlorophyll magnitudes are likely to respond quasi‐symmetrically to perturbations in the analyzed parameters, while depth and thickness of the subsurface chlorophyll maximum show an asymmetric response. Perturbing the phytoplankton growth tends to generate more symmetric responses than perturbations in the grazing rate. We identify parameters that affect chlorophyll magnitude, subsurface chlorophyll or both and discuss insights into the processes that determine chlorophyll vertical structure in B‐SOSE. We highlight turbulence, differences in phytoplankton traits, and grazing parameterizations as key areas for improvement in models of the Southern Ocean.

The Southern Ocean is rich in highly dynamic mesoscale eddies and substantially modulates global biogeochemical cycles. However, the overall surface and subsurface effects of eddies on the Southern Ocean biogeochemistry have not been quantified observationally at a large scale. Here, we co-locate eddies, identified in the Meta3.2DT satellite altimeter-based product, with biogeochemical Argo floats to determine the effects of eddies on the dissolved inorganic carbon (DIC), nitrate, and dissolved oxygen concentrations in the upper 1,500 m of the ice-free Southern Ocean, as well as the eddy effects on the carbon fluxes in this region. DIC and nitrate concentrations are lower in anticyclonic eddies (AEs) and increased in cyclonic eddies (CEs), while dissolved oxygen anomalies switch signs above (CEs: positive, AEs: negative) and below the mixed layer (CEs: negative, AEs: positive). We attribute these anomalies primarily to eddy pumping (isopycnal heave), as well as eddy trapping for oxygen. Maximum anomalies in all tracers occur at greater depths in the subduction zone north of the Antarctic Circumpolar Current (ACC) compared to the upwelling region in the ACC, reflecting differences in background vertical structures. Eddy effects on air-sea CO2 exchange have significant seasonal variability, with additional outgassing in CEs in fall (physical process) and additional oceanic uptake in AEs and CEs in spring (biological and physical process). Integrated over the Southern Ocean, AEs contribute ∼0.03± 0.01 Pg C yr-1 (7 ±2% ) to the Southern Ocean carbon uptake, and CEs offset this by ∼0.01± 0.01 Pg C yr-1 (2 ±2% ). These findings underscore the importance of considering eddy impacts in observing networks and climate models.

Abstract: Over 9 years of hourly surface current data from high‐frequency radar (HFR) off the US West Coast are analyzed using a Bayesian least‐squares fit for tidal components. The spatial resolution and geographic extent of HFR data allow us to assess the spatial structure of the non‐phase‐locked component of the tide. In the frequency domain, the record length and sampling rate allow resolution of discrete tidal lines corresponding to well‐known constituents and the near‐tidal broadband elevated continuum resulting from amplitude and phase modulation of the tides, known as cusps. The FES2014 tide model is used to remove the barotropic component of tidal surface currents in order to evaluate its contribution to the phase‐locked variance and spatial structure. The mean time scale of modulation is 243 days for the M2 constituent and 181 days for S2, with overlap in their range of values. These constituents' modulated amplitudes are significantly correlated in several regions, suggesting shared forcing mechanisms. Within the frequency band M2 ± 5 cycles per year, an average of 48% of energy is not at the phase‐locked frequency. When we remove the barotropic model, this increases to 64%. In both cases there is substantial regional variability. This indicates that a large fraction of tidal energy is not easily predicted (e.g., for satellite altimeter applications). The spatial autocorrelation of the non‐phase‐locked variance fraction drops to zero over a distance of 150 km, a scale that is comparable to the swath width of the Surface Water and Ocean Topography altimeter.

Abstract: Meltwater from the Greenland Ice Sheet can alter the continental shelf/slope circulation and cross-shelf freshwater fluxes and limit deep convection in adjacent basins through surface freshening. We explore the impacts on the West Greenland Current and eastern Labrador Sea with different vertical distributions of the meltwater forcing. In this study, we present the results from global coupled ocean/sea ice simulations, forced with atmospheric reanalysis, that are mesoscale eddy-active (∼2–3-km horizontal spacing) and eddy-permitting (∼6–7-km horizontal spacing) in the study region. We compare the West Greenland Current in mesoscale eddy-active and eddy-permitting without meltwater to highlight the role of small-scale features. The mesoscale eddy-active configuration is then used to assess the change in the eastern Labrador Sea when meltwater is added to the surface or vertically distributed to account for mixing within fjords. In both simulations with meltwater, the West Greenland and West Greenland Coastal Currents are faster than in the simulation with no meltwater; their mean surface speeds are the highest in the vertical distribution case. In the latter case, there is enhanced baroclinic conversion at the shelf break compared to the simulation with no meltwater. When meltwater is vertically distributed, there is an increase in baroclinic conversion at the shelf break associated with increased eddy kinetic energy. In addition, in the eastern Labrador Sea, the salinity is lower and the meltwater volume is greater when meltwater is vertically distributed. Therefore, the West Greenland Current is sensitive to how meltwater is added to the ocean with implications for the freshening of the Labrador Sea. Significance Statement: Our goal is to understand how the flux of freshwater across the West Greenland continental slope into the Labrador Sea is modified by meltwater from the Greenland Ice Sheet. We compare the simulations of the ocean that capture key dynamics along the West Greenland continental slope that have no meltwater, meltwater added to the ocean surface, and meltwater distributed vertically to represent the mixing within fjords. When meltwater is added, the currents along the continental slope are faster, with the greatest increase when meltwater is vertically distributed. In that case, there is enhanced freshening of the Labrador Sea because modified density gradients generate more eddies. Proper representation of the vertical structure of meltwater is important for projecting the impact of freshwater on the subpolar North Atlantic.

Open-ocean polynyas formed over the Maud Rise, in the Weddell Sea, during the winters of 2016-2017. Such polynyas are rare events in the Southern Ocean and are associated with deep convection, affecting regional carbon and heat budgets. Using an ocean state estimate, we found that during 2017, early sea ice melting occurred in response to enhanced vertical mixing of heat, which was accompanied by mixing of salt. The melting sea ice compensated for the vertically mixed salt, resulting in a net buoyancy gain. An additional salt input was then necessary to destabilize the upper ocean. This came from a hitherto unexplored polynya-formation mechanism: an Ekman transport of salt across a jet girdling the northern flank of the Maud Rise. Such transport was driven by intensified eastward surface stresses during 2015-2018. Our results illustrate how highly localized interactions between wind, ocean flow and topography can trigger polynya formation in the open Southern Ocean.

Modelled water-mass changes in the North Pacific thermocline from CMIP6, both in the subsurface and at the surface, reveal the impact of the competition between anthropogenic aerosols and greenhouse gases (GHGs) over the past 6 decades. The aerosol effect overwhelms the GHG effect during 1950-1985 in driving salinity changes on density surfaces, while after 1985 the GHG effect dominates. These subsurface water-mass changes are traced back to changes at the surface, of which ~70% stems from the migration of density surface outcrops, equatorward due to regional cooling by anthropogenic aerosols and subsequent poleward due to warming by GHGs. Ocean subduction connects these surface outcrop changes to the main thermocline. Both observations and models reveal this transition in climate forcing around 1985 and highlight the important role of anthropogenic aerosol climate forcing on our oceans’ water masses.

Abstract Background Testicular torsion (TT) is the abnormal twisting of the spermatic cord leading to strangulation of the testis. Any non-traumatic acute scrotal pain in adolescents and young adults is considered TT until proven otherwise. It is a urological emergency. The diagnosis is clinically based on symptoms and signs that raise a high index of suspicions. Scrotal exploration is necessary, which will provide diagnostic certainty and allow treatment. Methods This study aimed to assess the clinical predictors of TT. This was a retrospective cross-sectional study of patients aged 15 years and older who underwent surgical exploration for suspected TT from January 2005 to December 2021 at Teaching Hospital Yalgado Ouedraogo (Burkina Faso). We compared the demographic data and clinical findings of patients according to the intraoperative diagnosis. Results In total, 135 scrotal explorations were performed for suspected TT. The average age of the patients was 24.56 years. TT was found in 123 cases (91.11%), epididymo-orchitis in 8 cases (5.93%) and in 4 cases (2.96%) no abnormality was found. Scrotal swelling and elevation of the testicle were more common in TT than in other cases (p

BTEX (benzene, toluene, ethylbenzene, and m-xylene,p-xylene, and o-xylene) are significant urban air pollutants. This study examines BTEX variability across 7 European countries using data from 22 monitoring sites in different urban settings (urban background, traffic, industry, and suburban background). Results indicate that the relative abundance of BTEX in urban areas follows the order toluene > benzene > m,p-xylene > o-xylene > ethylbenzene, with median mixing ratios of 266 ± 152, 163 ± 74, 129 ± 88, 53 ± 35, and 45 ± 27 ppt during the years 2017–2022, respectively. Seasonal trends show benzene had similar median concentrations across urban background, traffic, and industrial sites, indicating mixed sources. Toluene levels were highest in traffic and industrial areas, highlighting road traffic and industrial emissions. Ethylbenzene and xylenes showed equivalent levels in traffic and industrial areas but were lower in urban backgrounds. Peak BTEX levels occurred during morning and evening rush hours, linked to traffic, heating, and atmospheric stagnation. B/T ratios ranged from 0.29 ± 0.11 to 1.35 ± 0.95, and X/E ratios ranged from 1.75 ± 0.91 to 3.68 ± 0.30, indicating primary pollution from local traffic, followed by solvents, coatings, and biomass burning. Lifetime cancer risk from BTEX exposure was below the definite risk threshold (10−4) but above the permissible risk level (10−6), suggesting moderate risk from benzene and ethylbenzene, particularly in traffic and industrial areas. Additionally, the health index of BTEX at monitoring sites was generally lower than the threshold limit value, suggesting a low non-carcinogenic risk overall. This study offers essential insights into BTEX pollution in urban European environments.

Freshwater fishes are increasingly facing extinction. Some species will require conservation intervention such as habitat restoration and/or population supplementation through mass-release of hatchery fish. In California, USA, a number of conservation strategies are underway to increase abundance of the endangered Delta Smelt (Hypomesus transpacificus); however, it is unclear how different estuarine conditions influence hatchery fish. The goal of this study was to evaluate a year of Delta Smelt field deployments to inform species conservation strategies of suitable conditions for smelt physiology. Hatchery-reared Delta Smelt was deployed in experimental cages (seven deployments) throughout the Estuary in the winter, summer and fall of 2019. Effects of season and location of cage deployments on fish health (condition factor and histological condition of liver and gill), growth, thermal tolerance and survival were evaluated. The results indicate both seasonal and location differences, with high survival in the winter (100%) and fall (88-92%) compared to lower survival in summer (67%). In the summer, one of the study sites had no surviving fish following high temperature exposure, which peaked ~26°C. After 29 days in the cages, surviving Delta Smelt in summer and fall showed signs of nutritional stress that may be related to biofouling of the cages limiting passive food inputs, restriction of natural foraging behaviour by containment in the cages, and water temperatures that were too high given the chronically low pelagic productivity in the Estuary overall. Field measurements of upper thermal tolerance (CTmax) following caging exposures suggest that laboratory measures of CTmax may overestimate the realized tolerance in a more stochastic field environment. This study demonstrates the utility of using cages as an experimental tool to better understand aspects of Delta Smelt physiological responses to environmental changes across estuarine habitats in a more natural-field setting, while also highlighting potential limitations of using cages.

This paper presents a first comprehensive analysis of long-term measurements of atmospheric aerosol components from aerosol chemical speciation monitor (ACSM) and multiwavelength Aethalometer (AE33) instruments collected between 2015 and 2021 at 13 (sub)urban sites as part of the French CARA (Chemical Characterization of Particles) program. The datasets contain the mass concentrations of major chemical species within submicron aerosols (PM1), namely organic aerosols (OAs), nitrate (NO3-), ammonium (NH4+), sulfate (SO42-), non-sea-salt chloride (Cl−), and equivalent black carbon (eBC). Rigorous quality control, technical validation, and environmental evaluation processes were applied, adhering to both guidance from the French Reference Laboratory for Air Quality Monitoring (LCSQA) and the Aerosol, Clouds, and Trace Gases Research Infrastructure (ACTRIS) standard operating procedures. Key findings include geographical differences in the aerosol chemical composition, seasonal variations, and diel patterns, which are influenced by meteorological conditions, anthropogenic activities, and proximity to emission sources. Overall, OA dominates PM1 at each site (43 %–60 % of total mass), showing distinct seasonality with higher concentrations (i) in winter, due to enhanced residential heating emissions, and (ii) in summer, due to increased photochemistry favoring secondary aerosol formation. NO3 is the second most important contributor to PM1 (15 %–30 %), peaking in late winter and early spring, especially in northern France, and playing a significant role during pollution episodes. SO4 (8 %–14 %) and eBC (5 %–11 %) complement the major fine-aerosol species, with their relative contributions strongly influenced by the origin of air masses and the stability of meteorological conditions, respectively. A comparison with the 3D chemical transport model (CTM) CHIMERE shows high correlations between simulations and measurements, albeit with an OA concentration underestimation of 46 %–76 %. Regional discrepancies in NO3 concentration levels emphasize the importance of these datasets with respect to validating air quality models and tailoring air pollution mitigation strategies. The datasets can be found at https://doi.org/10.5281/zenodo.13318298 (Chebaicheb et al., 2024).

Abstract: Submesoscale currents and internal gravity waves achieve an intense turbulent cascade near the ocean surface (0 m – O(100) m depth), which is thought to give rise to significant energy sources and sinks for mesoscale eddies. Here, we characterise the contributions of Non-Wave Currents (NWCs; including eddies and fronts) and Internal Gravity Waves (IGWs; including near-inertial motions, lee waves and the internal wave continuum) to near-surface submesoscale turbulence in the Drake Passage. Using a numerical simulation, we combine Lagrangian filtering and a Helmholtz decomposition to identify NWCs and IGWs and to characterise their dynamics (rotational vs. divergent). We show that NWCs and IGWs contribute in different proportions to the inverse and forward turbulent kinetic energy cascades, based on their dynamics and spatiotemporal scales. Purely rotational NWCs cause most of the inverse cascade, while coupled rotational– divergent components of NWCs and coupled NWC–IGWs cause the forward cascade. The cascade changes direction at a spatial scale at which motions become increasingly divergent. However, the forward cascade is ultimately limited by the motions’ spatiotemporal scales. The bulk of the forward cascade (80 – 95%) is caused by NWCs and IGWs of small spatiotemporal scales (L

Modeled water-mass changes in the North Pacific thermocline, both in the subsurface and at the surface, reveal the impact of the competition between anthropogenic aerosols (AAs) and greenhouse gases (GHGs) over the past 6 decades. The AA effect overwhelms the GHG effect during 1950-1985 in driving salinity changes on density surfaces, while after 1985 the GHG effect dominates. These subsurface water-mass changes are traced back to changes at the surface, of which ~70% stems from the migration of density surface outcrops, equatorward due to regional cooling by AAs and subsequent poleward due to warming by GHGs. Ocean subduction connects these surface outcrop changes to the main thermocline. Both observations and models reveal this transition in climate forcing around 1985 and highlight the important role of AA climate forcing on our oceans' water masses.

Recent developments in national health data platforms have the potential to significantly advance medical research, improve public health outcomes, and foster public trust in data governance. Across Europe, initiatives such as the NHS Research Secure Data Environment in England and the Data Room for Health-Related Research in Switzerland are underway, reflecting examples analogous to the European Health Data Space in two non-EU nations. Policy discussions in England and Switzerland emphasize building public trust to foster participation and ensure the success of these platforms. Central to building public trust is investing efforts into developing and implementing public involvement activities. In this commentary, we refer to three national research programs, namely the UK Biobank, Genomics England, and the Swiss Health Study, which implemented effective public involvement activities and achieved high participation rates. The public involvement activities used within these programs are presented following on established guiding principles for fostering public trust in health data research. Under this lens, we provide actionable policy recommendations to inform the development of trust-building public involvement activities for national health data platforms.

BackgroundThe increased use of digital data in health research demands interdisciplinary collaborations to address its methodological complexities and challenges. This often entails merging the linear deductive approach of health research with the explorative iterative approach of data science. However, there is a lack of structured teaching courses and guidance on how to effectively and constructively bridge different disciplines and research approaches. ObjectiveThis study aimed to provide a set of tools and recommendations designed to facilitate interdisciplinary education and collaboration. Target groups are lecturers who can use these tools to design interdisciplinary courses, supervisors who guide PhD and master’s students in their interdisciplinary projects, and principal investigators who design and organize workshops to initiate and guide interdisciplinary projects. MethodsOur study was conducted in 3 steps: (1) developing a common terminology, (2) identifying established workflows for research question formulation, and (3) examining adaptations of existing study workflows combining methods from health research and data science. We also formulated recommendations for a pragmatic implementation of our findings. We conducted a literature search and organized 3 interdisciplinary expert workshops with researchers at the University of Zurich. For the workshops and the subsequent manuscript writing process, we adopted a consensus study methodology. ResultsWe developed a set of tools to facilitate interdisciplinary education and collaboration. These tools focused on 2 key dimensions— content and curriculum and methods and teaching style—and can be applied in various educational and research settings. We developed a glossary to establish a shared understanding of common terminologies and concepts. We delineated the established study workflow for research question formulation, emphasizing the “what” and the “how,” while summarizing the necessary tools to facilitate the process. We propose 3 clusters of contextual and methodological adaptations to this workflow to better integrate data science practices: (1) acknowledging real-life constraints and limitations in research scope; (2) allowing more iterative, data-driven approaches to research question formulation; and (3) strengthening research quality through reproducibility principles and adherence to the findable, accessible, interoperable, and reusable (FAIR) data principles. ConclusionsResearch question formulation remains a relevant and useful research step in projects using digital data. We recommend initiating new interdisciplinary collaborations by establishing terminologies as well as using the concepts of research tasks to foster a shared understanding. Our tools and recommendations can support academic educators in training health professionals and researchers for interdisciplinary digital health projects.

User trust is pivotal for the adoption of digital health systems interventions (DHI). In response, numerous trust-building guidelines have recently emerged targeting DHIs such as artificial intelligence. The common aim of these guidelines aimed at private sector actors and government policy makers is to build trustworthy DHI. While these guidelines provide some indication of what trustworthiness is, the guidelines typically only define trust and trustworthiness in broad terms, they rarely offer guidance about economic considerations that would allow implementers to measure and balance trade-offs between costs and benefits. These considerations are important when deciding how best to allocate scarce resources (eg, financial capital, workforce, or time). The missing focus on economics undermines the potential usefulness of such guidelines. We propose the development of actionable trust-performance-indicators (including but not limited to surveys) to gather evidence on the cost-effectiveness of trust-building principles as a crucial step for successful implementation. Furthermore, we offer guidance on navigating the conceptual complexity surrounding trust and on how to sharpen the trust discourse. Successful implementation of economic considerations is critical to successfully build user trust in DHI.

Abstract The Southern Ocean is rich in highly dynamic mesoscale eddies and substantially modulates global biogeochemical cycles. However, the overall surface and subsurface effects of eddies on the Southern Ocean biogeochemistry have not been quantified observationally at a large scale. Here, we co‐locate eddies, identified in the Meta3.2DT satellite altimeter‐based product, with biogeochemical Argo floats to determine the effects of eddies on the dissolved inorganic carbon (DIC), nitrate, and dissolved oxygen concentrations in the upper 1,500 m of the ice‐free Southern Ocean, as well as the eddy effects on the carbon fluxes in this region. DIC and nitrate concentrations are lower in anticyclonic eddies (AEs) and increased in cyclonic eddies (CEs), while dissolved oxygen anomalies switch signs above (CEs: positive, AEs: negative) and below the mixed layer (CEs: negative, AEs: positive). We attribute these anomalies primarily to eddy pumping (isopycnal heave), as well as eddy trapping for oxygen. Maximum anomalies in all tracers occur at greater depths in the subduction zone north of the Antarctic Circumpolar Current (ACC) compared to the upwelling region in the ACC, reflecting differences in background vertical structures. Eddy effects on air–sea CO2 exchange have significant seasonal variability, with additional outgassing in CEs in fall (physical process) and additional oceanic uptake in AEs and CEs in spring (biological and physical process). Integrated over the Southern Ocean, AEs contribute ∼0.03± 0.01 Pg C yr−1 (7 ±2%) to the Southern Ocean carbon uptake, and CEs offset this by ∼0.01±0.01 Pg C yr−1 (2 ±2%). These findings underscore the importance of considering eddy impacts in observing networks and climate models.

Abstract: Physics-based simulations of Arctic sea ice are highly complex, involving transport between different phases, length scales, and time scales. Resultantly, numerical simulations of sea-ice dynamics have a high computational cost and model uncertainty. We employ data-driven machine learning (ML) to make predictions of sea-ice motion. The ML models are built to predict present-day sea-ice velocity given present-day wind velocity and previous-day sea-ice concentration and velocity. Models are trained using reanalysis winds and satellite-derived sea-ice properties. We compare the predictions of three different models: persistence (PS), linear regression (LR), and convolutional neural network (CNN). We quantify the spatio-temporal variability of the correlation between observations and the statistical model predictions. Additionally, we analyze model performance in comparison to variability in properties related to ice motion (wind velocity, ice velocity, ice concentration, distance from coast, bathymetric depth) to understand the processes related to decreases in model performance. Results indicate that a CNN makes skillful predictions of daily sea-ice velocity with a correlation up to 0.81 between predicted and observed sea-ice velocity, while the LR and PS implementations exhibit correlations of 0.78 and 0.69, respectively. The correlation varies spatially and seasonally; lower values occur in shallow coastal regions and during times of minimum sea-ice extent. LR parameter analysis indicates that wind velocity plays the largest role in predicting sea-ice velocity on one-day time scales, particularly in the central Arctic. Regions where wind velocity has the largest LR parameter are regions where the CNN has higher predictive skill than the LR.