SAI Simulator v1.3 Update: Sea Ice, Model Variability, and Data Export

What's new in version 1.3 of Reflective's SAI Simulator

Photo by NASA Earth Observatory

We’re introducing an updated version of the Stratospheric Aerosol Injection (SAI) Simulator that brings several new features: Arctic sea ice extent projections, data export for downstream research, and updated model variability estimates.

Sea ice extent is now included as a new impact in the Simulator, enabling users to explore how Arctic sea ice would respond under different SAI deployment scenarios.

The Arctic is warming roughly four times faster than the global average, and the rapid loss of sea ice has cascading effects on ecosystems, weather patterns, and communities that depend on polar environments. A key question for SAI research is whether, and to what extent, stratospheric aerosol injection could slow or reverse Arctic sea ice decline.

The new sea ice module uses CESM2-WACCM GAUSS and ARISE simulation data with a logistic-based model fit to project September sea ice extent (the annual minimum, when Arctic ice coverage reaches its lowest point after the summer melt season) under various SAI scenarios. Observational data from the National Snow and Ice Data Center (NSIDC Sea Ice Index v4) are also integrated, enabling direct comparison between model projections and the historical record. This allows users to investigate questions like: How much Arctic sea ice could be preserved under a given SAI deployment? How do different injection strategies compare in their ability to protect sea ice?

We are also rolling out data export capabilities that allow users to download model output for each impact variable directly from the Simulator. Previously, users could only view results within the application. Now, researchers and analysts can take the Simulator’s projections into their own environments, enabling extended analysis and integration into downstream research workflows. For example, a scientist studying food security could export projected temperature and precipitation data and combine it with crop models or other tools.

We’ve updated how model variability is calculated, improving the uncertainty estimates displayed across the Simulator. On a technical level, we revised regional estimates of model internal variability, which impacts the calculated p-value and significance hatching across temperature, precipitation, and other variables. These changes give users a more robust picture of where projected changes are, and are not, statistically distinguishable from the natural variability of the climate. Users familiar with the hatching feature introduced in v1.2 will notice that some regions may now appear hatched (or unhatched) differently, reflecting these improved estimates.

If you’d like to explore these new features, please visit the user guide page. If you have questions, feedback, or ideas for features you’d like to see in future versions, please get in touch.