Direct radiative forcing due to anthropogenic black carbon (BC) is highly uncertain but best estimates suggest a large positive effect (+0.71 [+0.08, +1.27] W m-2). The uncertainty in the total forcing is due to large uncertainties in the atmospheric burden of BC and its radiative properties. The uncertainty in the burden is in-turn due to the uncertainty in emissions (7500 [2000, 29000] Gg yr-1) and lifetime (removal rates). In comparison with the available observations GCMs tend to under-predict absorption near source (e.g. at Aeronet stations), and over-predict concentrations in remote regions (e.g. as measured by HIPPO). By exploring the uncertainties in the dominant emission and removal processes, and in the key radiative property (the imaginary part of the refractive index) and comparing with a variety of observations we hope to better constrain the radiative forcing.

We aim to address the uncertainty in direct radiative forcing in a unique way by developing a new approach to tackle two dominant sources of model uncertainty: structural uncertainty and parametric uncertainty. We will do this via a multi-model perturbed parameter ensemble (MMPPE).

Each participating model will run a 3-parameter perturbed parameter ensemble (PPE). This will consist of 39 pre-defined simulations that will be run for the years 2017 and 1850 + any required spin-up time. The 2017 simulations will be the priority but 1850 simulations are required to calculate the radiative forcing. This is a total of 78 years of simulation + spin-up. The pre-defined simulations will allow statistical modelling to be carried out for defined diagnostics producing sensitivity analyses that will be used to compare individual models following Lee, et al. 2011 and Carslaw et al. 2013. Participants are also requested to submit the results of the one-at-a-time high/low tests used to test the implementation of the perturbation for initial comparisons.