Optimizing regional regression coefficients for AIRS profile retrieval for direct broadcast users over Indian region

Abstract

The Atmospheric Infrared Sounder (AIRS) onboard Aqua satellite is providing a wealth of highly accurate atmospheric and surface information using 2378 high-spectral-resolution infrared (3.7 - 15.4 μ) channels. Cooperative Institute for Meteorological Satellite Studies (CIMSS) has developed International MODIS/AIRS Processing Package (IMAPP) to retrieve atmospheric and surface parameters from AIRS-L1B radiance measurements. CIMSS retrieval algorithm is based on principal component regression technique. In order to account for retrieval dependency on zenith angle and regional/seasonal variations a classification scheme is employed based on scan angle classification and window-channel brightness temperature classification. To improve atmospheric sounding retrieval for a specific region, which is useful for AIRS direct broadcast users, regional regression coefficients have been generated for Indian region. Training dataset of radiosonde observations over India and surrounding region have been used to generate regional regression coefficients for IMAPP-AIRS processing. Retrieval error statistics was generated using simulated radiances from independent dataset of radiosonde observations over Indian region. This study shows that the Root Mean Square (RMS) error in humidity profile is reduced by ~25% when compared to the global regression coefficients, whereas RMS error for temperature profile is reduced by ~0.2 K. This study is also useful for sounding retrieval from geostationary sounder measurements, for example, for Geostationary Operational Environmental Satellite (GOES) Sounder and INSAT-3D Sounder that have observations over a limited region with high spatial and temporal resolution.