Observations and mesoscale-model forecasts of low-level winds for the Salt Lake Valley, Utah region during an approximately 100-day period in winter 2002 are used to assess the importance of diurnally forced, land-atmosphere interactions to the regional climatology. For each observation location with a reasonably continuous sequence of wind data, the time series of observations and model simulations were spectrally decomposed in order to evaluate the spectral energy in each of three bands: the diurnally forced component with periods from 22-26 h, the super-diurnal component (period > 26 h) that contains the synoptic-scale motions with period of a few days, and the sub-diurnal component (period < 22 h) that contains mesoscale eddies that do not result from local forcing. The super-diurnal motions are assumed to be reasonably well predicted by most models because the processes can be generally resolved by the synoptic-scale observing network, and therefore represented in the initial conditions. The sub-diurnal motions will be poorly simulated by even mesoscale models because the circulations are too small to be well represented in three dimensions by the observing network. Lastly, the diurnal circulations should be predictable by mesoscale models with good physical representations of diurnally forced land-atmosphere interactions.

Spectral analyses of the data show that there is much station-to-station variation in the amount of energy in the diurnal band. For virtually all observation locations, however, there was considerably less spectral energy in the diurnal band than in the other two bands combined, even in this geographic region with complex local diurnal forcing, where wind-direction climatologies at some locations are dominated by thermal orographic effects. Simulations from two different models (MM5 and RUC-2) for this period show great differences in terms of the models’ abilities to represent the amount of energy in these bands.