The sensitivity of numerically simulated multicell convection to grid spacing and computational mixing coefficients

Previous studies have shown that numerically simulated deep moist convection using a large eddy simulation (LES) cloud model is sensitive to changes in user specified model parameters. Specifically, changes in model grid spacing have been shown to alter the simulated convection. Grid spacing on the order of tens of meters is necessary for a statistically converged solution of atmospheric boundary layer convection, but has not been shown to produce a converged solution for deep moist convection. Although this work does not fully investigate the sensitivity of an isolated multicell thunderstorm to grid spacing, results are presented from a series of numerical simulations using a range of grid spacings and other model parameters. Turbulence theory is used to evaluate the model's prediction of the energy cascade in the inertial subrange and select an appropriate grid spacing in the absence of a grid converged solution. The method involves using the Discrete Cosine Transform to extract spectral information from model-predicted kinetic energy. The predicted spectra are compared with the expected spectrum provided by the turbulence theory. Using this spectral method, the effects of computational mixing coefficients on the simulated convection are also investigated.