We propose a novel multiple-scale spatial marching method for flows with slow streamwise variation. The key idea is to couple the boundary region equations, which govern large-scale flow evolution, with local exact coherent structures that capture the small-scale dynamics. This framework is consistent with high-Reynolds-number asymptotic theory and offers a promising approach to constructing time-periodic finite-amplitude solutions in a broad class of spatially developing shear flows. As a first application, we consider a non-uniformly curved channel flow, assuming that a finite-amplitude travelling-wave solution of plane Poiseuille flow is sustained at the inlet. The method allows for the estimation of momentum transport and highlights the impact of the inlet condition on both the transport properties and the overall flow structure. We then consider a case with gradually decreasing curvature, starting with Dean vortices at the inlet. In this setting, small external oscillatory disturbances can give rise to subcritical self-sustained states that persist even after the curvature vanishes.

For a perturbed trefoil vortex knot evolving under the Navier–Stokes equations, a sequence of $\nu$-independent times $t_m$ are identified that correspond to a set of scaled, volume-integrated vorticity moments $\nu ^{1/4}\mathcal{O}_{\textit{Vm}}$, with this hierarchy $t_\infty \leqslant \ldots \leqslant t_m\ldots t_1=t_x\approx 40$ and $\mathcal{O}_{\textit{Vm}}=(\int _{V\ell }|\omega |^{2m}\,{\rm d}V)^{1/2m}$. For the volume-integrated enstrophy $Z(t)$, convergence of $\sqrt {\nu }Z(t)=\bigl (\nu ^{1/4}\mathcal{O}_{\textit{V}\text{1}}(t)\bigr )^2$ at $t_x=t_1$ marks the end of reconnection scaling. Physically, reconnection follows from the formation of a double vortex sheet, then a knot, which splits into spirals. Meanwhile $Z$ accelerates, leading to approximate finite-time $\nu$-independent convergence of the energy dissipation rate $\epsilon (t)=\nu Z(t)$ at $t_\epsilon \sim 2t_x$. This is sustained over a finite temporal span of at least $\Delta T_\epsilon \searrow 0.5 t_\epsilon$, giving Reynolds number independent finite-time, temporally integrated dissipation, $\Delta E_\epsilon =\int _{\Delta T_\epsilon }\epsilon \,{\rm d}t$, and thus satisfies one definition for a dissipation anomaly, with enstrophy spectra that are consistent with transient $k^{1/3}$ Lundgren-like inertial scaling over some of the $\Delta T_\epsilon$ time. A critical factor in achieving these temporal convergences is how the computational domain $V_\ell =(2\ell \pi )^3$ is increased as $\ell \sim \nu ^{-1/4}$, for $\ell =2$ to 6, then to $\ell =12$, as $\nu$ decreases. Appendix A shows compatibility with established $(2\pi )^3$ mathematics where $\nu \equiv 0$ Euler solutions bound small $\nu$ Navier–Stokes solutions. Two spans of $\nu$ are considered. Over the first factor of 25 decrease in $\nu$, most of the $\nu ^{1/4}\mathcal{O}_{\textit{Vm}}(t)$ converge to their respective $t_m$. For the next factor of 5 decrease (125 total) in $\nu$, with increased $\ell$ to $\ell =12$, there is initially only convergence of $\nu ^{1/4}\varOmega _{V\infty }(t)$ to $t_\infty$, without convergence for $9\gt m\gt 1$. Nonetheless, there is later $\sqrt {\nu }Z(t)$ convergence at $t_1=t_x$ and $\epsilon (t)=\nu Z$ over $t\sim t_\epsilon \approx 2t_x$.

Whilst surface-stress integration remains the standard approach for fluid force evaluation, control-volume integral methods provide deeper physical insights through functional relationships between the flow field and the resultant force. In this work, by introducing a second-order tensor weight function into the Navier–Stokes equations, we develop a novel weighted-integral framework that offers greater flexibility and enhanced capability for fluid force diagnostics in incompressible flows. Firstly, in addition to the total force and moment, the weighted integral methods establish, for the first time, rigorous quantitative connections between the surface-stress distribution and the flow field, providing potential advantages for flexible body analyses. Secondly, the weighted integral methods offer alternative perspectives on force mechanisms, through vorticity dynamics or pressure view, when the weight function is set as divergence-free or curl-free, respectively. Thirdly, the derivative moment transformation (DMT)-based integral methods (Wu et al., J. Fluid Mech. vol. 576, 2007, 265–286) are generalised to weighted formulations, by which the interconnections among the three DMT methods are clarified. In the canonical problem of uniform flow past a circular cylinder, weighted integral methods demonstrate advantages in yielding new force expressions, improving numerical accuracy over original DMT methods, and enhancing surface-stress analysis. Finally, a force expression is derived that relies solely on velocity and acceleration at discrete points, without spatial derivatives, offering significant value for experimental force estimation. This weighted integral framework holds significant promise for flow diagnostics in fundamentals and applications.

We present a back-in-time analysis for the origin of vorticity in viscous separated flows over immersed bodies, using the adjoint-vorticity framework recently introduced by Xiang et al. (2025 J. Fluid Mech. vol. 1011, A33. The solution of the adjoint-vorticity equations yields the volume density of mean deformation, which captures the stretching and tilting of the earlier vorticity that leads to the terminal value. The analysis also takes into account the boundary contributions of vorticity and its flux. Three examples are considered. Steady, axisymmetric separation in the flow over a sphere at Reynolds number $Re=200$ is shown to be established due to wall flux from both upstream and downstream of separation, the latter contribution being absent from the classical description by Lighthill. For unsteady separation at higher $Re=300$, the streamwise vorticity within the wake hairpin vortex is traced back, quantitatively, to the azimuthal vorticity on the sphere. The third configuration is the flow over a prolate spheroid at $Re=3000$. The null vorticity at three-dimensional separation originates from the cancellation of opposite interior contributions adjacent to the separation surface. The contribution from the downstream side migrates across the separation surface into the upstream region due to a tilting effect – a fundamental distinction between two- and three-dimensional separation. We also examine the detached vortical structures. The streamwise vorticity in the primary vortex originates from tilting of near-wall azimuthal vorticity, differing from Lighthill’s conjecture that the origin is streamwise near-wall vorticity that arises due to the reduced Coriolis force. Finally, a necklace vortex in the turbulent wake is traced back in time, and is shown to have contributions from the spheroid trailing-edge shed shear layer and the large-scale counter-rotating primary vortices.

The interaction between a coherent vortex ring and an inertial particle is studied through a combination of experimental and numerical methods. The vortex ring is chosen as a model flow ubiquitous in various geophysical and industrial flows. A detailed description of the vortex properties together with the evolution of the particle kinematics during the interaction is addressed thanks to time-resolved particle image velocimetry and three-dimensional shadowgraphy visualisations. Complementary, direct numerical simulations are realised with a one-way coupling model for the particle, allowing for the identification of the elementary forces responsible for the interaction behaviours. The experimental and numerical results unequivocally demonstrate the existence of three distinct interaction regimes in the parameter range of the present study: simple deviation, strong deviation and capture. These regimes are delineated as functions of key controlled dimensionless parameters, namely, the Stokes number and the initial radial position of the particle relative to the vortex ring axis of propagation.

A large-scale parametric study of the flow over the prolate spheroid is presented to understand the effect of Reynolds number and angle of attack on the separation, the wake formation and the loads. Large-eddy simulation is performed for six Reynolds numbers ranging from ${\textit{Re}} = 0.15\times 10^6$ to $4 \times 10^6$ and for eight angles of attack ranging from $\alpha = 10^\circ$ to $\alpha = 90^\circ$. For all the cases considered, the boundary layer separates symmetrically and forms a recirculation region. Several distinct flow topologies are observed that can be grouped into three categories: proto-vortex, coherent vortex and recirculating wake. In the proto-vortex state, the recirculation does not have a distinct centre of rotation, instead, a two-layer detached flow structure is formed. In the coherent vortex state, the separated shear layer rolls into a three-dimensional vortex that is aligned with the axis of the spheroid. This vortex has a clear centre of rotation corresponding to a minimum of pressure and transforms the transverse momentum from the separated shear layer into axial momentum. In the recirculating wake regime, the recirculation is incoherent and the primary separation forms a dissipative shear layer that is convected in the direction of the free stream. This symmetric pair of shear layers bounds a low-momentum recirculating cavity on the leeward side of the spheroid. The properties of these states are not constant, but evolve along the axis of the spheroid and are dictated by the characteristics of the boundary layer at separation. The variation of the flow with Reynolds number and angle of attack is described, and its connection to the loads on the spheroid are discussed.

Single-cell tornado-like vortices (TLVs) exhibit periodic wandering fluctuations around the time-averaged vortex core, a phenomenon known as vortex wandering, which constitutes the most prominent periodic behaviour in such flows. The coupling between vortex motion and wandering creates complex swirl dynamics, posing significant analytical challenges. However, the limited availability of experimental studies on vortex wandering decomposition hampers a deeper understanding of this phenomenon. To address this gap, a tornado simulator was designed to generate a controllable single-cell TLV, and high-frequency velocity data were obtained using particle image velocimetry. A sparsity-promoting dynamic mode decomposition (sp-DMD) method was developed to decouple coherent structures and analyse dynamic characteristics. Results show that as the swirl ratio increases, the vortex structure becomes more diffuse, with significant reductions in intensity. Vortex wandering is present across all swirl conditions, with its periodicity strongly modulated by the swirl ratio. Importantly, sp-DMD identified two primary modes, the time-averaged mode (first mode), representing the dominant rotational vortex motion, and the vortex-wandering-dominated modes (second and third conjugate modes), which correspond to persistent periodic velocity fluctuations and contribute the most significant pulsations. These modes exhibit a pair of oppositely rotating vortices symmetrically revolving around the central flow axis. Visualisations of the Q criterion reveal a symmetric dipole pattern. This suggests that rotational and shear effects are likely responsible for the periodic movement of the vortex core. Furthermore, as the swirl ratio increases, the energy of the vortex-wandering-dominated modes diminishes, and motion transitions from high-energy, organised dynamics to low-energy, disordered behaviour.

We investigate the effectiveness of linear optimal perturbation (LOP) for the flow past a finite span wing in reducing the lifespan of its trailing vortex system. Two approaches, referred to as local and model analysis, are introduced and used for our investigation. Both analyses assume that the baseflow is parallel. Local analysis is suited for intermediate distance from the wing where both tip vortices (TVs) and trailing edge wake (TEW) are present. Its results suggest that the unperturbed baseflow is stable. The separation between TVs and TEW increases downstream and their dynamics appear to be uncoupled at large distance from the wing. When perturbation corresponding to LOP is added to the baseflow, the vortices are displaced forming a helical twist. With time, the maximum displacement initially increases and then saturates. The perturbation retains its compact wavepacket-like structure, and perturbation energy within the tip vortex remains nearly constant. In the model analysis, the far wake is modelled as a pair of counter-rotating $q$-vortices. For low Reynolds number, the flow is stable. However, for higher Reynolds number, the trailing vortices develop Crow instability. Its growth rate is found to be in good agreement with earlier studies. Instability leads to contact of vortices, resulting in the formation of vortex rings. The time for vortex contact decreases with increase in the strength of the initial perturbation. The results suggest that LOP is effective in reducing the lifespan of trailing vortices.

A backward swept shape is one of the common features of the wings and fins in animals, which is argued to contribute to leading-edge vortex (LEV) attachment. Early research on delta wings proved that swept edges could enhance the axial flow inside the vortex. However, adopting this explanation to bio-inspired flapping wings and fins yields controversial conclusions, in that whether and how enhanced spanwise flow intensifies the vorticity convection and vortex stretching is still unclear. Here, the flapping wings and fins are simplified into revolving plates with their outboard 50 $\%$ span swept backward in either linear or nonlinear profiles. The local spanwise flow is found to be enhanced by these swept designs and further leads to stronger vorticity convection and vortex stretching, thus contributing to local LEV attachment and postponing bursting. These results further prove that a spanwise gradient of incident velocity is sufficient to trigger a regulation of LEV intensity, and a concomitant gradient of incident angle is not necessary. Moreover, an attached trailing-edge vortex is generated on a swept wing and induces an additional low-pressure region on the dorsal surface. The lift generation of swept wings is inferior to that of the rectangular wing because the extended stable LEV along the span and the additional suction force near the trailing edge are not comparable to the lift loss due to the reduced LEV intensity. Our findings evidence that a swept wing can enhance the spanwise flow and vorticity transport, as well as limit excessive LEV growth.

We investigate Lighthill’s proposed turbulent mechanism for near-wall concentration of spanwise vorticity by calculating mean flows conditioned on motion away from or toward the wall in an (friction Reynolds number) ${\textit{Re}}_\tau =1000$ database of plane-parallel channel flow. Our results corroborate Lighthill’s proposal throughout the entire logarithmic layer, but extended by counter-flows that help explain anti-correlation of vorticity transport by advection and by stretching/tilting. We present evidence also for Lighthill’s hypothesis that the vorticity transport in the log layer is a ‘cascade process’ through a scale hierarchy of eddies, with intense competition between transport outward from and inward to the wall. Townsend’s model of attached eddies of hairpin-vortex type accounts for half of the vorticity cascade, whereas we identify necklace type or ’shawl vortices’ that envelop turbulent sweeps as supplying the other half.

High-resolution particle image velocimetry (PIV) particle-to-velocity analyses using small interrogation areas (IAs) often require substantial processing time. To overcome this limitation, a generative adversarial network (GAN)-based model is proposed to achieve spatio-temporal super-resolution (SR) reconstruction from low-resolution PIV data with large IAs, thereby significantly reducing post-processing time. Time-resolved PIV measurements of plasma-induced vortex flows, covering vortex formation, growth, transition and breakdown stages, are employed to train and evaluate the model with multi-scale vortical structures. By sequentially constructing spatial and temporal datasets, the GAN-based model enables reliable SR reconstruction at different scaling factors. Reconstruction accuracy is systematically assessed using time-averaged, instantaneous and phase-averaged velocity fields. At SR factors of $\times$4 and $\times$8, the reconstructed fields closely match high-resolution references, effectively capturing both fluctuating velocities and small-scale vortical structures. In contrast, $\times$16 reconstructions exhibit diminished accuracy due to the loss of fine-scale information from highly downsampled inputs. For time-averaged fields, high-resolution reconstructions reliably capture plasma jet characteristics at all SR factors. To enhance generalisation, transfer learning is introduced to fine tune the parameters of SR-related layers in the generator, enabling accurate reconstructions under varying vortex dynamics. In addition, the efficiency gains in PIV particle-to-velocity analysis and the fundamental limitations on achievable SR factors imposed by spatio-temporal data correlations are discussed. This study demonstrates that GAN-based spatio-temporal SR models offer a promising approach to accelerate PIV analyses while maintaining high reconstruction fidelity with diverse flow conditions.

Amphibious unmanned vehicles promise next-generation water-based missions by eliminating the need for multiple vehicles to traverse water and air separately. Existing research-grade quadrotors can navigate in water and air and cross the water–air boundary, but it remains unclear how their transition is affected by rotor kinematics and geometry. We present here experimental results from isolated small rotors (diameters $\sim 10\,\mathrm{cm}$) dynamically transitioning from water to air. We discovered that rotors experience an abrupt change in frequency, lift and torque before reaching the interface, and the change is linked to the surface depression caused by a free surface vortex. We explored how the surface dynamics are affected by advance ratio, rotor diameter, number of rotor blades and input throttle. Free surface vortices above rotating objects have been studied in the context of unbaffled stirred tanks, but not in the field of small amphibious rotorcraft. We show that existing free surface vortex models can be adapted to explain water-to-air rotor performance. A better understanding of water–air rotor transitions helps to (i) assess the amphibious capability of existing aerial rotors, and (ii) suggest efficient water–air transition strategies for next-generation amphibious vehicles.

Quantum turbulence is characterised by the collective motion of mutually interplaying thin and discrete vortex filaments of fixed circulation which move in two mutually interacting fluid components. Despite this very peculiar nature determined by quantum-mechanical effects, turbulence in quantum fluids may exhibit very similar features to classical turbulence in terms of the vortex dynamics, energy spectrum and decay and intermittency. The recent work by Blaha et al. (2025 J. Fluid. Mech. 1015, A57) reveals an additional classical behaviour of quantum turbulence, by showing that the trajectories of starting vortices shed by accelerating airfoils in a quantum fluid are almost indistinguishable from their counterpart in classical viscous flows. These results strongly support the suggestive idea that turbulent flows, both classical and quantum, may be described by the collective dynamics of interacting, thin and discrete filaments of fixed circulation.

Vortex–magnetic interactions shape magnetohydrodynamic (MHD) turbulence, influencing energy transfer in astrophysical, geophysical and industrial systems. In the solar atmosphere, granular-scale vortex flows couple strongly with magnetic fields, channelling energy into the corona. At high Reynolds numbers, vorticity and magnetic fields are nearly frozen into the charged fluid, and MHD flows emerge from the Lorentz force mediated interactions between coherent vortex structures in matter and the field. To probe this competition in a controlled setting, we revisit the canonical problem of two antiparallel flux tubes. By varying the magnetic flux threading each tube – and thus sweeping the interaction parameter $N_i$, which gauges Lorentz-to-inertial force balance – we uncover three distinct regimes: vortex-dominated joint reconnection, instability-triggered cascade, and Lorentz-induced vortex disruption. At low $N_i$, classical vortex dynamics dominates, driving joint vortex–magnetic reconnection, and amplifying magnetic energy via a dynamo effect. At moderate $N_i$, the system oscillates between vorticity-driven attraction and magnetic damping, triggering instabilities and nonlinear interactions that spawn secondary filaments and drive an energy cascade. At high $N_i$, Lorentz forces suppress vortex interactions, aligning the tubes axially while disrupting vortex cores and rapidly converting magnetic to kinetic energy. These findings reveal how the inertial–Lorentz balance governs energy transfer and coherent structure formation in MHD turbulence, offering insight into vortex–magnetic co-evolution in astrophysical plasmas.

The paper discusses the stochastic dynamics of the vortex shedding process in the presence of external harmonic excitation and coloured multiplicative noise. The situation is encountered in a turbulent practical combustor experiencing combustion instability. Acoustic feedback and turbulent flow are imitated by the harmonic and stochastic excitations, respectively. The Ornstein–Uhlenbeck process is used to generate the noise. A low-order model for vortex shedding is used. The Fokker–Planck framework is used to obtain the evolution of the probability density function of the shedding time period. Stochastic lock-in and resonance characteristics are studied for various parameters associated with the harmonic (amplitude, frequency) and noise (amplitude, correlation time, multiplicative noise factor) excitations. We observed that: (i) the stochastic lock-in (s-lock-in) boundary strongly depends on the noise correlation time; (ii) the parameter sites for s-lock-in can be approximately identified from the noise-induced shedding statistics; and (iii) stochastic resonance is significant for some intermediate correlation times. The effects of the above-mentioned observations are discussed in the context of combustion instability.

Large numbers of relative periodic orbits (RPOs) have been found recently in doubly periodic, two-dimensional Kolmogorov flow at moderate Reynolds numbers ${\textit{Re}} \in \{40, 100\}$. While these solutions lead to robust statistical reconstructions at the ${\textit{Re}}$ values where they were obtained, it is unclear how their dynamical importance changes with ${\textit{Re}}$. Arclength continuation on this library of solutions reveals that large numbers of RPOs quickly become dynamically irrelevant, reaching dissipation values either much larger or smaller than the values typical of the turbulent attractor at high ${\textit{Re}}$. The scaling of the high-dissipation RPOs is shown to be consistent with a direct connection to solutions of the unforced Euler equation, and is observed for a wide variety of states beyond the ‘unimodal’ solutions considered in previous work (Kim & Okamoto, Nonlinearity vol. 28, 2015, p. 3219). However, the weakly dissipative states have properties indicating a connection to exact solutions of a forced Euler equation. The dynamical irrelevance of many solutions leads to poor statistical reconstruction at higher ${\textit{Re}}$, raising serious questions for the future use of RPOs for estimating probability densities. Motivated by the Euler connection of some of our RPOs, we also show that many of these states can be well described by exact relative periodic solutions in a system of point vortices. The point vortex RPOs are converged via gradient-based optimisation of a scalar loss function which (i) matches the dynamics of the point vortices to the turbulent vortex cores and (ii) insists the point vortex evolution is itself time-periodic.

We investigate interactions between two like-signed vortices over either an isolated seamount or a basin (a depression in the bathymetry), using a quasi-geostrophic, two-layer model on the $f$-plane. When the vortex pair is centred over the seamount, the vortices are pushed together by the secondary flow generated in the bottom layer, facilitating their merger. Over a basin, the deep anomalies are much stronger and their interaction strains out the surface vortices. The results are supported by an analytical estimation of the initial potential vorticity anomalies in the lower layer and by analysis of the linear stability of a single vortex over the bathymetry. Similar phenomena are observed when the vortex pair is displaced from the bathymetric centre and when the initial vortices are initially compensated. Sub-deformation-scale vortices are less influenced by bathymetry than larger vortices. The results help explain asymmetries noted previously in turbulence simulations over bathymetry.

Time-varying flow-induced forces on bodies immersed in fluid flows play a key role across a range of natural and engineered systems, from biological locomotion to propulsion and energy-harvesting devices. These transient forces often arise from complex, dynamic vortex interactions and can either enhance or degrade system performance. However, establishing a clear causal link between vortex structures and force transients remains challenging, especially in high-Reynolds-number nominally three-dimensional flows. In this study, we investigate the unsteady lift generation on a rotor blade that is impulsively started with a span-based Reynolds number of 25 500. The lift history from this direct-numerical simulation reveals distinct early-time extrema associated with rapidly evolving flow structures, including the formation, evolution and breakdown of leading-edge and tip vortices. To quantify the influence of these vortical structures on the lift transients, we apply the force partitioning method (FPM) that quantifies the surface pressure forces induced by vortex-associated effects. Two metrics – $Q$-strength and vortex proximity – are derived from FPM to provide a quantitative assessment of the influence of vortices on the lift force. This analysis confirms and extends qualitative insights from prior studies, and offers a simple-to-apply data-enabled framework for attributing unsteady forces to specific flow features, with potential applications in the design and control of systems where unsteady aerodynamic forces play a central role.

This numerical investigation focuses on the mechanisms, flow topology and onset of Kelvin–Helmholtz instabilities (KHIs), that drive the leading-edge shear-layer destabilisation in the wake of wall-mounted long prisms. Large-eddy simulations are performed at ${\textit{Re}} = 2.5\times 10^3, 5\times 10^3$ and $1\times 10^4$ for prisms with a range of aspect ratio (AR, height-to-width) between $0.25$ and $1.5$, and depth ratios (DR, length-to-width) of $1{-}4$. Results show that shear-layer instabilities enhance flow irregularity and modulate spanwise vortex structures. The onset of KHI is strongly influenced by depth ratio, such that long prisms (${\textit{DR}}= 4$) experience earlier initiation compared with shorter ones (${\textit{DR}}= 1$). At higher Reynolds numbers, the onset of KHI shifts upstream towards the leading-edge, intensifying turbulence kinetic energy and increasing flow irregularity, especially for long prisms. The results further show that in this configuration, energy transfer from the secondary recirculation region contributes to the destabilisation of the leading-edge shear layer by reinforcing low-frequency modes. A feedback mechanism is identified wherein energetic flow structures propagate upstream through reverse boundary-layer flow, re-energising the leading-edge shear layer. Quantification using probability density functions reveals rare, intense upstream energy convection events, driven by this feedback mechanism. These facilitate the destabilisation process regardless of Reynolds number. This study provides a comprehensive understanding of the destabilisation mechanisms for leading-edge shear layers in the wake of wall-mounted long prisms.

This paper provides direct experimental evidence for the coexistence of both a laminar separation bubble and a secondary vortex on the advancing side of a rotating sphere when subjected to the inverse Magnus effect. Detailed experiments were conducted in a wind tunnel on two spheres of varying surface roughness to investigate both ordinary and inverse Magnus effects. Experiments took place for $0.5\times 10^{5}\leqslant {\textit{Re}}\leqslant 3\times 10^{5}$ and rotation rates $0\leqslant \alpha \leqslant 0.45$, where the spheres were rotated via a shaft that was oriented perpendicularly to the free stream flow. Static pressure measurements were made on the non-shaft hemisphere using a spline of taps spanning from the equator to the pole. The ordinary Magnus effect was generally observed at the lowest ${\textit{Re}}$ tested, with a transition to the inverse Magnus effect occurring as ${\textit{Re}}$ increased. Time-averaged pressure coefficient distributions across the equatorial plane were obtained for the smooth and rough spheres. Cross-flow particle image velocimetry was used to visualise the downstream wake velocity field. A pair of counter-rotating wing-tip-like vortices were detected when the sphere experienced the ordinary Magnus effect, generated by flow leakage from the advancing to the retreating side. When the sphere experienced the inverse Magnus effect, the polarity of the counter-rotating vortex pair reversed. This is the first experimental observation of the vortex polarity reversal associated with the inverse Magnus effect in the wake of a rotating sphere. The results provide qualitative visualisation of the complex fluid dynamics and inform future applications of the Magnus effect.