Search

High-latitude explosive volcanic eruptions can cause substantial hemispheric cooling. Here, we use a whole-atmosphere chemistry-climate model to simulate Northern Hemisphere (NH) high-latitude volcanic eruptions of magnitude similar to the 1991 Mt. Pinatubo eruption. Our simulations reveal that the initial stability of the polar vortex strongly influences sulphur dioxide lifetime and aerosol growth by controlling the dispersion of injected gases after such eruptions in winter. Consequently, atmospheric variability introduces a spread in the cumulative aerosol radiative forcing of more than 20%. We test the aerosol evolution's sensitivity to co-injection of sulphur and halogens, injection season, and altitude, and show how aerosol processes impact radiative forcing. Several of these sensitivities are of similar magnitude to the variability stemming from initial conditions, highlighting the significant influence of atmospheric variability. We compare the modelled volcanic sulphate deposition over the Greenland ice sheet with the relationship assumed in reconstructions of past NH eruptions. Our analysis yields an estimate of the Greenland transfer function for NH extratropical eruptions that, when applied to ice core data, produces volcanic stratospheric sulphur injections from NH extratropical eruptions 23% smaller than in currently used volcanic forcing reconstructions. Furthermore, the transfer function's uncertainty, which propagates into the estimate of sulphur release, needs to be at least doubled to account for atmospheric variability and unknown eruption parameters. Our results offer insights into the processes shaping the climatic impacts of NH high-latitude eruptions and highlight the need for more accurate representation of these events in volcanic forcing reconstructions. [ABSTRACT FROM AUTHOR]

The climate of the Northern Hemisphere (NH) in the mid-6th century was one of the coldest during the last 2 millennia based on multiple paleo-proxies. While the onset of this cold period can be clearly connected to the volcanic eruptions in 536 and 540 Common Era (CE), the duration, extent, and magnitude of the cold period are uncertain. Proxy data are sparse for the first millennium, which compounds the uncertainties of the reconstructions. To better understand the mechanisms of the prolonged cooling, we analyze new transient simulations over the Common Era and enhance the representation of mid-6th to 7th century climate by additional ensemble simulations covering 520–680 CE. We use the Max Planck Institute Earth System Model to apply the external forcing as recommended in the Paleoclimate Modelling Intercomparison Project phase 4. After the four large eruptions in 536, 540, 574, and 626 CE, a significant mean surface climate response in the NH lasting up to 20 years is simulated. The 2 m air temperature shows a cooling over the Arctic in winter, corresponding to the increase in Arctic sea ice, mainly in the Labrador Sea and to the east of Greenland. The increase in sea-ice extent relates to a decrease in the northward ocean heat transport into the Arctic within the first 2 years after the eruptions and to an increase in the Atlantic meridional overturning circulation, which peaks 10 years after the eruptions. A decrease in the global ocean heat content is simulated after the eruptions that does not recover during the simulation period. These ocean–sea-ice interactions sustain the surface cooling, as the cooling lasts longer than is expected solely from the direct effects of the volcanic forcing, and are thus responsible for the multi-decadal surface cooling. In boreal summer, the main cooling occurs over the continents at midlatitudes. A dipole pattern develops with high sea level pressure and a decrease in both precipitation and evaporation poleward of 40 ∘ N. In addition, more pronounced cooling over land compared to ocean leads to an enhanced land–sea contrast. While our model ensemble simulations show a similar ∼20 -year summer cooling over NH land after the eruptions as tree ring reconstructions, a volcanic-induced century-long cooling, as reconstructed from tree ring data, does not occur in our simulations. [ABSTRACT FROM AUTHOR]