Uncertainty in Microwave Radiative Simulations Associated with Cloud Scattering Database Used in CRTM

Assimilation of microwave radiance measurements in numerical weather prediction (NWP) models improved significantly the global medium range forecast skill. Cloudy radiances contain rich information on atmospheric temperature, humidity and hydrometeor profiles except for heavy precipitation. However, large uncertainties exist in the forward radiative transfer models, NWP model physical parameterization, as well as an inconsistency of moisture physics between the two. This study investigates the uncertainty arising from uses of different cloud scattering lookup tables (LUTs) in the Community Radiative Transfer Model (CRTM) developed by the US Joint Center of Satellite Data Assimilation (JCSDA). The current database is a default LUT for scattering coefficients in microwave frequencies generated under the assumption of Mie theory, which assumes spherical liquid and ice water cloud particles. Large uncertainties are found in cloudy radiance simulations, especially in deep-convective clouds. A new LUT based on discrete dipole approximation (DDA) is developed and tested. In DDA, an ice particle is represented by a three-dimensional array with polarizable points. The CRTM with the new DDA based LUT provides more realistic optical properties of non-spherical ice particles than the current LUT. Numerical results are compared for the simulations of the Advanced Technology Microwave Sounder (ATMS) radiance measurements within and around tropical cyclones using the same CRTM but with two different LUTs. The atmospheric state profiles from European Centre for Medium-Range Forecast (ECMWF) and the hydrometeor profiles from collocated Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) are used as inputs to both simulations. Large differences are found in high frequency range for ATMS humidity sounding channels 17-22. An underestimation of the scattering effect on ATMS radiance simulations of these channels in areas with large ice water content by the Mie theory is corrected by the DDA method. The root causes of the discrepancy of simulated brightness temperatures between two LUTs are investigated. The bulk volume density of ice particles assumed differently in the default Mie-based and the new DDA-based LUTs seems to be a major cause, and the particle shape has a secondary effect.