Microwave Variational Retrieval of Surface and Atmospheric Parameters: Application to AMSU/MHS and SSMI/S sensors

The study presents results from a microwave 1D-VAR algorithm, labeled MIRS (Multi-sensor Integrated Retrieval System) used to retrieve simultaneously surface parameters as well as atmospheric profiles in all-weather conditions and over multiple-surface backgrounds. The final outputs of the algorithm include the humidity and temperature profiles, the cloud and precipitation profiles, the ground skin temperature and the surface emissivity spectra. To allow a stable inversion, the retrieval is performed in reduced space for the atmospheric profiles, including the cloud and precipitation parameters. This is done by performing an Eigenvalue decomposition of the covariance matrix. This approach allows the most pertinent components of a profile to be retrieved and ignores the less important ones, or those for which we do not have sufficient information, avoiding therefore the null-space. The exact number of components to be retrieved must be optimally found and depends on the sensor characteristics and the information content of its measurements. The forward operator used by MIRS is a prototype version of the Community Radiative Transfer Model (CRTM), based on a two-stream approximation for the modeling of multiple scattering effects of clouds and precipitation. The radiance derivatives with respect to the cloud and precipitation parameters are computed analytically as part of CRTM. The MIRS is being tested with the data from the recently launched NOAA-18 AMSU and MHS. It will be also applied for DMSP F-16 SSMI/S which is a first conical sounding microwave instrument. While this study is presented with the overall system design, our special attention will be given to the retrieval and validation results under cloudy and precipitating conditions.