Gulf Stream Region Research Articles

The Gulf Stream Structure and Meandering Based on the CTD and SADCP Measurements in 1989–1990 and 2014–2015

An review of field studies in the Gulf Stream region carried out by the authors in two periods with a break of 25 years is presented to summarize the results. The studies in the early period included hydrographic surveys in the area of a southern meander of the Gulf Stream (1989) and in the area of dividing of a single jet of the current into separate branches: in the Gulf Stream delta (1990). The second, recent, stage includes on-route surveys with SADCP profiler in 2014–2015 while crossing the meandering Gulf Stream at mid-latitudes to study its detailed high-resolution velocity field structure.

Near‐Surface Wind Convergence Along the Sea Ice Edge in the Greenland Sea: Its Mean State and Shaping Process

AbstractAt mid‐latitudes, a narrow band of near‐surface wind convergence (NSWC) overlies the western boundary currents in long‐term climatology as a response to steep sea surface temperature gradients. The underlying dynamics shaping mean convergence in the mid‐latitude region have been investigated in detail. In polar regions, surface temperature gradients are intense along the sea ice edges. However, literature concerning NSWC near sea ice edges is limited. This study investigates time‐mean NSWC along sea ice edges and its shaping processes, focusing on the Greenland Sea, based on atmospheric reanalysis. In cold‐season climatology, positive NSWC overlies the sea ice edge, resulting in a localized upward motion reaching the free atmosphere. The mean NSWC was insensitive to sea ice thickness and surface roughness in the regional model. This study suggests that, in addition to local atmospheric boundary processes, extreme NSWC events play a vital role in shaping the mean distribution. Although these features are similar to those along the Gulf Stream, atmospheric fronts appear to play a relatively minor role in the Greenland Sea. Instead, the frequent cyclone generation near the sea ice edge and the anticyclonic circulation over Greenland in conjunction with the transient synoptic circulation seem essential. In the warm season, positive NSWC was virtually missing in the Greenland Sea, unlike in the Gulf Stream region, reflecting the shallow virtual temperature response to the surface thermal forcing. This study contributes to understanding the mechanisms by which sea ice variability affects large‐scale atmospheric circulation in remote regions.

Internal Tide Energy Transfers Induced by Mesoscale Circulation and Topography Across the North Atlantic

AbstractThe interactions between the internal tide and the mesoscale circulation are studied from the internal tide energy budget perspective. To that end, the modal energy budget of the internal tide is diagnosed using a high resolution numerical simulation covering the North Atlantic. Compared to the topographic contribution, the advection of the internal tide by the low‐frequency flow component and the horizontal and vertical shear are found to be significant at global scale, while the buoyancy contribution is important locally. The advection of the internal tide by the low‐frequency currents is responsible for a net energy transfer from the large scale to smaller scale internal tide, without exchanges with the low‐frequency flow. On the opposite, the shear of the mesoscale circulation and the buoyancy field are responsible for exchanges between the internal tide and the low‐frequency flow. The importance of the shear increases in the northernmost part of the domain, and a partial compensation between the buoyancy and the shear contributions is found in some areas of the North Atlantic, such as in the Gulf Stream region. In addition, the temporal variability of these energy transfers is investigated. In contrast to topographic scattering, for which the spring‐neap cycle is the dominant frequency, the energy transfer terms driven by low‐frequency motions in areas of strong mesoscale activity are also modulated by variations of the low‐frequency current spatial distribution.

Gulf Stream Moisture Fluxes Impact Atmospheric Blocks Throughout the Northern Hemisphere

AbstractIn this study, we explore the impact of oceanic moisture fluxes on atmospheric blocks using the ECMWF IFS. Artificially suppressing surface latent heat flux over the Gulf Stream (GS) region reduces atmospheric blocking frequency across the Northern Hemisphere by up to 30%. Affected blocks show a shorter lifespan (−6%), smaller spatial extent (−10%), and reduced intensity (−0.4%), with an increased number of individual blocking anticyclones (+17%). These findings are robust across various blocking detection thresholds. Analysis reveals a qualitatively consistent response across all resolutions, with Tco639 (∼18 km) showing the largest statistically significant change across all blocking characteristics, although differences between resolutions are not statistically significant. Exploring the broader Rossby wave pattern, we observe that diminished moisture fluxes favor eastward propagation and higher zonal wavenumbers, while air‐sea interactions promote stationary and westward‐propagating waves with zonal wavenumber 3. This study underscores the critical role of the GS in modulating atmospheric blocking.

Training Neural Mapping Schemes for Satellite Altimetry With Simulation Data

AbstractSatellite altimetry combined with data assimilation and optimal interpolation schemes have deeply renewed our ability to monitor sea surface dynamics. Recently, deep learning schemes have emerged as appealing solutions to address space‐time interpolation problems. However, the training of state‐of‐the‐art neural schemes on real‐world case‐studies is hindered by the sparse space‐time coverage of the sea surface of real altimetry data set. Here, we introduce an innovative approach that leverages state‐of‐the‐art ocean models to train simulation‐based neural schemes for the mapping of sea surface height and demonstrate their performance on real altimetry data sets. We analyze further how the ocean simulation data set used during the training phase impacts this performance. This experimental analysis covers both the resolution from eddy‐present configurations to eddy‐rich ones, forced simulations versus reanalyzes using data assimilation and tide‐free versus tide‐resolving simulations. Our benchmarking framework focuses on a Gulf Stream region for a realistic 5‐altimeter constellation using NEMO ocean simulations and 4DVarNet mapping schemes. All simulation‐based 4DVarNets outperform the operational observation‐driven and reanalysis products, namely DUACS and GLORYS. The more realistic the ocean simulation data set used during the training phase, the better the mapping. The best 4DVarNet mapping was trained from an eddy‐rich and tide‐free simulation data sets. It improves the resolved longitudinal scale from 151 km for DUACS and 241 km for GLORYS to 98 km and reduces the root mean square error by 23% and 61%. These results open research avenues for new synergies between ocean modeling and ocean observation using learning‐based approaches.

Predicting Ocean Current Temperature Off the East Coast of America with XGBoost and Random Forest Algorithms Using Rstudio

This research investigates the comparative predictive efficacy of two leading machine learning methodologies, specifically the XGBoost and Random Forest models, in estimating ocean temperature dynamics in the TS Gulf Stream and Labrador Current regions along the east coast of North America. Using annual temperature datasets and relevant oceanographic parameters, the data is carefully processed, cleaned and sorted into training and test subsets via the RStudio Platform. The performance evaluation model is carried out using predetermined machine learning assessment criteria, including Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and R-squared. The results show the superiority of the XGBoost model compared to Random Forest in terms of prediction accuracy and minimizing prediction errors. The XGBoost model shows lower MSE values and higher R-squared values than the Random Forest model, indicating its better capacity in explaining data variations. XGBoost consistently provides more accurate predictions and shows higher sensitivity in identifying important factors influencing ocean temperature fluctuations than Random Forest. This research significantly improves understanding and prognostic capabilities regarding ocean temperature dynamics in the TS Gulf Stream and Labrador Current regions. Empirical evidence underlines the efficacy of the XGBoost model in predicting ocean temperatures in the studied region. Continuous model evaluation and parameter refinement for both methodologies is critical to establishing standards for optimal prediction performance. The findings of this research have implications for the fields of oceanography and climate science, and offer potential pathways to comprehensively understand and mitigate the impacts of climate change on marine ecosystems.

Impact of the Gulf Stream front on atmospheric rivers and Rossby wave train in the North Atlantic

The Gulf Stream (GS) ocean front exhibits intense ocean–atmosphere interaction in winter, which has a significant impact on the genesis and development of extratropical cyclones in the North Atlantic. The atmospheric rivers (ARs), closely related with the cyclones, transport substantial moisture from the North Atlantic towards the Western European coast. While the influence of the GS front on extratropical cyclones has been extensively studied, its effect on ARs remains unclear. In this study, two sets of ensemble experiments are conducted using a high-resolution global Community Atmosphere Model forced with or without the GS sea surface temperature front. Our findings reveal that the inclusion of the GS front leads to approximately 25% enhancement of water vapor transport and precipitation associated with ARs in the GS region, attributed to changes in both AR frequency and intensity. Furthermore, this leads to a more pronounced downstream response in Western Europe, characterized by up to 60% (40%) precipitation increases (reductions) around Spain (Norway) for the most extreme events (exceeding 90 mm/day). The influence of the GS front on ARs is mediated by both thermodynamic and dynamic factors. The thermodynamic aspect involves an overall increase of water vapor in both the GS region and Western Europe, promoting AR genesis. The dynamic aspect encompasses changes in storm tracks and Rossby wave train, contributing to downstream AR shift. Importantly, we find the co-occurrence of ARs and the GS front is crucial for inducing deep ascending motion and heating above the GS front, which perturbs the deep troposphere and triggers upper-level Rossby wave response. These findings provide a further understanding of the complex interaction between the oceanic front in the western boundary current regions and extratropical weather systems and the associated dynamics behind them.

Linking Gulf Stream air–sea interactions to the exceptional blocking episode in February 2019: a Lagrangian perspective

Abstract. The development of atmospheric blocks over the North Atlantic–European region can lead to extreme weather events like heat waves or cold air outbreaks. Despite their potential severe impact on surface weather, the correct prediction of blocking lifecycles remains a key challenge in current numerical weather prediction (NWP) models. Increasing evidence suggests that latent heat release in cyclones, the advection of cold air (cold air outbreaks, CAOs) from the Arctic over the North Atlantic, and associated air–sea interactions over the Gulf Stream are key processes contributing to the onset, maintenance, and persistence of such flow regimes. To better understand the mechanism connecting air–sea interactions over the Gulf Stream with changes in the large-scale flow, we focus on an episode between 20 and 27 February 2019, when a quasi-stationary upper-level ridge was established over western Europe accompanied by an intensified storm track in the northwestern North Atlantic. During that time, a record-breaking winter warm spell occurred over western Europe bringing temperatures above 20 ∘C to the United Kingdom, the Netherlands, and northern France. The event was preceded and accompanied by the development of several rapidly intensifying cyclones that originated in the Gulf Stream region and traversed the North Atlantic. To explore the mechanistic linkage between the formation of this block and air–sea interactions over the Gulf Stream, we adopt a Lagrangian perspective, using kinematic trajectories. This allows us to study the pathways and transformations of air masses that form the upper-level potential vorticity anomaly and interact with the ocean front. We establish that more than one-fifth of these air masses interact with the Gulf Stream in the lower troposphere, experiencing intense heating and moistening over the region due to the frequent occurrence of CAOs behind the cold front of the cyclones. Trajectories moistened by the advection of cold air over a warm ocean by one cyclone later ascend into the upper troposphere with the ascending airstream of a subsequent cyclone, fueled by the strong surface fluxes. These findings highlight the importance of CAOs in the Gulf Stream region, indicating that their intense coupling between the ocean and atmosphere plays a role in block development. Additionally, they provide a mechanistic pathway linking air–sea interactions in the lower troposphere and the upper-level flow.

The Gulf Stream Front Amplifies Large-Scale SST Feedback to the Atmosphere in North Atlantic Winter

The Gulf Stream (GS) ocean front releases intense moisture and heat to the atmosphere and regulates storm tracks and zonal jets in winter. The large-scale sea surface temperature (SST) anomaly in the central North Atlantic provides important feedback to the atmosphere in winter, but the role played in this feedback by the GS front inside the SST anomaly has not been extensively studied. In this study, two sets of ensemble experiments were conducted using a global community atmosphere model forced by SST in boreal winters from 2000 to 2013. The regional averaged SST and its variation in the experiments were identical, with the only difference being the strength of the SST front in the GS region. The large-scale SST anomaly in the central North Atlantic in our model provides feedback to the atmosphere and excites a wave train that extends across Eurasia. With the inclusion of the strong GS front, the first center of the wave train in the North Atlantic is strengthened by approximately 40%, and the wave activity flux toward downstream is highly intensified. When the large-scale SST anomaly is combined with a strong GS front, greatly increased water vapor is released from the GS region, resulting in a 50% increase in moisture transport toward Western Europe. In this scenario, precipitation and diabatic heating both increase greatly on the western Scandinavian Peninsula. With the release of deep diabatic heating, a strong upward wave activity flux is triggered, and the wave train excited by the large-scale SST variation is significantly intensified. These findings suggest that the strong SST front in the large-scale SST anomaly in the central North Atlantic significantly amplifies its feedback to the atmosphere in winter.

Tidal Modulation of Energy Dissipation Routes in the Gulf Stream

AbstractThe Gulf Stream (GS) is a powerful ocean current that is instrumental in regulating the global climate. While a correct reproduction of GS dynamics is contingent on an appropriate representation of energy dissipation, the specific role of tides in dissipation pathways of the wind‐driven circulation is not well understood. Here, we present evidence, using high‐resolution ocean simulations (∼2 km grid spacing), that ocean tides, by generating internal gravity waves, increase the forward cascade of energy in the GS region. This effect is greatest in summer, when the intensity of internal tides increases. However, the dissipation route associated with the forward energy cascade remains an order of magnitude weaker than frictional dissipation near the surface and bottom boundaries.

Near-Surface Wind Convergence over the Gulf Stream—The Role of SST Revisited

Abstract High-resolution observations have demonstrated the presence of strong time-mean near-surface wind convergence (NSWC) anchored across oceanic frontal zones, such as the western boundary currents. Initial analyses appeared to show a close association between this time-mean NSWC and time-mean properties of the underlying sea surface temperature (SST), such as the gradients and second derivatives (e.g., Laplacian of SST), acting through pressure-adjustment and vertical-mixing mechanisms. However, a series of recent papers have revealed the instantaneous NSWC to be dominated by atmospheric fronts and have suggested the importance of air–sea processes occurring instead on shorter, synoptic time scales. In this paper, using the ERA5 reanalysis dataset in the Gulf Stream region, we aim to reconcile these viewpoints by investigating the spatial and temporal dependence of NSWC and its relationship to SST. It is revealed that while atmospheric frontal processes govern the day-to-day variability of NSWC, the relatively weak but persistent pressure-adjustment and vertical-mixing mechanisms provide lower-frequency modulations in conditions both with and without atmospheric fronts. In addition to their temporal characteristics, each mechanism is shown through spectral analysis to dominate on specific spatial scales. In light of recent work that has tied remote atmospheric responses to NSWC anomalies in western boundary current regions, these results emphasize the importance of oceanic frontal zones for atmospheric variability on all spatiotemporal scales.

Skillful Multiyear Prediction of the Kuroshio and Gulf Stream Jets and Eddy Activity

AbstractThe Kuroshio and Gulf Stream (GS), Northern Hemisphere warm ocean currents, have narrow bands of intense jets and vigorous eddies. While the importance of them in regulating physical and biological ocean systems and climate has been increasingly recognized, the operational predictability of year‐to‐year variations in these jets and eddies has not been quantitatively assessed yet. Through a series of experiments using an eddy‐resolving dynamical ocean forecasting system, here we show that observed year‐to‐year variations in the intensities of the jets and eddies in the Kuroshio Extension and GS regions can be skillfully predicted up to a lead time of about 2 years. Sources of predictability mostly come from initial ocean memories, and variations in local atmospheric conditions play a secondary role. Our results demonstrate that the observed multiyear variations in ocean jets and eddies in these regions are operationally predictable when ocean models are properly initialized with sufficient spatial resolution.

Impact of Stochastic Ocean Density Corrections on Air‐Sea Flux Variability

AbstractAir‐sea flux variability has contributions from both ocean and atmosphere at different spatio‐temporal scales. Atmospheric synoptic scales and the air‐sea turbulent heat flux that they drive are well represented in climate models, but ocean mesoscales and their associated variability are often not well resolved due to non‐eddy‐resolving spatial resolutions of current climate models. We deploy a physics‐based stochastic subgrid‐scale parameterization for ocean density, that reinforces the lateral density variations due to oceanic eddies, and examine its effect on air‐sea heat flux variability in a comprehensive coupled climate model. The stochastic parameterization substantially modifies sea surface temperature (SST) and latent heat flux (LHF) variability and their co‐variability, primarily at scales near the resolution of the ocean model grid. Enhancement in the SST‐LHF anomaly covariance, and correlations, indicate that the ocean‐intrinsic component of the air‐sea heat flux variability is more consistent with high‐resolution satellite observations, especially in Gulf Stream region.

North Atlantic subtropical mode water formation controlled by Gulf Stream fronts.

The North Atlantic Ocean hosts the largest volume of global subtropical mode waters (STMWs) in the world, which serve as heat, carbon and oxygen silos in the ocean interior. STMWs are formed in the Gulf Stream region where thermal fronts are pervasive and result in feedback with the atmosphere. However, their roles in STMW formation have been overlooked. Using eddy-resolving global climate simulations, we find that suppressing local frontal-scale ocean-to-atmosphere (FOA) feedback leads to STMW formation being reduced almost by half. This is because FOA feedback enlarges STMW outcropping, attributable to the mixed layer deepening associated with cumulative excessive latent heat loss due to higher wind speeds and greater air-sea humidity contrast driven by the Gulf Stream fronts. Such enhanced heat loss overshadows the stronger restratification induced by vertical eddies and turbulent heat transport, making STMW colder and heavier. With more realistic representation of FOA feedback, the eddy-present/rich coupled global climate models reproduce the observed STMWs much better than the eddy-free ones. Such improvement in STMW production cannot be achieved, even with the oceanic resolution solely refined but without coupling to the overlying atmosphere in oceanic general circulation models. Our findings highlight the need to resolve FOA feedback to ameliorate the common severe underestimation of STMW and associated heat and carbon uptakes in earth system models.

Variability modes of September Arctic sea ice: drivers and their contributions to sea ice trend and extremes

The variability of September Arctic sea ice at interannual to multidecadal time scales in the midst of anthropogenically forced sea ice decline is not fully understood. Understanding Arctic sea ice variability at different time scales is crucial for better predicting future sea ice conditions and separating the externally forced signal from internal variability. Here, we study modes of variability, extreme events and trend in September Arctic sea ice in 100–150 year datasets by using time-frequency analysis. We extract the non-linear trend for sea ice area and provide an estimate for the sea ice loss driven by anthropogenic warming with a rate of ∼−0.25 million km2 per decade in the 1980s and accelerating to ∼−0.47 million km2 per decade in 2010s. Assuming the same accelerating rate for sea ice loss in the future and excluding the contributions of internal variability and feedbacks, a September ice-free Arctic could occur around 2060. Results also show that changes in sea ice due to internal variability can be almost as large as forced changes. We find dominant modes of sea ice variability with approximated periods of around 3, 6, 18, 27 and 55 years and show their contributions to sea ice variability and extremes. The main atmospheric and oceanic drivers of sea ice modes include the Arctic Oscillation and Arctic dipole anomaly for the 3 year mode, variability of sea surface temperature (SST) in the Gulf Stream region for the 6-year mode, decadal SST variability in the northern North Atlantic Ocean for the 18-year mode, Pacific Decadal Oscillation for the 27 year mode, and Atlantic Multidecadal Oscillation for the 55 year mode. Finally, our analysis suggests that over 70% of the sea ice area loss between the two extreme cases of 1996 (extreme high) and 2007 (extreme low) is caused by internal variability, with half of this variability being related to interdecadal modes.

Reconstructing subsurface temperature profiles with sea surface data worldwide through deep evidential regression methods

In this paper, the combination of the deep evidential regression network (DERN) and empirical orthogonal function (EOF) methods, namely, the DERN-EOF method, was proposed to reconstruct the subsurface temperature profiles. Firstly, the DERN method was compared with the previous machine learning methods used in the subsurface temperature field reconstruction, for example the Self-organizing map method. The new network was shown to have certain advantages of higher accuracy and the capability to provide uncertainty estimation, which enables the reconstruction of the subsurface field with higher belief. Analysis of the method performance suggests that the DERN could improve reconstruction accuracy by almost 20% and that the uncertainty statistics agrees well with the data distribution. Then the DERN method combined with the EOF method was used to reconstruct subsurface temperature profiles at a global scale. Results suggest that the reconstruction error in highly dynamic regions, like the Kuroshio Extension and Gulf Stream regions, is significantly higher than that in other regions, and the uncertainties increase with the reconstruction error, suggesting that the proposed method can provide a relatively believable uncertainty estimation.

The emergence of the Gulf Stream and interior western boundary as key regions to constrain the future North Atlantic carbon uptake

Abstract. In recent years, the growing number of available climate models and future scenarios has led to emergent constraints becoming a popular tool to constrain uncertain future projections. However, when emergent constraints are applied over large areas, it is unclear (i) if the well-performing models simulate the correct dynamics within the considered area, (ii) which key dynamical features the emerging constraint is stemming from, and (iii) if the observational uncertainty is low enough to allow for a considerable reduction in the projection uncertainties. We therefore propose to regionally optimize emergent relationships with the twofold goal to (a) identify key model dynamics associated with the emergent constraint and model inconsistencies around them and (b) provide key areas where a narrow observational uncertainty is crucial for constraining future projections. Here, we consider two previously established emergent constraints of the future carbon uptake in the North Atlantic (Goris et al., 2018). For the regional optimization, we use a genetic algorithm and pre-define a suite of shapes and size ranges for the desired regions. Independent of pre-defined shape and size range, the genetic algorithm persistently identifies the Gulf Stream region centred around 30∘ N as optimal as well as the region associated with broad interior southward volume transport centred around 26∘ N. Close to and within our optimal regions, observational data of volume transport are available from the RAPID array with relative low observational uncertainty. Yet, our regionally optimized emergent constraints show that additional measures of specific biogeochemical variables along the array will fundamentally improve our estimates of the future carbon uptake in the North Atlantic. Moreover, our regionally optimized emergent constraints demonstrate that models that perform well for the upper-ocean volume transport and related key biogeochemical properties do not necessarily reproduce the interior-ocean volume transport well, leading to inconsistent gradients of key biogeochemical properties. This hampers the applicability of emergent constraints over large areas and highlights the need to additionally evaluate spatial model features.

Towards the Upper-Ocean Unbalanced Submesoscale Motions in the Oleander Observations

Abstract Oceanic submesoscale dynamics with horizontal scales < 20 km have similar temporal and spatial scales as internal gravity waves (IGWs), but they differ dynamically and have distinct impacts on the ocean. Separating unbalanced submesoscale motions (USMs), quasi-balanced submesoscale motions (QBMs), and IGWs in observations remains a great challenge. Based on the wave–vortex decomposition and the vertical scale separation approach for distinguishing IGWs and USMs, the long-term repeat Oleander observations in the Gulf Stream region provide an opportunity to quantify these processes separately. Here in this study, the role of USMs in the divergence is emphasized, which has confounded the wave–vortex decomposition of wintertime data in previous analyses. We also adopt the vertical filtering approach to identify the USMs by applying a high-pass filter to the vertical scales, as USMs are characterized by smaller vertical scales. This approach is tested with submesoscale-permitting model data to confirm its effectiveness in filtering the submesoscale velocity perturbations, before being applied to the compiled velocity data of the Oleander dataset (years 2005–18). The results show that the averaged submesoscale eddy kinetic energy by USMs can reach ∼1 × 10−3 m2 s−2 at z = −30 m in winter, much stronger than found in other seasons. Importantly, this study exemplifies the possibility of obtaining USMs from existing ADCP observations and reveals the seasonal dynamical regimes for the submesoscales.

MESI: A Multiparameter Eddy Significance Index

AbstractIn this study, we introduce the Multiparameter Eddy Significance Index (MESI) for use in conjunction with mesoscale eddy tracking to estimate the impact of mesoscale eddies on the upper ocean and marine ecosystems. MESI combines blended satellite observations of sea level anomalies, sea surface temperature, sea surface salinity, and ocean color chlorophyll‐a and is a normalized index. When used in conjunction with mesoscale eddy tracking, MESI can highlight those eddies which extend into the deeper ocean and have a greater effect on nutrient cycling. Preliminary analysis in the Gulf Stream region of the western North Atlantic shows that MESI covaries with model estimates of nitrate, phosphate, iron, and pH with cross‐correlation values between 76% and 95%. These results suggest that when MESI is used in concert with eddy tracking, it can be used as an indicator of eddies and areas that are more likely to impact nutrient cycling in the upper ocean. An additional analysis of MESI values inside eddies during Hurricane Dorian (2019) along the Carolina coast further suggests that MESI is most effective when used to enhance eddy tracking as it provides valuable insight into mesoscale eddy activity and the upper ocean circulation.

Downscaling of ocean fields by fusion of heterogeneous observations using Deep Learning algorithms

We present a deep learning method to downscale low-resolution geophysical fields by merging them with high-resolution data. The downscaling was performed using an ensemble of convolutional neural networks (CNNs), whose prediction values are the average values of the outputs of 20 CNNs. Academic experiments were conducted on simulated ocean data in the Gulf Stream region, given by the outputs of the NATL60 model. The CNNs forced with low-resolution (120 × 120 km) sea surface high (SSH) data and mesoscale resolution (12 × 12 km) sea surface temperature (SST) data allowed us to obtain mesoscale resolution sea surface currents with good accuracy. Sensitivity experiments have shown that taking SST into account significantly increases the accuracy of the high-resolution velocity retrieval, even when noise is added to the SSH data. The velocity information embedded in the transport equation modeling the SST advection is taken into account by the CNN, which greatly increases the resolution of ocean currents provided by SSH. In the present work, we only consider spatial downscaling by assuming that SSH and SST are daily observations. The method we developed is generic and can be used to improve the resolution of a wide variety of large-scale fields by merging them with high-resolution fields.