Comparison of model predicted to observed winds in the coastal zone

Predictions of near-surface (10 to 100 m) wind velocities made by a mesoscale numerical model on a 10 km grid over and near the coastline are checked against observations. Two comparisons are made. The first is between observed and model-estimated mean annual wind power density at locations where surface observations exist in three coastal areas: the Chesapeake Bay, the Apalachee Bay and the South Texas coastal area. The second comparison is made between model predictions over the Delmarva Peninsula and adjacent ocean and observations made over a 120 x 30 km rectangle extending across the peninsula and out to sea. It is concluded that the unbiased error analysis skill ratings of 81% and 76% are attained for two days of prediction-observation comparisons. In the meantime, the skill of the model in duplicating individual coastal wind fields is taken as 78%. In addition, a qualitative comparison is made between the predicted fields of wind and the observed wind field. The predicted wind field unquestionably reproduces the observed field.