Accounting for Cloud Microphysical Properties in Variational Data Assimilation of Passive Microwave Brightness Temperatures

Recent emphasis has been placed on the assimilation of space-born passive microwave observations which are affected by clouds and precipitation toward improving the initialization and quality of Numerical Weather Prediction (NWP) forecasts. However, progress in fully exploiting these observations in data assimilation systems is still hindered by the many assumptions about the cloud microphysical and optical properties during radiative transfer simulations of the background fields, including water/ice content, density, and particle size and shape, not to mention errors resulting from displacement between background fields and observations. The result is that either the observations are removed by quality control (QC) procedures, or they are assimilated with inflated observation errors to minimize any negative impact on the state variables due to uncertainties from the microphysical properties assumptions. To address this issue, we test the impact of accounting for hydrometer microphysical properties in physical 1DVAR assimilation using the Multi-Instrument Inversion and Data Assimilation Preprocessing System (MIIDAPS). The variational assimilation in MIIDAPS is applied to passive microwave brightness temperatures to produce analyses of temperature, humidity, and hydrometeor profiles, as well as the skin temperature and surface emissivity, using the Community Radiative Transfer model for both forward and Jacobian operators. To reduce the forward model uncertainties in cloudy and precipitation-affected observations, we extend the analysis state vector to include the hydrometeor effective radius (re) for non-precipitating cloud, rain, and graupel-ice. In this study, we will present the sensitivity of passive microwave simulations to re along with an error analysis of the 1DVAR state variables when re is not included. Finally we will present the impact of including re in the 1DVAR state vector and future plans for extending this capability to full 3D/4DVAR data assimilation in order to see the impact on analysis and forecast as well.