Evaluation of limited-area models for the representation of the diurnal cycle and contrasting nights in CASES-99

This study evaluates the ability of three limited-area models [the fifth-generation Pennsylvania State University¿National Center for Atmospheric Research Mesoscale Model (MM5), the Coupled Ocean¿Atmosphere Mesoscale Prediction System (COAMPS), and the High-Resolution Limited-Area Model (HIRLAM)] to predict the diurnal cycle of the atmospheric boundary layer (ABL) during the Cooperative Atmosphere¿Surface Exchange Study (CASES-99) experimental campaign. Special attention is paid to the stable ABL. Limited-area model results for different ABL parameterizations and different radiation transfer parameterizations are compared with the in situ observations. Model forecasts were found to be sensitive to the choice of the ABL parameterization both during the day and at night. At night, forecasts are particularly sensitive to the radiation scheme. All three models underestimate the amplitude of the diurnal temperature cycle (DTR) and the near-surface wind speed. Furthermore, they overestimate the stable boundary layer height for windy conditions and underestimate the stratification of nighttime surface inversions. Favorable parameterizations for the stable boundary layer enable rapid surface cooling, and they have limited turbulent mixing. It was also found that a relatively large model domain is required to model the Great Plains low-level jet. A new scheme is implemented for the stable boundary layer in the Medium-Range Forecast Model (MRF). This scheme introduces a vegetation layer, a new formulation for the soil heat flux, and turbulent mixing based on the local scaling hypothesis. The new scheme improves the representation of surface temperature (especially for weak winds) and the stable boundary layer structure