SAI Research Roadmap

PHASEFoundational knowledgeWhat can existing evidence tell us, and where does it fall short?Time2–3 yrsCost~$30M–$75M

PHASEAerosol microphysicsDoes data from small outdoor experiments validate the processes in our models?Time4–8 yrsCost~$70M–$150MEst. uncertainty reduction (cooling efficacy)~25%

PHASEStratospheric distributionDo sulfate aerosols mix and distribute, globally, in the way our models predict?Time4–11 yrsCost~$270M–$1.14BEst. uncertainty reduction (cooling efficacy)~66%

PHASEMonitored coolingAs cooling begins, do observed effects match predictions? Are any unexpected effects emerging?TimeOngoingCostTBD

ImpactsClimate response, impacts, and risks

AerosolsMicrophysics, chemistry, transport

ObservationsMonitoring systems, instruments, and data

EngineeringAircraft R&D, materials handling, operations

Aerosols Observations

Observation needs for aerosol microphysics experiment

Simulated observing campaign to determine the measurements and sampling frequency an aerosol microphysics experiment requires

Est. direct cost: $300K–$500K
Est. duration

Why it’s on the roadmap

To define the observational capabilities necessary for a successful aerosol microphysics experiment, researchers need to simulate, in detail, the observations the experiment would take and how those observations would be used. Even with detailed simulations of how a plume (the cloud of injected aerosol particles) should evolve, it is difficult to predict whether real-world observations — which are inherently limited in spatial coverage, temporal frequency, and measurement precision — will contain sufficient information to distinguish between competing physical processes or detect deviations from predictions.

An observing system simulation experiment (OSSE) directly addresses this problem. It uses a detailed model simulation as synthetic "truth," applies realistic observational constraints to create synthetic data, and then tests whether those synthetic observations contain enough information to accurately reconstruct the original plume evolution using inverse methods (statistical techniques that work backward from measurements to the underlying physical state). This approach allows researchers to ask quantitative questions: if the plume is sampled every hour versus every day, or from two locations versus ten, can researchers still infer the underlying aerosol microphysics — how the injected particles form, grow, and interact? By testing different observational strategies before deploying expensive instruments, researchers can optimize the experimental design and ensure that the aerosol microphysics experiment yields scientifically actionable results.

Scope of work

This activity uses the detailed multi-scale plume representation from prior work as a truth model. Researchers apply an observation operator — specialized code that simulates how different instruments would measure the plume — to this truth model to create synthetic observations with realistic characteristics such as measurement precision, temporal resolution, and detection limits. These synthetic observations then update a separate microphysical model (the assimilation model) using inverse methods. The goal is to determine how well the assimilation model can recover the properties of the original truth model using only the synthetic observations.

Research groups will test multiple observational strategies, varying parameters such as measurement cadence (e.g., hourly versus daily sampling), spatial sampling density (e.g., single-location versus multi-location deployments), and measurement capabilities (e.g., which aerosol properties are measured and with what precision). For each observational strategy, researchers will apply the inverse method to assess how closely the updated assimilation model matches the truth model.

This activity will determine the minimum observational requirements needed to confidently characterize aerosol microphysical processes during an aerosol microphysics experiment. The results will directly inform decisions about which instruments to deploy, how frequently to sample, and how many measurement locations are necessary. The activity will also identify which measurements provide redundant information versus which are critical for distinguishing between different physical processes, thereby optimizing both the scientific value and cost-effectiveness of the observational campaign.

Open full page