The PoEMS experiment will involve each interactive stratospheric aerosol model running a perturbed parameter ensemble (PPE) of simulations through the 1991-1995 Pinatubo-perturbed period. Variation-based sensitivity analysis will derive a probability distribution function (PDF) for each model’s predicted Pinatubo forcing, following techniques applied successfully to quantify and attribute sources of uncertainty in tropospheric aerosol forcings (e.g. Carslaw et al., 2013). The approach will teach us which aspects of the radiative forcing from major eruptions is most uncertain, and will enable us to identify how sensitive model predictions of key features (e.g. timing and value of peak forcing and decay timescales) are to uncertainties in several model parameters.  By comparing the time-signatures of different underlying aerosol metrics (mid-visible AOD, effective radius, particle number) between models, and crucially also against observations, may also help to reduce the natural forcing uncertainty, potentially thereby making the next generation of climate models more robust.

* Note that when imposing the parameter-scaling, the models should only enact the change in volcanically-enhanced air masses (where the total sulphur volume mixing ratio exceeds a threshold suitable for their model). Perturbing only the volcanically-enhanced air masses will ensure, pre-eruption conditions and tropospheric aerosol properties remains unchanged by the scalings.