Examination of model output reveal that the two cases were accurately simulated by the model. For the December case, the northwesterly flow of air over Lake Superior created superadiabatic lapse rates over the water surfaces with mixing continuing over land surfaces. Only a slight decrease in boundary layer depth occurred as air passed over the Upper Peninsula of Michigan. With the lake surface temperatures increased by 5°K, the well-mixed boundary layer and cloud bands expanded in depth and more lake-effect band formation occurred to the south as more heat and moisture became available from the lake surface. The opposite effect happened when the lake surface temperatures were decreased by 5°K; a reduction in bands over land masses was apparent. The February case showed similar boundary layer conditions to the December case initially. However, when the lake surface temperatures were increased by 5°K, the bands and boundary layer remained very similar to the base simulation as surface air temperatures were already significantly colder than the lake. The decrease in 5°K again produced thinner cloud and boundary layers. For both events, the change in land-surface roughness length produced no variations in the simulations.
Overall, inversion heights were comparable with time between the two cases. Lake surface-850 hPa temperature differences were typically higher for the February simulations, but L2L bands were generally observed to form between 15°C and 24°C when comparing all of the simulations. Wind speeds appear to affect L2L band formation the most in this study, as wind speeds during the start of L2L convection were approximately 15 m s-1 in both events. Lake surface to 2-m land-surface air temperature differences are also presented for each of the simulations.