Abstract
A three-dimensional primitive model was used to predict the early afternoon wind velocity field over St. Louis, Missouri. Four case studies were then performed where model results at various levels were compared with observed data from the METROMEX network. With proper initialization, the model very closely simulated actual conditions, the main feature of which was the urban heat island circulation. In-consistencies were due mainly to synoptic-scale changes and air mass anomalies, which the model is not designed to handle, and to limitations in the data associated with the network density.
