84th AMS Annual Meeting

Monday, 12 January 2004
Micro-Scale Simulation of a Lake Michigan Land Breeze Front Through the Downsizing of a Regional Scale Simulation
Hall 4AB
Gijs De Boer, University of Wisconsin - Madison, Madison, WI; and G. J. Tripoli and E. W. Eloranta
As part of the Lake Induced Convection Experiments (LAKE-ICE) on 21 December 1997, the University of Wisconsin Volume Imaging Lidar (UW-VIL) observed a visually intriguing, very shallow (100m) land breeze front undulating off the Lake Michigan shoreline at Sheboygan (Wisconsin) Point. Backscattering returns revealed pollutants trapped within the cold air extending offshore from the mainland. Eventually, as snow showers moved in from the east, the front withdrew westward across the shoreline into Wisconsin, clearing the air of pollutants as it moved westward. It was hypothesized that these small scale undulations and movements were actually deterministic and predictable in the 24 hour time frame because it was a localization of the synoptic system, modulated by the mesoscale effects of the land/lake thermal and moisture contrast. To investigate this hypothesis, we attempted to simulate the observed structure at the micro-scale by simply downscaling a regional synoptic scale simulation of the days weather to resolve microscale topographical and land use features that are believed to have created the observed frontal behavior in response to synoptic scale movements. For this we employed the University of Wisconsin Nonhydrostatic Modeling System (UW-NMS), initialized from standard NCEP EDAS (Etad Data Assimilation System) analysis. Previous efforts to simulate such microscale phenomena have largely focused around so called Large Eddy Simulations. Those studies seek to understand the structure and evolution of small scale turbulence, and its connection to local topographical and land use structures or regional scale atmospheric structures. Since turbulence structures evolve over hours, and turbulent eddies have lifetimes of only a couple of minutes, LES seeks to simulate the statistics of turbulence and structure types, rather than deterministically predicting a particular weather state. Moreover, LES typically employs periodic boundary conditions so that turbulence can evolve as naturally as possible, without obvious boundary influences. In the present study, the motivation is not to simulate the evolution of turbulence, but to examine to what extent the complex behavior of microscale flow is simply a localization of a more easily predictable large scale pattern. Compared to mesoscale simulations, which often nest down to around 1 km resolution using two or three grids, this simulation uses a larger number of two-way nested grids in order to achieve resolutions in the range of 10s of meters. To localize the synoptic scale prediction, our simulation includes very high resolution shoreline geometry of the Wisconsin coastline where the observations were taken, as well as specific land-use and topographical features derived from high-resolution topographical databases in order to more accurately represent the Sheboygan area. A simple down-gradient turbulence closure is used to represent the smoothing effects of turbulence. With this simple setup, the non-hydrostatic model was able to simulate the observed local behavior of the land breeze front with striking accuracy. These results imply that a great deal of the variance of small scale flow behavior may actually feature large scale flow-like predictability, and also have very interesting implications on the merits of forecasting at the microscale in the 24-48 hour time frame.

Supplementary URL: