Assimilating Cloud Water Path as a Function of Model Cloud Microphysics in an Idealized Simulation

Abstract

Abstract The sensitivity of assimilating satellite retrievals of cloud water path (CWP) to the microphysics scheme used by a convection-allowing numerical model is explored. All experiments use the Advanced Research core of the Weather Research and Forecasting Model (WRF-ARW), with observations assimilated using the Data Assimilation Research Testbed ensemble adjustment Kalman filter and a 40-member ensemble. Three-dimensional idealized supercell simulations are generated from a deterministic WRF nature run started from a homogeneous set of initial conditions. Four cloud microphysics schemes are tested: Lin–Farley–Orville (LFO), Thompson (THOMP), Morrison double-moment (MOR), and Milbrandt–Yau (MY). For the idealized experiments, assimilating CWP generates a mature supercell after approximately 1 h for all microphysics schemes. Vertical profiles of ensemble covariances show large differences in the relationship between CWP and various hydrometeor mixing ratios. While the differences in overall CWP are small, the experiments generate very different reflectivity analyses of the simulated storm, with MOR and MY underestimating reflectivity by a large margin. Vertical profiles of hydrometeor mixing ratios from each experiment are generally consistent with scheme design, such that the Thompson scheme characterizes the storm top as mostly snow whereas the Milbrandt–Yau scheme characterizes the storm top as mostly ice. The impacts of these differences on 30-min forecasts show that MOR and MY are unable to maintain convection within the model while THOMP and LFO perform somewhat better, though all fail to capture the divergent movement of the storm split in the nature run.