Seafloor Bumps Bend Ocean Currents in Ways Climate Models Miss

Deep along the continental slopes of the Arctic Ocean and the Nordic Seas, narrow boundary currents carry warm, saline Atlantic Water around the basin rims, shaping sea ice, water mass transformation, and the freshwater exported back toward the North Atlantic. For decades, oceanographers have assumed that these slope-following currents respond to changing winds in a broadly linear way: push harder and the flow strengthens, ease off and it weakens. A new study published in Ocean Science by Anna Lina Petruseviciute Sjur of the University of Oslo and the Norwegian Meteorological Institute, together with Pål Erik Isachsen, Johan Nilsson, and Susan Elizabeth Allen, shows that this tidy picture breaks down in a striking and systematic way. The seafloor itself, dotted with ridges and corrugations, bends the response of the currents so that they end up flowing faster in one direction than the other, even under winds that average to zero.

The motivation came from earlier analyses of realistic, high-resolution simulations and observations. Sjur and colleagues had previously examined a five-year, eddy-permitting simulation of the Nordic Seas and Arctic Ocean and compared the circulation around closed isobaths with estimates from a simple linear theory first developed in 2003. That theory treats the flow as a low-pass-filtered, lagged response to surface wind stress, damped only by linear bottom friction. The correlation between the linear prediction and the simulated circulation was impressive, suggesting that much of the variability could be understood with linear dynamics. Yet two systematic deviations emerged: the simulated circulation was consistently shifted toward stronger cyclonic, counter-clockwise flow, and this shift was larger during anticyclonic forcing than during cyclonic forcing. A companion analysis of Argo float observations in the eddy-rich Lofoten Basin found cyclonic circulation nearly twice as strong as the linear model predicted from observed winds.

To untangle the mechanism, the team stripped the problem down to its essentials. They ran idealized shallow-water simulations of barotropic flow over a continental slope in a re-entrant channel, a periodic domain that mimics circulation around a closed basin. The slope was decorated with sinusoidal corrugations along its length, and the entire system was driven by an along-slope surface stress oscillating sinusoidally in time, with realistic amplitudes of about 0.1 newtons per square meter. Two forcing periods bracketed the relevant physics: a short period of 16 days, roughly comparable to synoptic weather variability, and a long period of 128 days, evoking seasonal-scale forcing. Crucially, the forcing had zero time mean, so any residual current that emerged had to be produced by the ocean’s own nonlinear dynamics.

The experiments reproduced the anomalies seen in the realistic Arctic simulations. Under long-period forcing, the circulation displayed a pronounced prograde-retrograde asymmetry: flow in the prograde direction, aligned with the intrinsic propagation of topographic Rossby waves, grew much stronger than flow in the opposite, retrograde direction. Most dramatically, the retrograde flow saturated. About 25 days into the retrograde phase, the current simply stopped strengthening no matter how hard the wind pushed, while transient eddies bloomed in the vorticity field at scales smaller than the topographic wavelength. Under short-period forcing, the asymmetry was muted but still present as a steady prograde offset, because the flow cannot fully adjust before the forcing reverses, a low-pass filtering effect governed by the ratio of the forcing period to the bottom friction damping timescale, which ranges from about 2 to 21 days across the modeled depths.

The key to the mechanism lies in how water columns interact with the bumpy bottom. As fluid columns are stirred back and forth across the slope, they tend to conserve their potential vorticity, gaining positive relative vorticity over deeper regions and losing it over shallower ones. In the presence of irreversibility, here represented by linear bottom drag, some of that vorticity is lost on each excursion, producing a net flux of potential vorticity down the topographic gradient, from shallow toward deeper water. This down-slope flux is strongest during retrograde phases and acts to accelerate prograde flow, yielding a systematic cyclonic bias around the basin. The team showed that when the diagnosed vorticity fluxes were added as a forcing term to the linear model, the discrepancy with the simulations almost entirely vanished, confirming that a single nonlinear mechanism explains the deviations.

A second, complementary perspective emerges when the momentum budget is evaluated along straight transects crossing the slope rather than along depth contours. In that framework, the nonlinear effects appear as a convergence of momentum flux together with topographic form stress, the net pressure force exerted by the flow on the bottom irregularities. Form stress requires a phase offset between the pressure field and the corrugations, and that offset is generated by nonlinear advection of relative vorticity. Critically, form stress is far stronger for retrograde flow, where the current can arrest topographic Rossby waves and build large pressure anomalies. When averaged over a full forcing cycle, this asymmetry leaves a residual prograde circulation, a result that echoes classic work from the 1980s on topographic rectification.

Perhaps the study’s most elegant contribution is showing that these two seemingly different viewpoints are, under quasi-geostrophic scaling, the same thing. Whether one integrates along isobaths, where vorticity fluxes appear, or along straight transects, where momentum flux convergence and form stress appear, the combined effect can be recast as a single flux of quasi-geostrophic potential vorticity across the slope. The thickness flux through the corrugations, a linear process, is essential: without along-slope depth variations, the thickness contribution vanishes, no net domain-integrated potential vorticity flux is possible, and oscillatory winds with zero mean cannot produce any residual transport. The team verified this directly with simulations over a smooth slope, which responded perfectly linearly with no residual flow, even when an explicit cross-slope wind component was imposed.

The saturation of retrograde flow turned out to be a resonance phenomenon. When retrograde velocities approach the arrest speed of topographic Rossby waves whose wavelength matches the corrugation wavelength, standing waves form and form stress increases sharply, halting further acceleration. The diagnosed saturation speeds scaled with wavelength exactly as arrested-wave theory predicts, falling between the phase speeds of the second and third wave modes, and the cross-slope pressure structure at the onset of saturation closely resembled a mode-2 arrested wave. Varying the corrugation amplitude revealed stepwise transitions: as the bumps shrank, the saturation velocity jumped from a mode-2 arrest speed to a mode-1 arrest speed before the nonlinearity disappeared altogether, showing that the height of the topography determines which wave mode can sustain the saturation against the applied stress.

The implications for climate modeling are sobering. The nonlinear flow-topography interactions identified here require the bathymetry to be well resolved, and they are almost certainly under-represented in the coarse-resolution ocean models used for climate projections. This matters particularly for the Arctic, where CMIP6 ensemble analyses indicate that surface stress is expected to intensify as sea ice thins, giving the nonlinear mechanisms a larger dynamical imprint. A future blue Arctic with reduced ice cover is also projected to see surging eddy activity, potentially amplifying eddy-topography interactions further. The authors note that stratification, absent from their idealized model, would likely concentrate these interactions near the seabed, and that additional eddy sources such as baroclinic instability could reinforce the cyclonic bias they identify.

What emerges is a unified and rather beautiful picture of how the ocean’s rough floor quietly edits the currents above it. Winds that blow back and forth with no net push still leave a mark, because the bumps on the seafloor extract momentum more efficiently from currents trying to travel against the grain of the topography than from those traveling with it. The result is a persistent cyclonic tendency around Arctic basins and a ceiling on how fast retrograde flow can become, set by the arrest of Rossby waves resonating with the ridges beneath. Capturing that asymmetry, the authors argue, will require ocean models fine enough to resolve the topographic spectrum, or new parameterizations that carry its nonlinear signature into the coarse grids on which our climate projections depend.

Subject of Research: Nonlinear interaction between wind-driven slope circulation and bottom topography in the Arctic Ocean

Article Title: Nonlinear dynamics of time-variable slope circulation

Article References: Sjur, A. L. P., Isachsen, P. E., Nilsson, J., & Allen, S. E. (2026). Nonlinear dynamics of time-variable slope circulation. Ocean Science, 22(5), 3055-3078. https://doi.org/10.5194/os-22-3055-2026

Image Credits: AI Generated

DOI: 10.5194/os-22-3055-2026

Keywords: ocean circulation, continental slope, topographic form stress, potential vorticity, topographic Rossby waves, Arctic Ocean, Nordic Seas, shallow-water model, wind forcing, bottom friction, ocean modeling, geophysical fluid dynamics