Regionality and seasonality of submesoscale and mesoscale turbulence in the North Pacific Ocean

The kinetic energy (KE) seasonality has been revealed by satellite altimeters in many oceanic regions. Question about the mechanisms that trigger this seasonality is still challenging. We address this question through the comparison of two numerical simulations. The first one, with a 1/10° horizontal grid spacing, 54 vertical levels, represents dynamics of physical scales larger than 50 km. The second one, with a 1/30° grid spacing, 100 vertical levels, takes into account the dynamics of physical scales down to 16 km. Comparison clearly emphasizes in the whole North Pacific Ocean, not only a significant KE increase by a factor up to three, but also the emergence of seasonal variability when the scale range 16–50 km (called submesoscales in this study) is taken into account. But the mechanisms explaining these KE changes display strong regional contrasts. In high KE regions, such the Kuroshio Extension and the western and eastern subtropics, frontal mixed-layer instabilities appear to be the main mechanism for the emergence of submesoscales in winter. Subsequent inverse kinetic energy cascade leads to the KE seasonality of larger scales. In other regions, in particular in subarctic regions, results suggest that the KE seasonality is principally produced by larger-scale instabilities with typical scales of 100 km and not so much by smaller-scale mixed-layer instabilities. Using arguments from geostrophic turbulence, the submesoscale impact in these regions is assumed to strengthen mesoscale eddies that become more coherent and not quickly dissipated, leading to a KE increase.     

A review of the LATEX project: mesoscale to submesoscale processes in a coastal environment

The main objective of the LAgrangian Transport EXperiment (LATEX) project was to study the influence of coastal mesoscale and submesoscale physical processes on circulation dynamics, cross-shelf exchanges, and biogeochemistry in the western continental shelf of the Gulf of Lion, Northwestern Mediterranean Sea. LATEX was a five-year multidisciplinary project based on the combined analysis of numerical model simulations and multi-platform field experiments. The model component included a ten-year realistic 3D numerical simulation, with a 1 km horizontal resolution over the gulf, nested in a coarser 3 km resolution model. The in situ component involved four cruises, including a large-scale multidisciplinary campaign with two research vessels in 2010. This review concentrates on the physics results of LATEX, addressing three main subjects: (1) the investigation of the mesoscale to submesoscale processes. The eddies are elliptic, baroclinic, and anticyclonic; the strong thermal and saline front is density compensated. Their generation processes are studied; (2) the development of sampling strategies for their direct observations. LATEX has implemented an adaptive strategy Lagrangian tool, with a reference software available on the web, to perform offshore campaigns in a Lagrangian framework; (3) the quantification of horizontal mixing and cross-shelf exchanges. Lateral diffusivity coefficients, calculated in various ways including a novel technique, are in the range classically encountered for their associated scales. Cross-shelf fluxes have been calculated, after retrieving the near-inertial oscillation contribution. Further perspectives are discussed, especially for the ongoing challenge of studying submesoscale features remotely and from in situ data.     

Submesoscale features and their interaction with fronts and internal tides in a high-resolution coupled atmosphere-ocean-wave model of the Bay of Bengal

Large freshwater fluxes into the Bay of Bengal by rainfall and river discharges result in strong salinity fronts in the bay. In this study, a high-resolution coupled atmosphere-ocean-wave model with comprehensive physics is used to model the weather, ocean circulation, and wave field in the Bay of Bengal. Our objective is to explore the submesoscale activity that occurs in a realistic coupled model that resolves mesoscales and allows part of the submesoscale field. Horizontal resolution in the atmosphere varies from 2 to 6 km and is 13 km for surface waves, while the ocean model is submesoscale permitting with resolutions as high as 1.5 km and a vertical resolution of 0.5 m in the upper 10 m. In this paper, three different cases of oceanic submesoscale features are discussed. In the first case, heavy rainfall and intense downdrafts produced by atmospheric convection are found to force submesoscale currents, temperature, and salinity anomalies in the oceanic mixed layer and impact the mesoscale flow. In a second case, strong solitary-like waves are generated by semidiurnal tides in the Andaman Sea and interact with mesoscale flows and fronts and affect submesoscale features generated along fronts. A third source of submesoscale variability is found further north in the Bay of Bengal where river outflows help maintain strong salinity gradients throughout the year. For that case, a comparison with satellite observations of sea surface height anomalies, sea surface temperature, and chlorophyll shows that the model captures the observed mesoscale eddy features of the flow field, but in addition, submesoscale upwelling and downwelling patterns associated with ageostrophic secondary circulations along density fronts are also captured by the model.     

Correction to: Observation of submesoscale turbulence in a cyclonic eddy

Ocean Dynamics - In this study, we report results from a high-resolution measurement of submesoscale turbulence in a mesoscale eddy. The kinetic energy spectra in the mesoscale eddy are decomposed...     

Effects of cyclonic mesoscale eddies on the marine ecosystem in the Kuroshio Extension region using an eddy-resolving coupled physical-biological model

Effects of mesoscale eddies on the marine ecosystem in the Kuroshio Extension (KE) region are investigated using an eddy-resolving coupled physical-biological model. The model captures the seasonal and intra-seasonal variability of chlorophyll distribution associated with the mesoscale eddies, front variability, Kuroshio meanders, and upwelling. The model also reproduces the observed interannual variability of sea surface height anomaly (SSHA) in the KE region along a zonal band of 32–34°N from 2002 to 2006. The distribution of high surface chlorophyll corresponds to low SSHA. Cyclonic eddies are found to detach from the KE jet near 150°E and 158°E and propagate westward. The westward propagating cyclonic eddies lift the nutrient-rich thermocline into the euphotic zone and maintain high levels of chlorophyll in summer. In the subsurface layer, the pattern in chlorophyll is influenced by both lateral and vertical advection. In winter, convection inside the eddy entrains high levels of nutrients into the mixed layer, increasing production, and resulting in high chlorophyll concentration throughout the surface mixed layer. There is significant interannual variability in both the cyclonic eddy activity and the surface phytoplankton bloom south of the KE jet, although whether or not there is a causal link is unclear.     

Regional turbulence patterns driven by meso- and submesoscale processes in the Caribbean Sea

The surface ocean circulation in the Caribbean Sea is characterized by the interaction between anticyclonic eddies and the Caribbean Upwelling System (CUS). These interactions lead to instabilities that modulate the transfer of kinetic energy up- or down-cascade. The interaction of North Brazil Current rings with the islands leads to the formation of submesoscale vorticity filaments leeward of the Lesser Antilles, thus transferring kinetic energy from large to small scales. Within the Caribbean, the upper ocean dynamic ranges from large-scale currents to coastal upwelling filaments and allow the vertical exchange of physical properties and supply KE to larger scales. In this study, we use a regional model with different spatial resolutions (6, 3, and 1 km), focusing on the Guajira Peninsula and the Lesser Antilles in the Caribbean Sea, in order to evaluate the impact of submesoscale processes on the regional KE energy cascade. Ageostrophic velocities emerge as the Rossby number becomes O(1). As model resolution is increased submesoscale motions are more energetic, as seen by the flatter KE spectra when compared to the lower resolution run. KE injection at the large scales is greater in the Guajira region than in the others regions, being more effectively transferred to smaller scales, thus showing that submesoscale dynamics is key in modulating eddy kinetic energy and the energy cascade within the Caribbean Sea.     

Intercomparison of oceanic and atmospheric forced and coupled mesoscale simulations. Part I: Surface fluxes

A mesoscale non-hydrostatic atmospheric model has been coupled with a mesoscale oceanic model. The case study is a four-day simulation of a strong storm event observed during the SEMAPHORE experiment over a 500 × 500 km² domain. This domain encompasses a thermohaline front associated with the Azores current. In order to analyze the effect of mesoscale coupling, three simulations are compared: the first one with the atmospheric model forced by realistic sea surface temperature analyses; the second one with the ocean model forced by atmospheric fields, derived from weather forecast re-analyses; the third one with the models being coupled. For these three simulations the surface fluxes were computed with the same bulk parametrization. All three simulations succeed well in representing the main oceanic or atmospheric features observed during the storm. Comparison of surface fields with in situ observations reveals that the winds of the fine mesh atmospheric model are more realistic than those of the weather forecast re-analyses. The low-level winds simulated with the atmospheric model in the forced and coupled simulations are appreciably stronger than the re-analyzed winds. They also generate stronger fluxes. The coupled simulation has the strongest surface heat fluxes: the difference in the net heat budget with the oceanic forced simulation reaches on average 50 Wm⁻² over the simulation period. Sea surface-temperature cooling is too weak in both simulations, but is improved in the coupled run and matches better the cooling observed with drifters. The spatial distributions of sea surface-temperature cooling and surface fluxes are strongly inhomogeneous over the simulation domain. The amplitude of the flux variation is maximum in the coupled run. Moreover the weak correlation between the cooling and heat flux patterns indicates that the surface fluxes are not responsible for the whole cooling and suggests that the response of the ocean mixed layer to the atmosphere is highly non-local and enhanced in the coupled simulation.     

Using an eddy-tracking algorithm to understand the impact of assimilating altimetry data on the eddy characteristics of the Agulhas system

A complex and highly dynamical ocean region, the Agulhas Current System plays an important role in the transfer of energy, nutrients and organic material from the Indian to the Atlantic Ocean. Its dynamics are not only important locally, but affect the global ocean-atmosphere system. In working towards improved ocean reanalysis and forecasting capabilities, it is important that numerical models simulate mesoscale variability accurately—especially given the scarcity of coherent observational platforms in the region. Data assimilation makes use of scarce observations, a dynamical model and their respective error statistics to estimate a new, improved model state that minimises the distance to the observations whilst preserving dynamical consistency. Qualitatively, it is unclear whether this minimisation directly translates to an improved representation of mesoscale dynamics. In this study, the impact of assimilating along-track sea-level anomaly (SLA) data into a regional Hybrid Coordinate Ocean Model (HYCOM) is investigated with regard to the simulation of mesoscale eddy characteristics. We use an eddy-tracking algorithm and compare the derived eddy characteristics of an assimilated (ASSIM) and an unassimilated (FREE) simulation experiment in HYCOM with gridded satellite altimetry-derived SLA data. Using an eddy tracking algorithm, we are able to quantitatively evaluate whether assimilation updates the model state estimate such that simulated mesoscale eddy characteristics are improved. Additionally, the analysis revealed limitations in the dynamical model and the data assimilation scheme, as well as artefacts introduced from the eddy tracking scheme. With some exceptions, ASSIM yields improvements over FREE in eddy density distribution and dynamics. Notably, it was found that FREE significantly underestimates the number of eddies south of Madagascar compared to gridded altimetry, with only slight improvements introduced through assimilation, highlighting the models’ limitation in sustaining mesoscale activity in this region. Interestingly, it was found that the threshold for the maximum eddy propagation velocity in the eddy detection scheme is often exceeded when data assimilation relocates an eddy, causing the algorithm to interpret the discontinuity as eddy genesis, which directly influences the eddy count, lifetime and propagation velocity, and indirectly influences other metrics such as non-linearity. Finally, the analysis allowed us to separate eddy kinetic energy into contributions from detected mesoscale eddies and meandering currents, revealing that the assimilation of SLA has a greater impact on mesoscale eddies than on meandering currents.     

Multi-nest high-resolution model of submesoscale circulation features in the Gulf of Taranto

Recent oceanographic field measurements and high-resolution numerical modelling studies have revealed intense, transient, submesoscale motions characterised by a horizontal length scale of 100–10,000 m. This submesoscale activity increases in the fall and winter when the mixed layer (ML) depth is at its maximum. In this study, the submesoscale motions associated with a large-scale anticyclonic gyre in the central Gulf of Taranto were examined using realistic submesoscale-permitting simulations. We used realistic flow field initial conditions and multiple nesting techniques to perform realistic simulations, with very-high horizontal resolutions (> 200 m) in areas with submesoscale variability. Multiple downscaling was used to increase resolution in areas where instability was active enough to develop multi-scale interactions and produce 5-km-diameter eddies. To generate a submesoscale eddy, a 200-m resolution was required. The submesoscale eddy was formed through small-scale baroclinic instability in the rim of a large-scale anticyclonic gyre leading to large vertical velocities and rapid restratification of the ML in a time-scale of days. The submesoscale eddy was confirmed by observational data from the area and we can say that for the first time we have a proof that the model reproduces a realistic submesoscale vortex, similar in shape and location to the observed one.     

Mechanisms controlling the seasonal mixed-layer temperature and salinity of the Indonesian seas

We examine the seasonal mixed-layer temperature (MLT) and salinity (MLS) budgets in the Banda–Arafura Seas region (120–138° E, 8–3° S) using an ECCO ocean-state estimation product. MLT in these seas is relatively high during November–May (austral spring through fall) and relatively low during June–September (austral winter and the period associated with the Asian summer monsoon). Surface heat flux makes the largest contribution to the seasonal MLT tendency, with significant reinforcement by subsurface processes, especially turbulent vertical mixing. Temperature declines (the MLT tendency is negative) in May–August when seasonal insolation is smallest and local winds are strong due to the southeast monsoon, which causes surface heat loss and cooling by vertical processes. In particular, Ekman suction induced by local wind stress curl raises the thermocline in the Arafura Sea, bringing cooler subsurface water closer to the base of the mixed layer where it is subsequently incorporated into the mixed layer through turbulent vertical mixing; this has a cooling effect. The MLT budget also has a small, but non-negligible, semi-annual component since insolation increases and winds weaken during the spring and fall monsoon transitions near the equator. This causes warming via solar heating, reduced surface heat loss, and weakened turbulent mixing compared to austral winter and, to a lesser extent, compared to austral summer. Seasonal MLS is dominated by ocean processes rather than by local freshwater flux. The contributions by horizontal advection and subsurface processes have comparable magnitudes. The results suggest that ocean dynamics play a significant part in determining both seasonal MLT and MLS in the region, such that coupled model studies of the region should use a full ocean model rather than a slab ocean mixed-layer model.     

Geophysical studies of the global electric circuit

Physical mechanisms of formation of the global electric circuit (GEC) are considered and energy estimates are obtained for aeroelectric processes. Global thunderstorm activity, the electrodynamics of mesoscale convective systems, the electric fields of the magnetospheric dynamo and ionospheric dynamo region, and the electrostatic field of the global unipolar generator form a quasi-stationary aeroelectric state and maintain the balance of currents of the GEC atmospheric interval. In essence, the GEC is an open dissipative system including microphysical and electrohydrodynamic processes of generation and dissipation of the aeroelectric energy. The atmospheric electric field in the range of short-period aeroelectric pulsations has power-law spectra and contains coherent aeroelectric structures. The main GEC characteristics can serve as an indicator of a stationary state and spatiotemporal dynamics of atmospheric processes.     

Lateral stirring of large-scale tracer fields by altimetry

Ocean surface fronts and filaments have a strong impact on the global ocean circulation and biogeochemistry. Surface Lagrangian advection with time-evolving altimetric geostrophic velocities can be used to simulate the submesoscale front and filament structures in large-scale tracer fields. We study this technique in the Southern Ocean region south of Tasmania, a domain marked by strong meso- to submesoscale features such as the fronts of the Antarctic Circumpolar Current (ACC). Starting with large-scale surface tracer fields that we stir with altimetric velocities, we determine ‘advected’ fields which compare well with high-resolution in situ or satellite tracer data. We find that fine scales are best represented in a statistical sense after an optimal advection time of ～2 weeks, with enhanced signatures of the ACC fronts and better spectral energy. The technique works best in moderate to high EKE regions where lateral advection dominates. This technique may be used to infer the distribution of unresolved small scales in any physical or biogeochemical surface tracer that is dominated by lateral advection. Submesoscale dynamics also impact the subsurface of the ocean, and the Lagrangian advection at depth shows promising results. Finally, we show that climatological tracer fields computed from the advected large-scale fields display improved fine-scale mean features, such as the ACC fronts, which can be useful in the context of ocean modelling.     

The Mutual Response of the Tropical Squall Line and the Ocean

—The Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) is used to investigate the mutual response of a tropical squall line and the ocean. Simulated squall line compares well with the observations, and consists of counterrotating vortices, and has a bow shape bulge toward the leading edge. In addition to these features, which are also shown in the previous numerical simulations, the unique results from the coupled simulation indicate that the air–sea interaction processes within the squall line are important. They affect both the atmosphere and the ocean locally. Simulated upper ocean displays significant response to the squall line with upwelling and baroclinicity. Depth of the ocean mixed layer in the coupled simulation becomes modified due to feedback processes. Ocean temperature acts as a destabilizing factor, and the salinity as a stabilizing factor. Surface turbulent fluxes from the coupled simulation are about 10% less than that of the uncoupled simulation. The SST in the coupled simulation decreases by about 0.21°C. Predicted squall line in the coupled simulation is weaker as compared to the uncoupled simulation. This is reflected in terms of differences in surface fluxes, cloud water, rain water and vertical velocities between the two simulations.     

Variability of the Subtropical Front in the Indian Ocean south of Australia

A detailed high resolution survey of a small region (68 × 68 km) of the Subtropical Front south of Australia over a period of 14 days is used to study the interaction between the mixed layer and the permanent frontal structure underneath during summer conditions. The front extends through the mixed layer as a salinity front, while its temperature structure is modified by seasonal warming. Wind-driven movement of the mixed layer combines with the short-time development of indentations and filaments in the front to produce some degree of decoupling between the mixed layer and the underlying structure, and the front is not always found at the same location in and below the mixed layer. Intrusions and parcels of distinct water properties are found just below the mixed layer, produced as a result of the relative movement of the front in and below the mixed layer. These parcels are typically 10 km in width and 10–50 m in depth. Successive surveys of the front with a time separation of 2 days showed that these features persist over at least 1 week. Large scale surveys of the front show that parcels are ubiquitous along the Subtropical Front over a distance of several hundred kilometres. The results suggest that any study aimed at understanding the intricate interaction between the mixed layer and the layers below in oceanic fronts will have to address wind-driven dynamics and frontal dynamics together.This revised version was published online February 2005 with corrections to figures. Unfortunately the figures were reproduced in black and white and in the new version they are in color.     

The transition from symmetric to baroclinic instability in the Eady model

Here, we explore the transition from symmetric instability to ageostrophic baroclinic instability in the Eady model; an idealised representation of a submesoscale mixed layer front. We revisit the linear stability problem considered by Stone (J Atmos Sci, 23, 390–400, (Stone 1966)), Stone (J Atmos Sci, 27, 721–726, (Stone 1970)), Stone (J Atmos Sci, 29, 419–426, (Stone 1972)) with a particular focus on three-dimensional ‘mixed modes’ (which are neither purely symmetric or baroclinic) and find that these modes can have growth rates within just a few percent of the corresponding two-dimensional growth rate maximum. In addition, we perform very high resolution numerical simulations allowing an exploration of the transition from symmetric to baroclinic instability. Three-dimensional mixed modes represent the largest contribution to the turbulent kinetic energy during the transition period between symmetric and baroclinic instability. In each simulation, we see the development of sharp fronts with associated high rms vertical velocities of up to 30 mm s^?1. Furthermore, we see significant transfer of energy to small scales, demonstrated by time-integrated mixing and energy dissipation by small-scale three-dimensional turbulence totalling about 30 % of the initial kinetic energy in all cases.     

A case study of the mesoscale dynamics in the North-Western Mediterranean Sea: a combined data–model approach

The Northern current is the main circulation feature of the North-Western Mediterranean Sea. While the large-scale to mesoscale variability of the northern current (NC) is well known and widely documented for the Ligurian region, off Nice or along the Gulf of Lions shelf, few is known about the current instabilities and its associated mesoscale dynamics in the intermediate area, off Toulon. Here, we took advantage of an oceanographic cruise of opportunity, the start of a HF radar monitoring programme in the Toulon area and the availability of regular satellite sea surface temperature and chlorophyll a data, to evaluate the realism of a NEMO-based regional high-resolution model and the added value brought by HF radar. The combined analysis of a 1/64° configuration, named GLAZUR64, and of all data sets revealed the occurrence of an anticyclonic coastal trapped eddy, generated inside a NC meander and passing the Toulon area during the field campaign. We show that this anticyclonic eddy is advected downstream along the French Riviera up to the study region and disturbs the Northern current flow. This study aims to show the importance of combining observations and modelling when dealing with mesoscale processes, as well as the importance of high-resolution modelling.     

Tidal wave transformations in the German Bight

Mesoscale and submesoscale dynamics associated with tidal wave transformations were addressed in the German Bight using numerical simulations. Tidal gauge and velocity observations in several locations were used to validate the numerical model. A downscaling approach included analysis of simulations with horizontal resolutions of 1, 0.4, and 0.2 km. It was shown that the modified tidal wave lost most of its energy after reflection or refraction over the eastern part of the German Bight. Energy loss resulted in a pronounced change of the wave’s spectral composition and generation of overtides. Tidal oscillations were modified by mesoscale processes associated with bathymetric channels. Semidiurnal and quarterdiurnal tides revealed very different spatial patterns. The former were aligned with the bathymetric channels, while the latter were rather “patchy” and had about half the spatial scales. In numerous areas around the bathymetric channels, the major axis of the M₄ ellipses was normal or at some angle with the major axis of the M₂ ellipses. Thus, higher harmonics developed “orthogonal” patterns that drove secondary circulations. Moreover, the ratio between spring and neap tidal amplitudes was relatively low in the Wadden Sea, showing reduced sensitivity of this very shallow area to fortnightly tidal variations. It was demonstrated that simulated hydrodynamics patterns help explain the physical mechanism shaping the median grain size distribution in the German Bight.     

Using a dynamical advection to reconstruct a part of the SSH evolution in the context of SWOT,application to the Mediterranean Sea

The main oceanographic objective of the future SWOT mission is to better characterize the ocean mesoscale and sub-mesoscale circulation, by observing a finer range of ocean topography dynamics down to 20 km wavelength. Despite the very high spatial resolution of the future satellite, it will not capture the time evolution of the shorter mesoscale signals, such as the formation and evolution of small eddies. SWOT will have an exact repeat cycle of 21 days, with near repeats around 5–10 days, depending on the latitude. Here, we investigate a technique to reconstruct the missing 2D SSH signal in the time between two satellite revisits. We use the dynamical interpolation (DI) technique developed by Ubelmann et al. (2015). Based on potential vorticity (hereafter PV) conservation using a one and a half layer quasi-geostrophic model, it features an active advection of the SSH field. This model has been tested in energetic open ocean regions such as the Gulf Stream and the Californian Current, and has given promising results. Here, we test this model in the Western Mediterranean Sea, a lower energy region with complex small scale physics, and compare the SSH reconstruction with the high-resolution Symphonie model. We investigate an extension of the simple dynamical model including a separated mean circulation. We find that the DI gives a 16–18% improvement in the reconstruction of the surface height and eddy kinetic energy fields, compared with a simple linear interpolation, and a 37% improvement in the Northern Current subregion. Reconstruction errors are higher during winter and autumn but statistically, the improvement from the DI is also better for these seasons.     

Wind-induced subduction at the South Atlantic subtropical front

The South Atlantic Subtropical Front, associated with the eastward-flowing South Atlantic Current, separates the colder, nutrient-rich waters of the subpolar gyre from the warmer, nutrient-poor waters of the subtropical gyre. Perturbations to the quasi-geostrophic, eastward flow generate meanders and filaments which induce cross-frontal exchange of water properties. Down-front winds transport denser waters from the South over warm waters from the North, inducing convective instability and subduction. Such processes occur over spatial scales of the order of 1 km and thus require high horizontal spatial resolution. In this modeling study, a high-resolution (4 km) regional grid is embedded in a basin-wide configuration (12 km) of the South Atlantic Ocean in order to test the importance of submesoscale processes in water mass subduction along the subtropical front. Stronger and more numerous eddies obtained in the high-resolution run yield more intense zonal jets along the frontal zone. Such stronger jets are more susceptible to instabilities, frontogenesis, and the generation of submesoscale meanders and filaments with \(\mathcal {O}(1)\) Rossby number. As a consequence, vertical velocities larger than 100 md ¹ are obtained in the high-resolution run, one order of magnitude larger than in the low-resolution run. Wind-driven subduction occurs along the frontal region, associated with negative Ertel potential vorticity in the surface layer. Such processes are not observed in the low-resolution run. A passive tracer experiment shows that waters with density characteristics similar to subtropical mode waters are preferentially subducted along the frontal region. The wind-driven buoyancy flux is shown to be much larger than thermal or haline fluxes during the wintertime, which highlights the importance of the frictional component in extracting PV from the surface ocean and inducing subduction, a process that has been overlooked in subtropical mode water formation in the region.     

Deep Radiation Fog in a Wide Closed Valley: Study by Numerical Modeling and Remote Sensing

The Ebro river basin, in the northeastern part of the Iberian Peninsula in Europe, very often experiences radiation fog episodes in winter that can last for several days. The impact on human activities is high, especially on road and air transportation. The installation in July 2009 of a WindRASS in the area, which is able to work in the presence of fog, now allows inspecting the vertical structure of the temperature and wind profiles across the roughly 300-m-thick fog layer. We present a case study of a long-lasting (60 h) deep radiation fog that took place in December 2009 to obtain a deeper understanding of the dynamic processes governing such persistent fog. Field observations of vertical profiles of temperature, wind and turbulent kinetic energy are compared with a high-resolution mesoscale simulation, satellite imagery of fog distribution and observations taken in the area to understand why the fog is so persistent and how it dissipates only for a short period in the afternoon despite intermittent turbulence within the fog deck. The confinement of the fog inside a practically closed basin allows us to study the relevant physical processes in the establishment and subsequent evolution of the fog episode using a limited-area mesoscale model. The contribution of the WindRASS measurements allowed us to validate the numerical simulations, particularly inspecting the role of turbulence that can link the bottom and top of the fog through moderate episodic mixing. The fog layer has very weak winds inside, but is well mixed and experiences intermittent top-bottom turbulence generated in its upper part by convection due to radiative cooling and by wind shear due to the topographically generated flows that blow just above the top of the fog.