Assimilation of NEXRAD data to improve numerical simulations of mesoscale convective systems with cloud initialization

Accurate numerical prediction of mesoscale convective systems (MCSs) is of great importance yet it remains a challenging problem. Many poor MCS forecast can be attributed to the errors in the initial conditions. Specifically, cloud initialization is not included in many data assimilation systems. Thus, the representation of clouds in the numerical models much relies on spin-up of the cumulus and microphysical processes. In this study we investigate the impacts of cloud initialization on numerical simulations of MCSs at a cloud-permitting scale. The NCEP Gridpoint Statistical Interpolation (GSI) data assimilation system and its cloud analysis package were used. The NEXRAD observations were assimilated into an advanced research version of the Weather Research and Forecasting (WRF) model. First, radar reflectivity observations were assimilated with the GSI cloud analysis package to derive the initial cloud analysis of hydrometeors. The subsequent forecasts indicated that the cloud initialization reduced the model spin-up effects and thus improved the simulations of convective systems. Cloud analysis results were sensitive to the choice of the method of hydrometer retrievals and cloud temperature analysis. In addition, the radar radial velocities were also assimilated, in conjunction with cloud initialization, to obtain a realistic local dynamical field. The impacts of assimilation of NEXRAD observations, especially the cloud initialization, are evaluated in details. Two cases during May 2011 are presented.