Model vs observation comparison tools

Confidence in SAI modeling depends on how well models reproduce observed stratospheric behavior, since this underpins predictions of where aerosols go and how long they reside in the stratosphere, as well as their chemical composition and growth, which control their sunlight reflection and absorption. Determining how well models represent reality and better constraining uncertainty requires a systematic comparison of model output against observations across key metrics of stratospheric representation. These metrics include stratosphere-troposphere coupling and stratospheric circulation.

Today, performing these comparisons requires significant technical effort, which limits how often and how broadly they are done. The field also lacks a systematic assessment of the full set of specific models and versions used for recent SAI simulations (through GeoMIP). These GeoMIP models show a wide range of behavior in the stratosphere, with, for example, nearly a factor of two difference between two leading models in the sulfate aerosol lifetime under a recent set of SAI simulations.

A standardized, easy-to-use suite of tools would lower that barrier and enable modeling centers to routinely evaluate their own model's performance in the stratosphere. These tools would also make the analyses reproducible by others in the field, accelerating model improvement and building the kind of shared, verifiable evidence base that sound decision-making on SAI requires.

A team of scientists led by Dr. Abalos produced a comprehensive set of model-observation analyses for chemistry-climate models (CCMs) (Abalos et al., 2026). This research activity builds on their work by adapting the existing analyses to cover the models used specifically for SAI experiments. Rather than starting from scratch, the primary task is making those analyses more flexible and generally usable across modeling centers. The practical goal is a standardized, accessible suite of tools that researchers familiar with a given Earth system or chemistry-climate model can use on that model quickly and without significant technical overhead. Where possible, this suite should connect to existing evaluation infrastructure, such as the CMIP7 Rapid Evaluation Framework, to maximize reach and ease of adoption.

The metrics will cover key aspects of stratospheric dynamics relevant to SAI — including stratosphere-troposphere coupling and stratospheric circulation — evaluated across mean state, annual cycle, and interannual variability. A first step in the project will be to define the full list of metrics most appropriate for the SAI case. The project is also expected to include a first inter-model comparison paper on the stratospheres simulated by the models participating in GeoMIP (G6-1.5K experiments), demonstrating how they compare to observations, one another, and the full CMIP6 ensemble.