Evaluation of Turbulence Parameterization in Gray Zone Simulations of Real-Case Precipitation Events

Abstract In practice, the deep convection parameterization schemes are often switched off in convection-permitting simulations, which enable the turbulence parameterization scheme to be solely responsible for the representation of subgrid-scale (SGS) turbulent fluxes inside convective clouds. However, traditional 3D turbulence schemes or 1D planetary boundary layer (PBL) schemes in convection-permitting simulations adopt the eddy-diffusivity formulation to represent SGS turbulent fluxes in the free atmosphere, leading to insufficient mixing inside convective clouds above the PBL, which further results in the overestimation of heavy precipitation in convection-permitting simulations. In this study, we applied a nonlinear horizontal gradient ( H -gradient) term, which is capable of representing countergradient SGS turbulent transports inside convective clouds, to a 3D turbulent kinetic energy–based turbulence scheme. The results of two heavy precipitation case studies showed that the new 3D turbulence scheme, modified by the H -gradient term, produces satisfactory results that agree well with observations in terms of the magnitude and distribution of precipitation. A 2-month experiment further showed that the modified scheme can reduce the systematic bias of overestimated heavy and extreme precipitation in convection-permitting simulations. Significance Statement The switching off of the convection parameterization scheme is a common practice in convection-permitting simulations. However, traditional turbulence schemes are incapable of representing subgrid-scale turbulent fluxes inside deep convective clouds. The lack of representation for subgrid-scale turbulent transport associated with convective clouds can cause forecast errors in the magnitude and distribution of precipitation in convection-permitting simulations. By implementing a nonlinear term, which accounts for countergradient subgrid-scale turbulent transports inside convective clouds, a 3D turbulence parameterization scheme was found to reduce the systematic bias of overestimated heavy and extreme precipitation in a convection-permitting simulation. It is encouraging that the nonlinear term can help mitigate the overestimation of heavy precipitation, which is a common issue in convection-permitting simulations.