Following the success of Zeus AI’s NASA Phase I & II SBIR, we have been selected for Phase III funding to research, design, and build a Regional Air Quality Digital Twin.
Wildfire driven phenomena such as smoke plumes, aerosol transport, and heat signatures are challenging to track in real-time; and failure to do so can result in cascading hazards. This cost is acutely felt in public health, with vulnerable populations bearing disproportionate risk. Furthermore, the energy sector can see solar generation capacity reductions over 30%, where operational disruptions and volatility impact the grid and it’s derivatives.
Current monitoring systems face limitations in spatial resolution, temporal frequency, and gap-filling capabilities, particularly during nighttime and cloudy conditions. While TEMPO provides hourly observations at 2-4km resolution, measurements are limited to cloud-free pixels during daylight. Traditional physics-based models operate at coarser resolutions and require significant computational resources.
Regional Air Quality Digital Twin
During Phases I and II, Zeus AI developed EarthNet, a high-frequency global atmospheric profile foundation model, as a production-grade platform capable of ingesting vast amounts of satellite data to accurately correlate cross-domain modalities at high frequency, low latency, and global scale. EarthNet streams 6-hour atmospheric profile nowcasts with hourly refresh.
Phase III extends this foundation to regional air quality monitoring with the inclusion of datasets including TEMPO, CrIS hyperspectral sounders, VIIRS imagery, and GEOS-Chem model outputs. Leveraging insight from Visual Autoregressive (VAR) model design, we will produce gap-filled air quality analysis at 2-4km resolution with hourly updates, tracking Aerosol Optical Depth (AOD), Nitrogen Dioxide (NO₂), Ozone (O₃), PM2.5, Sulfur Dioxide (SO₂), and Formaldehyde (HCHO).
Beyond real-time monitoring, the digital twin will benefit “what-if” scenario exploration. Users can modify wildfire footprints, adjust emission scenarios, or alter meteorological conditions—transforming the system from an observational tool into an experimental platform for scientific research and operational planning.
This investment, combining Phase III funding from NASA’s Research and Applications branches within the Earth Science Division and Phase II-e matching funds from the Small Business Innovation Research program, we aim to enhance and rigorously contribute to the atmospheric monitoring toolkit regionally across the continental United States. We deeply appreciate the support from NASA Earth Science Division and look forward to collaborating with the excellent NASA POWER and FireSense teams advancing this critical capability.