Enhancing Atmospheric Profiling with the AMETHYST Hyperspectral Inversion System

In this talk, we will discuss plans to maximize the extraction of information from the inversion of hyperspectral infrared (IR) and microwave data obtained from polar-orbiting instruments.  Hyperspectral remote sensing, which involves capturing Earth’s emitted energy in the IR spectrum, has become a pivotal tool for understanding atmospheric and surface conditions. A central challenge in this field is the efficient and accurate inversion of hyperspectral IR data to extract quantitative physical and chemical properties. The AdaptiveMETeo HYperSpectral Transformer (AMETHYST) inversion system represents a significant advancement in this regard. It enables the characterization of vertical atmospheric columns by considering variables such as temperature, pressure, and humidity at various altitudes. By comparing observed hyperspectral IR data with simulated radiance, AMETHYST adjusts model parameters to accurately retrieve vertical atmospheric structures.

AMETHYST leverages both hyperspectral data and numerical weather prediction (NWP) model forecasts to produce thermodynamic profiles and transformed retrievals (TRs). These TRs, developed using Migliorini’s transformation, are particularly suited for regional model assimilation due to their reduced data volume, instrument-specific adaptations, and simplified observation error covariance. Recent studies, including Cherubini et al. (2023), have demonstrated the effectiveness of TR assimilation in improving moisture field characterization over the central North Pacific Ocean, particularly at mid-atmospheric levels. This improvement is crucial for refining cloud and precipitation predictions.

Additionally, our team is extending the capabilities of AMETHYST to analyze physical profiles above cloud tops in the Arctic by integrating cloud masks and properties from various satellite instruments, differentiating it from its predecessor, MIRTO.