This project will integrate DeepCTM, an advanced machine-learning model that emulates traditional chemical transport models (CTMs), into NOAA’s operational National Air Quality Forecast Capability (NAQFC). The goal is to improve the speed, accuracy, and resolution of air quality forecasts at both national and global scales. Additionally, the model generates sensitivity information that helps adjust key inputs, such as emissions and initial conditions, to improve forecast performance in a scientifically informed way.
The primary outcome will be a DeepCTM-enabled air quality forecasting system embedded within the Unified Forecast System (UFS). This system will:
- Produce faster forecasts
- Combine diverse observational data more efficiently
- Support full-chemistry simulations at finer spatial resolution
- Extend forecasting to larger regional and global domains
With its online coupling design, the system will also support future research on the interactions between atmospheric chemistry and weather. As it transitions from offline testing to full operational use, NOAA will gain a faster, high-resolution global air quality forecasting capability that provides more efficient and accurate daily guidance.
