We study the motion of a viscoelastic fluid within a rigid spherical cavity with the aim of improving understanding of the motion of the vitreous humour in the human eye. The flow of vitreous humour leads to traction on the retina, which, once the retina is torn or damaged, can cause it to detach from the choroid, leading to loss of sight if left untreated. In the first part of the paper we investigate the relaxation behaviour of the fluid, the transient flow that would be observed in the stationary sphere starting from non-stationary initial conditions. For a general viscoelastic fluid we calculate the growth rates and eigenfunctions associated with the system, and we discuss two particular rheological models of the vitreous humour taken from the literature. In the second part of the paper we consider forced oscillations of the fluid, due to small-amplitude rotations of the sphere about a diameter, representing saccades of the eyeball. We conclude with a discussion of the possible occurrence of resonant phenomena and their clinical relevance.

Feedback control applications for flows with a large number of degrees of freedom require the reduction of the full flow model to a system with significantly fewer degrees of freedom. This model-reduction process is accomplished by Galerkin projections using a reduction basis composed of modal structures that ideally preserve the input–output behaviour between actuators and sensors and ultimately result in a stabilized compensated system. In this study, global modes are critically assessed as to their suitability as a reduction basis, and the globally unstable, two-dimensional flow over an open cavity is used as a test case. Four criteria are introduced to select from the global spectrum the modes that are included in the reduction basis. Based on these criteria, four reduced-order models are tested by computing open-loop (transfer function) and closed-loop (stability) characteristics. Even though weak global instabilities can be suppressed, the concept of reduced-order compensators based on global modes does not demonstrate sufficient robustness to be recommended as a suitable choice for model reduction in feedback control applications. The investigation also reveals a compelling link between frequency-restricted input–output measures of open-loop behaviour and closed-loop performance, which suggests the departure from mathematically motivated -measures for model reduction toward more physically based norms; a particular frequency-restricted input–output measure is proposed in this study which more accurately predicts the closed-loop behaviour of the reduced-order model and yields a stable compensated system with a markedly reduced number of degrees of freedom.

The fine-scale response of a subsurface stable stratified jet subject to the forcing of surface wind stress and surface cooling is investigated using direct numerical simulation. The initial velocity profile consists of a symmetric jet located below a surface layer driven by a constant wind stress. The initial density profile is well-mixed in the surface layer and linearly stratified in both upper and lower flanks of the jet. The minimum value of the gradient Richardson number in the upper flank of the jet exceeds the critical value of 0.25 for linear shear instability. Broadband finite-amplitude fluctuations are introduced to the surface layer to initiate the simulation. Turbulence is generated in the surface layer and deepens into the jet upper flank. Internal waves generated by the turbulent surface layer are observed to propagate downward across the jet. The momentum flux carried by the waves is significantly smaller than the Reynolds shear stress extracted from the background velocity. The wave energy flux is also smaller than the turbulence production by mean shear. Ejections of fluid parcels by horseshoe-like vortices cause intermittent patches of intense dissipation inside the jet upper flank where the background gradient Richardson number is larger than 0.25. Drag due to the wind stress is smaller than the drag caused by turbulent stress in the flow. Analysis of the mean and turbulent kinetic energy budgets suggests that the energy input by surface forcing is considerably smaller than the energy extracted from the initially imposed background shear in the surface layer.

Deep water waves in two-dimensional flow can have curvature singularities on the surface profile; for example, the limiting Stokes wave has a corner of radians and the limiting standing wave momentarily forms a corner of radians. Much less is known about the possible formation of curvature singularities in general. A novel way of exploring this possibility is to consider the curvature as a complex function of the complex arclength variable and to seek the existence and nature of any singularities in the complex arclength plane. Highly accurate boundary integral methods produce a Fourier spectrum of the curvature that allows the identification of the nearest singularity to the real axis of the complex arclength plane. This singularity is in general a pole singularity that moves about the complex arclength plane. It approaches the real axis very closely when waves break and is associated with the high curvature at the tip of the breaking wave. The behaviour of these singularities is more complex for standing waves, where two singularities can be identified that may collide and separate. One of them approaches the real axis very closely when a standing wave forms a very narrow collapsing column of water almost under free fall. In studies so far, no singularity reaches the real axis in finite time. On the other hand, the surface elevation has square-root singularities in the complex plane that do reach the real axis in finite time, the moment when a wave first starts to break. These singularities have a profound effect on the wave spectra.

Particle image velocimetry (PIV) is used to investigate the three-dimensional wakes of rigid pitching panels with a trapezoidal geometry, chosen to model idealized fish caudal fins. Experiments are performed for Strouhal numbers from 0.17 to 0.56 for two different trailing edge pitching amplitudes. A Lagrangian coherent structure (LCS) analysis is employed to investigate the formation and evolution of the panel wake. A classic reverse von Kármán vortex street pattern is observed along the mid-span of the near wake, but the vortices realign and exhibit strong interactions near the spanwise edges of the wake. At higher Strouhal numbers, the complexity of the wake increases downstream of the trailing edge as the spanwise vortices spread transversely and lose coherence as the wake splits. This wake transition is shown to correspond to a qualitative change in the LCS pattern surrounding each vortex core, and can be identified as a quantitative event that is not dependent on arbitrary threshold levels. The location of this transition is observed to depend on both the pitching amplitude and free stream velocity, but is not constant for a fixed Strouhal number. On the panel surface, the trapezoidal planform geometry is observed to create additional vortices along the swept edges that retain coherence for low Strouhal numbers or high sweep angles. These additional swept-edge structures are conjectured to add to the complex three-dimensional flow near the tips of the panel.

In order to investigate continuous vortex dynamics based on a Lagrangian-like formulation, we develop a theoretical framework and a numerical method for computation of the evolution of a vortex-surface field (VSF) in viscous incompressible flows with simple topology and geometry. Equations describing the continuous, timewise evolution of a VSF from an existing VSF at an initial time are first reviewed. Non-uniqueness in this formulation is resolved by the introduction of a pseudo-time and a corresponding pseudo-evolution in which the evolved field is ‘advected’ by frozen vorticity onto a VSF. A weighted essentially non-oscillatory (WENO) method is used to solve the pseudo-evolution equations in pseudo-time, providing a dissipative-like regularization. Vortex surfaces are then extracted as iso-surfaces of the VSFs at different real physical times. The method is applied to two viscous flows with Taylor–Green and Kida–Pelz initial conditions respectively. Results show the collapse of vortex surfaces, vortex reconnection, the formation and roll-up of vortex tubes, vorticity intensification between anti-parallel vortex tubes, and vortex stretching and twisting. A possible scenario for understanding the transition from a smooth laminar flow to turbulent flow in terms of topology of vortex surfaces is discussed.

The characteristics of the intense vorticity structures (IVSs) near the turbulent/non-turbulent (T/NT) interface separating the turbulent and the irrotational flow regions are analysed using a direct numerical simulation (DNS) of a turbulent plane jet. The T/NT interface is defined by the radius of the large vorticity structures (LVSs) bordering the jet edge, while the IVSs arise only at a depth of about from the T/NT interface, where is the Kolmogorov micro-scale. Deep inside the jet shear layer the characteristics of the IVSs are similar to the IVSs found in many other flows: the mean radius, tangential velocity and circulation Reynolds number are , , and , where , and are the root mean square of the velocity fluctuations and the Reynolds number based on the Taylor micro-scale, respectively. Moreover, as in forced isotropic turbulence the IVSs inside the jet are well described by the Burgers vortex model, where the vortex core radius is stable due to a balance between the competing effects of axial vorticity production and viscous diffusion. Statistics conditioned on the distance from the T/NT interface are used to analyse the effect of the T/NT interface on the geometry and dynamics of the IVSs and show that the mean radius , tangential velocity and circulation of the IVSs increase as the T/NT interface is approached, while the vorticity norm stays approximately constant. Specifically , and exhibit maxima at a distance of roughly one Taylor micro-scale from the T/NT interface, before decreasing as the T/NT is approached. Analysis of the dynamics of the IVS shows that this is caused by a sharp decrease in the axial stretching rate acting on the axis of the IVSs near the jet edge. Unlike the IVSs deep inside the shear layer, there is a small predominance of vortex diffusion over stretching for the IVSs near the T/NT interface implying that the core of these structures is not stable i.e. it will tend to grow in time. Nevertheless the Burgers vortex model can still be considered to be a good representation for the IVSs near the jet edge, although it is not as accurate as for the IVSs deep inside the jet shear layer, since the observed magnitude of this imbalance is relatively small.

The effect of an inclination angle of the shear with respect to the stratification on the linear properties of the shear instability is examined in the work. For this purpose, we consider a two-dimensional plane Bickley jet of width and maximum velocity in a stably stratified fluid of constant Brunt–Väisälä frequency in an inviscid and Boussinesq framework. The plane of the jet is assumed to be inclined with an angle with respect to the vertical direction of stratification. The stability analysis is performed using both numerical and theoretical methods for all the values of and Froude number . We first obtain that the most unstable mode is always a two-dimensional Kelvin–Helmholtz (KH) sinuous mode. The condition of stability based on the Richardson number , which reads here , is recovered for . But when , that is, when the directions of shear and stratification are not perfectly aligned, the Bickley jet is found to be unstable for all Froude numbers. We show that two modes are involved in the stability properties. We demonstrate that when is decreased below , there is a ‘jump’ from one two-dimensional sinuous mode to another. For small Froude numbers, we show that the shear instability of the inclined jet is similar to that of a horizontal jet but with a ‘horizontal’ length scale . In this regime, the characteristics (oscillation frequency, growth rate, wavenumber) of the most unstable mode are found to be proportional to . For large Froude numbers, the shear instability of the inclined jet is similar to that of a vertical jet with the same scales but with a different Froude number, . It is argued that these results could be valid for any type of shear flow.

A detailed study into the rheology and microstructure of dense suspensions of initially spherical capsules is presented, where capsules are composed of a fluid-filled interior surrounded by an elastic membrane. This study couples a lattice-Boltzmann fluid solver to a finite-element membrane model creating a robust and scalable method for the simulation of these suspensions. A Lees–Edwards boundary condition is used to simulate periodic simple shear to obtain bulk rheological properties, and three-dimensional results are presented for capsules in the regime of negligible inertia, Brownian motion and colloidal interparticle forces. The simulation results focus on describing the suspension rheology as a function of the particle concentration and deformability, and relating these macroscopic rheological findings to changes at the particle level, i.e. the suspension microstructure. Several important findings are made: suspensions of deformable capsules are found to be shear thinning, and the initially compressive normal stresses associated with rigid spherical suspensions undergo rapid changes with moderate levels of particle deformation. These normal stress changes are particularly evident in the first normal stress difference, which undergoes a sign change at fairly minor levels of deformation, and the particle pressure, which decreases rapidly with increasing particle deformability. Changes in the microstructure as quantified by the single-body microstructure and the pair distribution function are reported. Also, results calculating particle self-diffusion are presented and related to changes in the normal stresses.

The thermal effects on the wake flow behind a heated circular cylinder operating in the mixed convection regime were investigated experimentally in the present study. The experiments were conducted in a vertical water channel with the heated cylinder placed horizontally and the flow approaching the cylinder downwards. With such a flow arrangement, the direction of the thermally induced buoyancy force acting on the fluid surrounding the heated cylinder would be opposite to the approach flow. During the experiments, the temperature and Reynolds number of the approach flow were held constant. By adjusting the surface temperature of the heated cylinder, the corresponding Richardson number () was varied between 0.0 (unheated) and 1.04, resulting in a change in the heat transfer process from forced convection to mixed convection. A novel flow diagnostic technique, molecular tagging velocimetry and thermometry (MTV&T), was used for qualitative flow visualization of thermally induced flow structures and quantitative, simultaneous measurements of flow velocity and temperature distributions in the wake of the heated cylinder. With increasing temperature of the heated cylinder (i.e. Richardson number), significant modifications of the wake flow pattern and wake vortex shedding process were clearly revealed. When the Richardson number was relatively small (), the vortex shedding process in the wake of the heated cylinder was found to be quite similar to that of an unheated cylinder. As the Richardson number increased to , the wake vortex shedding process was found to be ‘delayed’, with the wake vortex structures beginning to shed much further downstream. As the Richardson number approached unity (), instead of having ‘Kármán’ vortices shedding alternately at the two sides of the heated cylinder, concurrent shedding of smaller vortex structures was observed in the near wake of the heated cylinder. The smaller vortex structures were found to behave more like ‘Kelvin–Helmholtz’ vortices than ‘Kármán’ vortices, and adjacent small vortices would merge to form larger vortex structures further downstream. It was also found that the shedding frequency of the wake vortex structures decreased with increasing Richardson number. The wake closure length and the drag coefficient of the heated cylinder were found initially to decrease slightly when the Richardson number was relatively small (), and then to increase monotonically with increasing Richardson number as the Richardson number became relatively large (). The average Nusselt number () of the heated cylinder was found to decrease almost linearly with increasing Richardson number.

In the absence of significant body forces the passive manipulation of fluid interfacial flows is naturally achieved by control of the specific geometry and wetting properties of the system. Numerous ‘microfluidic’ systems on Earth and ‘macrofluidic’ systems aboard spacecraft routinely exploit such methods and the term ‘capillary fluidics’ is used to describe both length-scale limits. In this work a collection of analytic solutions is offered for passive and weakly forced flows where a bulk capillary liquid is slowly drained or supplied by a faster capillary flow along at least one interior edge of the container. The solutions are enabled by an assumed known pressure (or known height) dynamical boundary condition. Following a series of assumptions this boundary condition can be in part determined a priori from the container dimensions and further quantitative experimental evidence, but not proof, is provided in support of its expanded use herein. In general, a small parameter arises in the scaling of the problems permitting a decoupling of the edge flow from the global bulk meniscus flow. The quasi-steady asymptotic system of equations that results may then be easily solved in closed form for a useful variety of geometries including uniform and tapered sections possessing at least one critically wetted interior edge. Draining, filling, bubble displacement and other imbibing flows are studied. Cursory terrestrial and drop tower experiments agree well with the solutions. The solutions are valued for the facility they provide in computing designs for selected capillary fluidics problems by way of passive transport rates and meniscus displacement. Because geometric permutations of any given design are myriad, such analytic tools are capable of efficiently identifying and comparing critical design criteria (i.e. shape and size) and the impact of various wetting conditions resulting from the fluid properties and surface conditions. Sample optimizations are performed to demonstrate the utility of the method.

Electrokinetic streaming-potential phenomena are driven by imposed relative motion between liquid electrolytes and charged solids. Owing to non-uniform convective ‘surface’ current within the Debye layer Ohmic currents from the electro-neutral bulk are required to ensure charge conservation thereby inducing a bulk electric field. This, in turn, results in electro-viscous drag enhancement. The appropriate modelling of these phenomena in the limit of thin Debye layers ( denoting the dimensionless Debye thickness) has been a matter of ongoing controversy apparently settled by Cox’s seminal analysis (J. Fluid Mech., vol. 338, 1997, p. 1). This analysis predicts electro-viscous forces that scale as resulting from the perturbation of the original Stokes flow with the Maxwell-stress contribution only appearing at higher orders. Using scaling analysis we clarify the distinction between the normalizations pertinent to field- and motion-driven electrokinetic phenomena, respectively. In the latter class we demonstrate that the product of the Hartmann & Péclet numbers is contrary to Cox (1997) where both parameters are assumed . We focus on the case where motion-induced fields are comparable to the thermal scale and accordingly present a singular-perturbation analysis for the limit where the Hartmann number is and the Péclet number is . Electric-current matching between the Debye layer and the electro-neutral bulk provides an inhomogeneous Neumann condition governing the electric field in the latter. This field, in turn, results in a velocity perturbation generated by a Smoluchowski-type slip condition. Owing to the dominant convection, the present analysis yields an asymptotic structure considerably simpler than that of Cox (1997): the electro-viscous effect now already appears at and is contributed by both Maxwell and viscous stresses. The present paradigm is illustrated for the prototypic problem of a sphere sedimenting in an unbounded fluid domain with the resulting drag correction differing from that calculated by Cox (1997). Independently of current matching, salt-flux matching between the Debye layer and the bulk domain needs also to be satisfied. This subtle point has apparently gone unnoticed in the literature, perhaps because it is trivially satisfied in field-driven problems. In the present limit this requirement seems incompatible with the uniform salt distribution in the convection-dominated bulk domain. This paradox is resolved by identifying the dual singularity associated with the limit in motion-driven problems resulting in a diffusive layer of thickness beyond the familiar -wide Debye layer.

We present planforms of line plumes formed on horizontal surfaces in turbulent convection, along with the length of line plumes measured from these planforms, in a six decade range of Rayleigh numbers () and at three Prandtl numbers (). Using geometric constraints on the relations for the mean plume spacings, we obtain expressions for the total length of near-wall plumes on horizontal surfaces in turbulent convection. The plume length per unit area (), made dimensionless by the near-wall length scale in turbulent convection (), remains constant for a given fluid. The Nusselt number is shown to be directly proportional to for a given fluid layer of height . The increase in has a weak influence in decreasing . These expressions match the measurements, thereby showing that the assumption of laminar natural convection boundary layers in turbulent convection is consistent with the observed total length of line plumes. We then show that similar relationships are obtained based on the assumption that the line plumes are the outcome of the instability of laminar natural convection boundary layers on the horizontal surfaces.

Detailed knowledge of the dynamics of vortex structures in an oscillatory boundary layer is essential for the correct modelling of transport processes in many engineering problems and, in particular, of the pick-up and transport of sediments at the bottom of sea waves. In the present contribution, the formation of turbulent spots in an oscillatory boundary layer is investigated by means of direct numerical simulations. Two of the laboratory experiments of Carstensen, Sumer and Fredsøe are reproduced and, after a comparison of the numerical results with laboratory measurements, a detailed and quantitative characterization of the turbulent spots is also given on the basis of further simulations. The speeds of the head () and tail () of the spots are found to scale with the instantaneous free stream velocity and to be similar to those observed in steady boundary layers. The ratios and seem to increase with the Reynolds number () while the streamwise expansion rate of the spots appears to be independent of .

This study offers an explanation of the recently observed effect of destabilization of free convective flows by weak rotation. After studying several models where flows are driven by the simultaneous action of convection and rotation, it is concluded that destabilization is observed in cases where the centrifugal force acts against the main convective circulation. At relatively low Prandtl numbers, this counter-action can split the main vortex into two counter-rotating vortices, where the interaction leads to instability. At larger Prandtl numbers, the counter-action of the centrifugal force steepens an unstable thermal stratification, which triggers the Rayleigh–Bénard instability mechanism. Both cases can be enhanced by advection of azimuthal velocity disturbances towards the axis, where they grow and excite perturbations of the radial velocity. The effect was studied by considering a combined convective and rotating flow in a cylinder with a rotating lid and a parabolic temperature profile at the sidewall. Next, explanations of the destabilization effect for rotating-magnetic-field-driven flow and melt flow in a Czochralski crystal growth model were derived.

Consider low-speed potential flow past a ship modelled as a semi-infinite two-dimensional body with constant draught. Is it possible to design the hull in such a way as to eliminate the waves produced downstream of the ship? In 1977, Vanden-Broeck & Tuck had conjectured that a single-cornered piecewise-linear hull will always generate a wake; in this paper, we show how recently developed tools in exponential asymptotics can be used to confirm this conjecture. In particular, we show how the formation of waves near a ship is a necessary consequence of singularities in the ship’s geometry (or its analytic continuation). Comprehensive numerical computations confirm the analytical predictions.

An experimental study of wall shear stress in an accelerating flow of water in a pipe ramping between two steady turbulent flows has been undertaken in a large-scale experimental facility. Ensemble averaged mean and r.m.s. of the turbulent fluctuations of wall shear stresses have been derived from hot-film measurements from many repeated runs. The initial Reynolds number and the acceleration rate were varied systematically to give values of a non-dimensional acceleration parameter ranging from 0.16 to 14. The wall shear stress has been shown to follow a three-stage development. Stage 1 is associated with a period of minimal turbulence response; the measured turbulent wall shear stress remains largely unchanged except for a very slow increase which is readily associated with the stretching of existing turbulent eddies as a result of flow acceleration. In this condition of nearly ‘frozen’ turbulence, the unsteady wall shear stress is driven primarily by flow inertia, initially increasing rapidly and overshooting the pseudo-steady value, but then increasing more slowly and eventually falling below the pseudo-steady value. This variation is predicted by an analytical expression derived from a laminar flow formulation. The start of Stage 2 is marked by the generation of new turbulence causing both the mean and turbulent wall shear stress to increase rapidly, although there is a clear offset between the responses of these two quantities. The turbulent wall shear, reflecting local turbulent activities near the wall, responds first and the mean wall shear, reflecting conditions across the entire flow field, responds somewhat later. In Stage 3, the wall shear stress exhibits a quasi-steady variation. The duration of the initial period of nearly frozen turbulence response close to the wall increases with decreasing initial Reynolds number and with increasing acceleration. The latter is in contrast to the response of turbulence in the core of the flow, which previous measurements have shown to be independent of the rate of acceleration.

The flow around a cylinder rolling or sliding on a wall was investigated analytically and numerically for small Reynolds numbers, where the flow is known to be two-dimensional and steady. Both prograde and retrograde rotation were analytically solved, in the Stokes regime, giving the values of forces and torque and a complete description of the flow. However, solving Navier–Stokes equation, a rotation of the cylinder near the wall necessarily induces a cavitation bubble in the nip if the fluid is a liquid, or compressible effects, if it is a gas. Therefore, an infinite lift force is generated, disconnecting the cylinder from the wall. The flow inside this interstice was then solved under the lubrication assumptions and fully described for a completely flooded interstice. Numerical results extend the analysis to higher Reynolds number. Finally, the effect of the upstream pressure on the onset of cavitation is studied, giving the initial location of the phenomenon and the relation between the upstream pressure and the flow rate in the interstice. It is shown that the flow in the interstice must become three-dimensional when cavitation takes place.

In the present study we investigate three-scalar mixing in a turbulent coaxial jet. In this flow a centre jet and an annular flow, consisting of acetone-doped air and ethylene respectively, are mixed with the co-flow air. A unique aspect of this study compared to previous studies of three-scalar mixing is that two of the scalars (the centre jet and air) are separated by the third (annular flow); therefore, this flow better approximates the mixing process in a non-premixed turbulent reactive flow. Planar laser-induced fluorescence and Rayleigh scattering are employed to measure the mass fractions of the acetone-doped air and ethylene. The results show that the most unique aspects of the three-scalar mixing occur in the near field of the flow. The mixing process in this part of the flow are analysed in detail using the scalar means, variances, correlation coefficient, joint probability density function (JPDF), conditional diffusion, conditional dissipation rates and conditional cross-dissipation rate. The diffusion velocity streamlines in scalar space representing the conditional diffusion generally converge quickly to a manifold along which they continue at a lower rate. A widely used mixing model, interaction through exchange with mean, does not exhibit such a trend. The approach to the manifold is generally in the direction of the ethylene mass fraction. The difference in the magnitudes of the diffusion velocity components for the two scalars cannot be accounted for by the difference in their dissipation time scales. The mixing processes during the approach to the manifold, therefore, cannot be modelled by using different dissipation time scales alone. While the three scalars in this flow have similar distances in scalar space, mixing between two of the scalars can occur only through the third, forcing a detour of the manifold (mixing path) in scalar space. This mixing path presents a challenging test for mixing models since most mixing models use only scalar-space variables and do not take into account the spatial (physical-space) scalar structure. The scalar JPDF and the conditional dissipation rates obtained in the present study have similarities to those of mixture fraction and temperature in turbulent flames. The results in the present study provide a basis for understanding and modelling multiscalar mixing in reactive flows.

We present an experimental investigation of the stability of a baroclinic front in a rotating two-layer salt-stratified fluid. A front is generated by the spin-up of a differentially rotating lid at the fluid surface. In the parameter space set by rotational Froude number, , dissipation number, (i.e. the ratio between disk rotation time and Ekman spin-down time) and flow Rossby number, a new instability is observed that occurs for Burger numbers larger than the critical Burger number for baroclinic instability. This instability has a much smaller wavelength than the baroclinic instability, and saturates at a relatively small amplitude. The experimental results for the instability regime and the phase speed show overall a reasonable agreement with the numerical results of Gula, Zeitlin & Plougonven (J. Fluid Mech., vol. 638, 2009, pp. 27–47), suggesting that this instability is the Rossby–Kelvin instability that is due to the resonance between Rossby and Kelvin waves. Comparison with the results of Williams, Haines & Read (J. Fluid Mech., vol. 528, 2005, pp. 1–22) and Hart (Geophys. Fluid Dyn., vol. 3, 1972, pp. 181–209) for immiscible fluid layers in a small experimental configuration shows continuity in stability regimes in space, but the baroclinic instability occurs at a higher Burger number than predicted according to linear theory. Small-scale perturbations are observed in almost all regimes, either locally or globally. Their non-zero phase speed with respect to the mean flow, cusped-shaped appearance in the density field and the high values of the Richardson number for the observed wavelengths suggest that these perturbations are in many cases due to Hölmböe instability.