Unified Cloud Retrieval and Initialization Using Radar Observations and the Impact on Numerical Simulations of Deep Convective Clouds during Mesoscale Convective Systems

Inaccurate cloud specification is one of the problems encountered in numerical simulations of clouds and precipitation. It causes erroneous simulations of mesoscale convective systems (MCSs), clouds, and precipitation. In addition, a long spin-up period is usually required for a numerical model to produce reasonable simulations of cloud and precipitation. In this study, a unified method of cloud retrieval and initialization was tested in a framework of variational data assimilation. NEXRAD reflectivity observations are used to derive the cloud hydrometeors in conjunction with the assimilation of radar radial velocity and other conventional observations. It is found that the cloud initialization scheme not only retrieves the cloud hydrometeors from radar observations, but also provides reasonable cloud initial conditions for numerical simulations of MCSs, deep convective clouds and precipitation. The cloud initialization reduces the model spin-up effects and thus improves the simulations of MCSs and their associated clouds and precipitation. Case studies with recent major MCSs are presented.