Convergence behavior of idealized convection-resolving simulations of summertime deep moist convection over land

Abstract

Convection-resolving models (CRMs) can explicitly simulate deep convection and resolve interactions between convective updrafts. They are thus increasingly used in numerous weather and climate applications. However, the truncation of the continuous energy cascade at scales of $$\mathcal {O}$$ (1 km) poses a serious challenge, as in kilometer-scale simulations the size and properties of the simulated convective cells are often determined by the horizontal grid spacing ( $$\Delta x$$ ).In this study, idealized simulations of deep moist convection over land are performed to assess the convergence behavior of a CRM at $$\Delta x =$$ 8, 4, 2, 1 km and 500 m. Two types of convergence estimates are investigated: bulk convergence addressing domain-averaged and integrated variables related to the water and energy budgets, and structural convergence addressing the statistics and scales of individual clouds and updrafts. Results show that bulk convergence generally begins at $$\Delta x =4$$ km, while structural convergence is not yet fully achieved at the kilometer scale, despite some evidence that the resolution sensitivity of updraft velocities and convective mass fluxes decreases at finer resolution. In particular, at finer grid spacings the maximum updraft velocity generally increases, and the size of the smallest clouds is mostly determined by $$\Delta x$$ . A number of different experiments are conducted, and it is found that the presence of orography and environmental vertical wind shear yields more energetic structures at scales much larger than $$\Delta x$$ , sometimes reducing the resolution sensitivity. Overall the results lend support to the use of kilometer-scale resolutions in CRMs, despite the inability of these models to fully resolve the associated cloud field.