Applications of Principal Component Analysis (PCA) to AIRS Data

The Atmospheric Infrared Sounder (AIRS) on the NASA EOS AQUA platform is providing vastly improved information of the temporal and spatial structure of key atmospheric parameters, such as temperature, moisture and clouds, which are needed to significantly improve real-time weather forecasting, and climate monitoring and prediction capabilities. Also important trace gases such as ozone, carbon dioxide, carbon monoxide, and methane can be derived. High spectral resolution infrared radiances from AIRS are also being tested for direct assimilation into numerical weather prediction models. The soundings and radiances are provided in near real-time by NOAA/NESDIS to NWP community.

A very important part of our AIRS processing is to apply Principal Component Analysis (PCA) to the original AIRS 2000+ channel radiances. PCA is used for detector monitoring and noise filtering/estimating, channel recovery and radiance reconstruction, and for deriving profiles of temperature, moisture, ozone and other geophysical parameters. Since PCA has the ability to reduce the dimensionality of a dataset while retaining most of the information, investigations are being done on its applications to AIRS data compression and archiving. Data compression is one of the key issues for the new generation of high spectral resolution instruments.

Our research and prototyping will allow us to provide valuable information and lessons-learned to the future sensors, such as NPOESS CrIS advanced infrared sounder. Examples of each application, along with the details on the generation and application of eigenvectors, will be presented at the meeting.