Abstract
The National Meteorological Center's 72-b spectral model forecasts for the 1983–84 cool season are examined in an appraisal of the model's ability to simulate 500-mb cyclones and anticyclones, defined by the existence of at least one (60-m interval) closed contour. Position and intensity errors we determined from comparison between forecast and observed 500-mb height fields. On the basis of this sample it is concluded that there is a tendency, with some geographical exceptions, for the model to overpredict the heights in these systems. This is particularly true of high latitude anticyclones. One noteworthy error characteristic in the model is a recurring failure to predict closed 500-mb cyclonic circulations which evolve from troughs crossing western North America. This suggests either initial data problems with troughs originating over the Pacific or that the model does not simulate the troughs' interaction with the Rocky Mountains very well on some occasions. Selected examples of model error are presented, accompanied by corresponding 48-h forecasts from the operational, limited-area, fine-mesh, primative equation model in one example. Computations of 850-mb geostrophic temperature advection from this latter model are compared with observed computations in this example and suggest that model errors in surface weather systems may he coupled with model errors in 500-mb systems, in this case through incorrect prediction of lower tropospheric temperature advection.
