Search

Numerical simulations are used to study transitions between boundary layer rolls and more cellular convective structures observed during a lake-effect snow event over Lake Michigan on 17 December 1983. Weak lake-effect nonroll convection was observed near the eastern (downwind) shore preceding passage of a secondary cold front. After frontal passage horizontal wind speeds in the convective boundary layer increased, with subsequent development of linear convective patterns. Thereafter the convective pattern became more three-dimensional as low-level wind speeds decreased. Little directional shear was observed in any of the wind profiles. Numerical simulations with the Advanced Regional Prediction System model were initialized with an upwind sounding and radar-derived wind profiles corresponding to each of the three convective structure regimes. Model-derived reflectivity fields were in good agreement with the observed regimes. These simulations differed primarily in the initial wind speed profiles, and suggest that wind speed and shear in the lower boundary layer are critical in determining the linearity of convection. Simulation with an upwind-overlake wind profile, with strong low-level winds, produced the most linear model reflectivity structure. Fluxes and measures of shear-to-buoyancy ratio for this case were comparable to observations. Model sensitivity tests were conducted to determine the importance of low-level wind speed and speed shear in determining the linearity of convection. Results are consistent with trends expected from ratios of buoyancy to shear (but not proposed numerical threshold values). Eliminating all directional shear from the initial wind profile for the most linear case did not reduce the degree of linearity, thus showing that directional shear is not a requirement for rolls in lake-effect convection. Elimination of clouds (principally latent heating) reduced the vertical velocities by about 50%. It was found that variations in wind... [ABSTRACT FROM AUTHOR]