Simulations of Orographic Convection across Multiple Grey Zones

As weather and climate models move to finer grids, they traverse so-called “grey zones”, or ranges of horizontal grid spacings where important processes transition from fully parameterized to fully explicit. Within these grey zones, scale-separation between the process of interest and the grid scale, a fundamental assumption of most subgrid parameterizations, breaks down. For the moist convection problem, which is characterized by horizontal scales of O(10 km) down to O(100 m), two important numerical grey zones exist: the deep convection [O(10 km) to O(1 km)] and turbulence [O(1 km) to O(100 m)] grey zones. In this study, idealized simulations of orographic convection crossing both grey zones are conducted to quantify the errors stemming from inadequate resolution of cloud processes. These experiments consider two different mechanisms by which orography initiates moist convection: mechanical and thermal forcing. To aid the interpretation, a new method of diagnosing entrainment/detrainment in large-eddy simulations is developed. For both convection-initiation mechanisms, parameterized convection causes large biases in the timing, intensity, and/or horizontal distribution of clouds and precipitation. Although the results improve for “convection-permitting” grids of O(1 km) and below, they do not necessarily converge to a robust solution as the grid spacing is decreased. Robust solutions are found only in the thermally forced cases, where well-developed turbulence forms in the subcloud layer prior to convection initiation. By contrast, in the mechanically forced flows that lack widespread turbulence, the orographic convection manifests as a turbulent-transition process. Because the rate of this transition depends on the grid spacing, numerical convergence does not occur. However, if boundary-layer turbulence is seeded by a patch of surface heating in the upstream flow, a similar degree of numerical convergence as the thermally forced cases is obtained.