The Penn State/NCAR Mesoscale Model version 5 (MM5) was run for seven cases from 2001 in which a sea breeze develops along the eastern New England coast. The results from model runs with 36, 12 and 4 km grid spacing are compared with both actual observations as well as output from the National Centers for Environmental Prediction Eta model on a 40 km grid. As expected, analysis of the individual cases shows that the higher resolution grids are able to resolve realistic details in the flow, details the 36 km grids and the 40 km NCEP Eta model miss entirely. Contrary to expectations, the error statistics show that the ability to forecast specific variables at specific locations does not improve significantly when using higher resolution. Even more surprisingly, the 4 km grids produced the biggest errors in forecasting surface dewpoint.

Details from these cases will be presented, concentrating on the 4 km grids, showing the development of the circulation and the forcing mechanisms operating in the model. The role of the surface energy balance in these runs will also be shown, with some of the differences between surface energy balance parameterizations brought out as well.