Barents Sea atlantification driven by a shift in atmospheric synoptic timescale

nature climate change Article https://doi.org/10.1038/s41558-025-02535-3 Barents Sea atlantification driven by a shift in atmospheric synoptic timescale Received: 17 February 2025 Accepted: 3 December 2025 Published online: 2 January 2026 Check for updates Robinson Hordoir 1,2 , Vahidreza Jahanmard 3, Pål Erik Isachsen 4,5, Ulrike Löptien 6,7, Heiner Dietze 6,7,8, Anne Britt Sandø1,2 & Vidar S. Lien 1,2 Climate change impinges on the Arctic Ocean, leading to sea-ice loss and potentially drastic cascading ecosystem changes. A recent process is atlantification, the growing influence of warm and salty waters from the Atlantic on the Arctic with increasing ocean volume transport from the Nordic Seas to the Barents Sea playing a key role. Despite its importance and a multitude of hypotheses that have been tested, this trend remains mainly unexplained. Here we explore nonlinear effects and successfully link the flow trend through the Barents Sea Opening to a frequency shift of atmospheric synoptic. We show that a part of the flow through Barents Sea Opening is driven by topographic Rossby waves, and that they have a very sensitive response to atmospheric frequency over the Nordic Seas. These findings highlight how anthropogenic changes to the atmosphere are altering ocean processes, with implications for sea-ice extent and ecosystems in the Arctic. The Arctic Ocean is one of the most vulnerable regions on the planet, changing at an unprecedented rate1. Attempts to advance understanding of the Arctic climate system have triggered comprehensive international observational efforts2 along with joint modelling experiments3–5. Among the processes of major concern in the Arctic is the so-called atlantification, which denotes an increasing influence of Atlantic waters on the Arctic Ocean6–10. Consequences are major, ranging from sea-ice decline11–13, with potentially far-reaching alterations of weather patterns through teleconnections14,15, to regional ecosystem changes9,16. Atlantic water (AW) driving the atlantification is transported towards the Arctic through two gateways: Fram Strait, which receives AW through the West Spitsbergen current (WSC) and the Barents Sea Opening (BSO) located between continental Norway and Bear Island. A number of studies attempt to link the variability of the two AW inflow branches to ocean dynamical responses to changing atmospheric forcing17,18 but failed to explain the positive trend in the volume transport of the BSO flow. Likewise, studies exploring the effects of trends in atmospheric patterns19–21 and effects of processes at other Arctic gateways22 have been inconclusive. Also, hypotheses that a decline in sea ice leads to changes in the atmosphere, which then influence the BSO flow, could not be confirmed in a model study23. Further, a recent study suggested a relationship between one of the leading atmospheric modes over the Arctic, the Arctic dipole and the BSO flow trend24. This relationship could, however, not be supported by idealized model experiments25. Here we provide a physical explanation for the BSO flow trend by using ocean modelling combined with deep learning (DL). We use a DL-based surrogate model25,26, which enables us to study complex atmosphere–ocean relationships in a controlled environment, alongside a regional ocean model25,27 to investigate the response of the flow towards the Arctic Ocean to changing atmospheric forcing. Our results establish a causal link between the BSO flow trend and a frequency shift in the atmospheric synoptic variability over the Nordic Seas, as a nonlinear indirect response through topographic waves and rectified currents. The nature of the flow trend at Barents Sea Opening Our regional ocean model27, simulates a clear positive trend of barotropic flow through the Norway-Svalbard section towards the Arctic (NorSva) for the period 1980–2021 (Fig. 1) in line with earlier findings28. Our model also simulates a trend (Fig. 1g) through the WSC section. Interestingly, this WSC trend disappears when the ocean model is Oceanography and Climate Research Group, Institute of Marine Research, Bergen, Norway. 2Bjerknes Centre for Climate Research, Bergen, Norway. Department of Civil Engineering and Architecture, Tallinn University of Technology, Tallinn, Estonia. 4Department of Geosciences, University of Oslo, Oslo, Norway. 5Norwegian Meteorological Institute, Oslo, Norway. 6Department of Computer Science, University of Kiel, Kiel, Germany. 7Institute of Geosciences, University of Kiel, Kiel, Germany. 8Department of Chemistry, King’s College London, London, UK. e-mail: 1 3 Nature Climate Change | Volume 16 | February 2026 | 179–186 179 Article https://doi.org/10.1038/s41558-025-02535-3 a b 4,000 80° N Svalbard Archipelago c 3,500 3,000 2,500 2,000 Bear Island Depth (m) Spitsbergen Bank 1,500 1,000 70° N 500 Norway 10° E 20° E 30° E d 0 –1 e NorSva flow 3.2 2.8 Flow (Sv) Flow (Sv) 0.02 m s BSO-north flow 1.6 1.4 3.0 2.6 2.4 1.2 1.0 0.8 2.2 0.6 2.0 1.8 –1 0.05 m s 1980 1985 1990 1995 2000 2005 2010 2015 2020 1980 1985 1990 1995 Years 2005 2010 2015 2020 2010 2015 2020 Years g BSO-south flow 1.8 WSC flow 8.0 1.7 7.0 1.6 6.0 Flow (Sv) Flow (Sv) f 2000 1.5 5.0 4.0 1.4 3.0 1.3 1980 1985 1990 1995 2000 2005 2010 2015 2020 Years 1980 1985 1990 1995 2000 2005 Years Fig. 1 | Barotropic currents and transports in the vicinity of BSO, schematic view and results of a regional numerical model. a, Sketch of the currents through the BSO-South section (red), through the BSO-North section (green) and across the WSC section (white). Colour shading denotes bathymetry. b, Simulated25,27 time interval (1980–2000) mean barotropic velocities. c, Simulated25,27 difference in mean barotropic currents between time intervals 2001–2021 and 1980–2000. The difference is computed by subtracting velocity fields between the two time intervals, interval 2001–2021 minus interval 1980–2000. d–g, Annual mean values of the NorSva (yellow section, as in ref. 25) (d), BSO-South (red section) (e), BSO-North (green section) (f) and WSC flows (white section) (g). The curves and linear regressions refer to the reference experiment25 (plain black), the reference experiment that uses an atmospheric forcing with linearly detrended SLP and wind fields (dotted blue) and to the reference experiment with detrended SLP and wind fields that uses climatological open boundary conditions for the barotropic mode (dasheddotted red). The reference experiment NorSva, BSO-North and WSC trends are 0.014 Sv, 0.012 Sv and 0.039 Sv per year, respectively. The BSO-South flow does not have a statistically significant trend. Flows heading east or north are counted as positive, whereas flows heading west or south are counted as negative. Basemaps in a–c from Natural Earth (https://www.naturalearthdata.com/). forced with detrended sea-level pressure (SLP) and winds, while t (...truncated)