Idealized simulations of supercell thunderstorm as a testbed for two fine-scale models

Three-dimensional idealized simulations of deep moist convection (supercell) are used to evaluate the ability of two mesoscale models, AROME and MesoNH, at kilometric and sub-kilometric scales. AROME is the new operational forecast model of Météo-France with a 2.5-km horizontal resolution (Seity et al. 2010), it is a semi-implicit, semi-Lagrangian spectral model based on the ALADIN-NH dynamical core (Bubnova et al., 1995). MesoNH is a grid-point Eulerian model, this research model can simulate atmospheric motions ranging from the synoptic scale down to LES (Lafore et al. 1998). At 2.5-km resolution, the physics package (including microphysical, shallow convection and 1D turbulence schemes) is the same for both models. At higher resolution, Meso-NH can use a 3D turbulence. Thus, the main differences of the two models come from their dynamical cores (fully compressible system for AROME, “pseudo-incompressible” system of Durran (1989) for MesoNH).

As in previous studies, the environmental conditions are simplified. To trigger convective cells, a warm bubble perturbation (with a maximum of 2 K at the center, a 10 km horizontal radius and a 1.4 km vertical radius) is superimposed on an initial homogeneous base state derived from analytic temperature and moisture profiles, similar to those used in Weisman and Klemp (1982, 1984). Directionally varying wind shear profiles are used. The lower condition is represented by flat terrain. Simulations are conducted without surface physics package and atmospheric radiative heating. The Coriolis force is neglected.

Sensitivity to horizontal resolution is first tested with horizontal grid-spacing ranging from 4 km to 500 m. The impact of different parameters (such as vertical resolution, time step duration, …) and model configuration (numerical diffusion, microphysical set-up with or without ice processes…) is also assessed in the context of different environmental conditions (different wind shear profiles…) favoring supercell formation.

The dynamic evolution of storm cells is analyzed and the experiments are assessed using different diagnostics (cell size, cold pool intensity, propagation speed, time evolution of vertical velocity, integrated precipitation, convective cell trajectories, mean vertical profile of hydrometeors …). Kinetic energy spectra are also used to evaluate the energy distribution into both models and their effective resolution (Lindborg, 1999; Skamarock, 2004).