Modeling for detection

International agreements governing SAI would likely require verification mechanisms to ensure that injections comply with agreed strategies and to detect any unauthorized or undeclared deployments. Detecting whether and how SAI has occurred requires understanding what observable signatures different injection scenarios would create in the stratosphere and troposphere and at the surface. It is unlikely that any climate-relevant SAI could go undetected (Smith et al., 2026), but injection at different locations, magnitudes, and times would produce distinct patterns of aerosol sizes and spatial distributions.

By modeling these deployment signatures in advance, researchers can develop detection methodologies that compare actual observations to model predictions, demonstrating whether observed aerosol anomalies are consistent with reported SAI activity. This modeling is essential for establishing credible verification procedures that could support international governance frameworks and for providing confidence that potential deployments can be reliably detected.

This activity involves modeling SAI deployment scenarios across a range of injection locations (tropical, mid-latitude, and polar), magnitudes (from small pilot deployments to large-scale operations), altitudes, and durations to characterize the resulting stratospheric aerosol signatures. Model simulations will produce predictions of the aerosol optical depth patterns, particle size distributions, and spatiotemporal evolution that would result from each deployment scenario. The work includes: (1) identifying observable features that distinguish SAI-induced aerosol changes from natural variability; (2) assessing the ability to detect the impacts of injection on the stratospheric aerosol burden as a function of injection magnitude, location, and duration; (3) evaluating how detection capability depends on observational system characteristics (spatial resolution, temporal frequency, measurement precision); and (4) developing detection methodologies that could be applied to real observational data. A key deliverable will be a quantitative assessment of the minimum injection magnitude detectable for specific injection locations and times, given current and near-future observational capabilities, and the identification of observational configurations that optimize detection sensitivity.