Initial error growth in a simulated supercell storm

In the preceding studies by many authors, in particular they found that moist processes were responsible for the strong initial error growth in mesoscale. hi the present study they take a more systematic look at the processes by means of the initial introduced mall errors and found that the errors first grow as small-scale differences associated with moist convection, then spread upscale as their growth begin to slow. In the context, we use vastly different initial perturbation methodologies to investigate the initial error growth in the storm scale with open boundary conditions. Comparison of the perturbation methodologies indicates that the ensuing patterns of ensemble spread converge within only a few minutes, irrespective of the initial perturbations employed. In the vertical direction, the largest errors in different variable fields concentrated in different layers (e.g., the largest errors in the temperature field concentrated in the upper tropopause, but in the horizontal wind field, the largest errors converged in the troposphere.). The error growth in the first and middle time contact with the storm tightly, but at last, the error growth goes their ways very slow and flat. The growth of the uncertainties is limited by the saturation effects, which in turn is controlled by the larger-scale atmospheric environment.