Fig. 4: Atmospheric and oceanic heat transport compete as drivers for marine heatwaves and marine cold spells in the southeastern Barents Sea.

Panel a, b show heat fluxes calculated following the methodology by Carton et al.¹⁷; Each variable is calculated for the southeastern Barents Sea and averaged for each year’s summer (Jul–Oct) and additionally smoothed using a 5-year rolling mean. The colored dashed lines are half a standard deviation for each individual time series, and the colored areas indicate exceedance of that threshold, either upward or downward. a Anomalies of the time derivative of ocean heat content (HC_t; solid line) and net surface air-sea heat fluxes (Q_net; dashed line). b Anomalies of the residual of the time derivative of the ocean heat content (HC_t) and the surface heat fluxes (Q_net), representing the ocean’s local heat convergence. In gray, the marine heatwave coverage (solid) and marine cold spell coverage (dashed). c Shown are normalized 3-year rolling means of: (blue, dashed) the residual heat flux seen in b; (black, solid) the joint coverage of marine heatwaves and marine cold spells, the sum of the two gray lines in b, relative to the linearly shifting baseline (Fig. 2b); (green, dash-dotted) the NAO index; (orange, dotted) the east-west pressure gradient ΔP between the Barents Sea and the Nordic Seas, positive values indicate continental and Atlantic air, and negative values Arctic air inflows (see Fig. 5a, d); and (black, dash-plus) the sum of the east-west pressure gradient and the residual heat flux. The correlations to the joint marine extreme coverage metric are: r = 0.48 (p < 0.01) for the residual heat flux; r = 0.32 for the NAO (p < 0.05); r = 0.54 (p < 0.01) for the pressure gradient; and r = 0.66 (p < 0.01) for the combined effect of the east-west pressure gradient and the residual heat flux.