Incorporation of Parameterized Convection in the Synoptic Study of Large Scale Effects of the Great Lakes

Abstract

This study deals with the influence of convective precipitation on the synoptic effects of the Great Lakes in winter during a strong cold air outbreak. The detailed synoptic situation during the period of study is presented. Various parameterization schemes are reviewed. Kuo's (1965) procedure of parameterization is modified by permitting mixing of the parcel and environment before the parcel is lifted. This modified procedure is incorporated in the eight-level primitive equations model of Danard and Rao (1972) and Danard and McMilan (1974). The atmospheric water balance components are computed to evaluate diagnostically the parameters used in the parameterization procedure. A value of 60 min is obtained for the time parameter and 0.5 for the mixing factor. Thus these parameters, which are used in numerical predictions described below, are not arbitrary. To isolate the large-scale effects of the lakes, three 36 h numerical forecasts are performed by assuming all grid points are land (identified as NL), by incorporating fluxes of sensible heat and water vapour for open water points (identified as L), and by incorporating convection (identified as CU). Seven grid points have been chosen to represent conditions over the Great Lakes. The salient features of these three 36 h forecasts are as follows: The average observed and predicted heights over the Great Lakes points indicate that CU forecast heights are closest to the observed at 850, 700, 500, and 300 mb while NL predictions are best at 1000 mb. The CU prediction is better than L at all levels. The Great Lakes cause a lowering of heights at 1000 mb by 8 m for the CU prediction and about 24 m for the L forecasts. At 500 mb the heights are raised by 19 m for the CU forecast while no increase is seen for the L forecast. The locations of the precipitation maxima for CU predictions are found to agree well with the observed pattern. The precipitation amounts for the CU forecasts are better than NL. The results demonstrate the feasibility of incorporating convective heating and precipitation in the numerical model during an intense cold air outbreak. Suggestions for improving the parameterization procedure are given.