Recent advancements by researchers at Google have demonstrated the potential to improve monitoring of atmospheric carbon dioxide (CO2) levels using existing weather satellites. This innovative approach utilizes a physics-guided neural network to analyze high-frequency data from the GOES East satellite, enabling estimates of CO2 concentrations every 10 minutes.
This development is crucial as it addresses the global need for enhanced greenhouse gas monitoring amid climate change. Traditional satellites like NASA's OCO-2 provide high-precision data but are limited in coverage, revisiting locations only every 16 days. In contrast, geostationary satellites can cover significant areas more frequently, though they lack specific capabilities for CO2 mapping.
By employing the Enhanced Research Applications (ERA) framework, Google’s model integrates data from 16 wavelength bands along with meteorological factors, achieving more detailed CO2 tracking. At the International Workshop on Greenhouse Gas Measurements from Space, the researchers validated their model's effectiveness against OCO-2 data, showcasing its ability to capture real variability in CO2 levels.
This research holds significant implications, not only for scientific understanding but also for the efficient use of existing observational infrastructure in monitoring greenhouse gases.