Dry Air Above the PBL As a Convection Initiation Failure Mode in Idealized Simulations

Recent work by the authors and others has highlighted the importance of using a greater variety of realistic thermodynamic and kinematic vertical profiles as the base state in idealized simulations of deep moist convection. Since the inception of three-dimensional numerical modeling of convective clouds, modelers have noted the difficulty in initiating sustained convection in idealized models with vertical profiles from some environments that support real convection. While numerous deficiencies in an idealized model could be blamed for this disparity depending on the situation, one “trick” modelers often resort to is increasing the moisture in the shallow layer directly above the PBL in the vicinity of the lifting condensation level (LCL) and the level of free convection (LFC).

In this presentation, we test the hypothesis that overly dry layers between the LCL and LFC (or surrounding these levels if they are at similar altitudes) is a common cause of convection initiation (CI) failure in numerical models. To test this hypothesis, we divide a dataset of Rapid Refresh (RAP) model analysis proximity soundings from supercellular severe convection into quartiles based on the relative humidity (RH) in the 1-3 km AGL layer. Distributions of sounding-derived parameters in each group are compared to investigate what other features of the dry subsets may have compensated for low RH to allow for CI. The same is done for environments of non-supercellular severe convection. Each group of soundings (both for supercells and non-supercells) is composited and used to initialize horizontally homogeneous idealized simulations. We analyze which composites promote CI in the simulations and which do not. For those that do not promote CI, we systematically increase the RH in the LCL-to-LFC layer to determine which, if any, threshold of RH in this layer promotes CI. For select environments, we also explore the sensitivity to model grid-spacing, subgrid-scale turbulence parameterization, and trigger mechanism (warm bubble, updraft nudging, surface fluxes, low-level convergence, etc.). Finally, recommendations for sounding modification for simulations with realistic environments will be presented, and the implications of our findings for the effects of dry air in this layer for real convection initiation will be discussed.