Fig. 5: Holocene changes in wind regime contextualized.

A comparison between our Principal Component (PC)-based Easterly and Westerly wind reconstructions and relevant reconstructions of storminess and paleoclimate data located in Fig. 1a, b using matching colors. a Iceland storm events by Ref. ⁶⁵. b PC-1 derived Storm Magnitude Index (SMI) values for polar Easterly winds, shown as scaled circles, Lake Steinbruvatnet. c Principal Component 1 (PC 1)-derived polar Easterlies reconstruction from Lake Steinbruvatnet, highlighting 30-year averages in bold, using a stippled line to mark the (μ + 1σ) cut-off for extremes. d The number of sand-sized particles (larger than 255 µm) from Lake Vårfluesjøen in Northern Spitsbergen³³. e Sum of cold periods, based on a global set of mostly summer-biased Holocene temperature timeseries⁶⁹. f the stacked chronology of Holocene Storm Periods (HSPs) from the North Atlantic, built on core data from the Seine Estuary and Mont-Saint-Michel Bay in northwestern France⁶⁶. g PC-2 derived SMI values for Westerly winds shown as scaled circles, Lake Steinbruvatnet. h Principal Component 2 (PC 2)-derived Westerly storm track reconstruction from Lake Steinbruvatnet, highlighting 30-year averages in bold, using a stippled line to mark the (μ + 1σ) cut-off for extremes. i P_BIP₂₅-derived sea ice cover data from the Fram Strait (gray), and j from the northern Barents Sea (black)^70,73,93. k A \({U}_{37}^{K}\)-based sea surface temperature (SST) reconstruction from the Barents Sea margin^71,94. The PCA and SMI source data for the data shown in this figure are provided as a Source Data file tab.