Long-term volcanic signal in 21st-century climate projections with a 25-member stochastic ensemble using SOCOL-MPIOM

Future volcanic eruptions are largely absent from standard CMIP6 simulations, introducing a systematic source of uncertainty in 21st-century climate projections. Explosive tropical eruptions inject SO₂ into the stratosphere, forming sulphate aerosol layers that perturb the global energy budget, stratospheric chemistry, and ocean–sea ice dynamics on interannual-to-decadal timescales. Quantifying this volcanic signal probabilistically within a fully coupled chemistry–climate framework remains an open challenge.

We performed an 80-year (2020-2100) 25-member stochastic ensemble simulation using the coupled chemistry–climate model SOCOL-MPIOM under the SSP3-7.0 emission scenario. Each ensemble member incorporates five independently sampled tropical eruptions (three strong, one moderate, one weak), targeting the upper bound of plausible centennial volcanic activity. The volcanic signal is isolated by comparing the ensemble mean against a single volcanic-free baseline simulation under identical anthropogenic forcing. We examine global and regional temperature, ozone, precipitation, Arctic sea ice, and ETCCDI climate extremes indices.

On global and annual mean scales, recurrent volcanic eruptions leave centennial warming trends statistically indistinguishable from the volcanic-free baseline (+0.338 vs. +0.334 K dec⁻¹). However, on local and seasonal scales the volcanic signal is substantial. Over Northern Europe, volcanic forcing produces winter surface warming of up to 1.0 K (~30% of the reference warming) and an annual precipitation deficit of −36 mm yr⁻¹. Stratospheric temperature increases by ~+0.4 K at 18 km, while mid-stratospheric polar ozone is enriched. The arctic sea ice area increases by +0.44% through albedo feedback. Probabilistic analysis reveals that annual maximum daily temperature (TXx) exceeds +0.5 K over 16% of global land with a more-likely-than-not probability—a perturbation absent from standard CMIP6 results.

The results demonstrate that moderate volcanic activity, though invisible in long-term global mean trends, generates robust regional climate signals and shifts the probability distribution of climate extremes. This study provides the first probabilistic characterisation of the centennial volcanic climate signal within a fully coupled chemistry–climate model, underscoring the necessity of including stochastic volcanic forcing in future projection frameworks.