Initialization and dynamical stabilization of a cloud-resolving model

A multiscale cloud-resolving model (CRM) composed of the anelastic grid equations coupled to viscous turbulence subgrid equations for the wind velocity is presented along with an initialization and dynamical stabilization method named Insertion & Brownian Motion (IBM). The subgrid model avoids the saturation of small scales. Its closure was driven by the assumption that the equilibrium law of energy production holds at small scales. The IBM method is shown to be effective in a 10-day long imperfect run with the CRM where the horizontal components of the grid velocity and the potential temperature are set to their reference value at the initial time. The model imperfection consists in a 30% raise in the heat transfer coefficient at surface, which directly impacts the potential temperature which is a control variable for the vertical component of the grid velocity. Future work will test the viability of the viscous turbulence equations as a subgrid model for global models that can be properly initialized and is less computationally demanding than 3D super-parametrization.