Developing a set of simple metrics to evaluate the performance of models with interactive stratospheric aerosols

Various sets of metrics have been used to rank the performance of CMIP6 models for a variety of purpose.

Here we present a potential set of metrics that would simplify a similar ranking for the purpose of evaluating the skill of climate models with a fully resolved stratosphere, including dynamics, chemistry, and aerosol microphysics in representing perturbations to stratospheric composition, such as past volcanic eruptions, as a first step to determine the reliability in future cases such as new volcanic eruptions or Stratospheric Aerosol Intervention.

Our purpose is to find metrics that include available observations of large and medium volcanic eruptions, such as Mt. Pinatubo in 1991, and that also consider the uncertainties in past retrievals, and that are representative of models skills across time (i.e. considering not just the initial plume development, but also the e-folding time and ultimate fate of the aerosol cloud) and space. We include metrics that consider aerosol spatial distribution, local and global size distribution and chemical properties through surface area density.

Our set of metrics could be of great use as more models in CMIP7 start including prognostic aerosols schemes and higher tops, and could inform future strategies for better observations of the stratospheres, as well as identify necessary variables to be requested by CMIP as part of the data requests prioritization for the Atmosphere Working Group.