Skip to content

Reducing the model simulations -- proxy reconstructions discrepancies on the volcanic cooling using reduced complexity models.

Magali Verkerk 1, Thomas J. Aubry1, James Salter1, Camilla Mathison2,3, Chris Smith4,5, Peter O. Hopcroft6, Michael Sigl7.

  • Affiliations: 1Department of Earth and Environmental Sciences, University of Exeter, Penryn, UK; 2Met Office Hadley Centre, Exeter, UK; 3University of Leeds, Leeds, UK; 4Department of Water and Climate, Vrije Universiteit Brussel, Brussels, Belgium; 5International Institute for Applied Systems Analysis, Laxenburg, Austria; 6School of Geography, Earth & Environmental Sciences, University of Birmingham, Birmingham, UK; 7Climate and Environmental Physics & Oeschger Centre for Climate Change Research, University of Bern, Bern, Switzerland

  • Presentation type: Talk

  • Presentation time: Monday 09:15 - 09:30, Room R290

  • Programme No: 6.1.4

  • Theme 6 > Session 1

Abstract

Volcanic eruptions are an important factor of climate variability at annual to multidecadal timescales, but the induced cooling is stronger in climate model simulations than in proxy-based reconstructions. This discrepancy may result from uncertainties in the eruption parameters, the aerosol-climate modelling and natural variability, but the computational cost of complex Earth system models prevents their quantification. Here, we combine a simple aerosol model (EVA_H) and a simple climate model (FaIR). Using ice-core and geological records to constrain SO2 injection parameters, we generate an ensemble of 1000 simulations for 6755 BCE--1900 CE in which SO2 injection and model parameters are resampled within their uncertainties. Our simulations reveal that the mean volcanic cooling over the last 9000 years is -0.12 ± 0.04 K, with a maximum yearly cooling of -2.0 ± 0.5 K in 5229 BCE. Critically, our simulations are in excellent agreement with tree ring-based reconstructions of the Northern Hemisphere summer temperature for the 20 largest eruptions of the period 750--1900 CE. This highlights the excellent skill of our modelling approach for estimating the radiative forcing and temperature response to volcanic eruptions. We also explore the suitability of data-driven emulators to simulate the spatial response to volcanic eruption. We adapt several methods of spatial emulation (pattern scaling and machine learning methods) to the climate response (local temperature and precipitation) to volcanic eruptions. The simplicity and versatility of our approach make it accessible to non-expert climate modelers, and ideal to propagate the numerous uncertainties affecting modelling of past eruption impacts.