Forecast Accuracy with Optimum Vertical Model Truncation

Ferdinand Baer Department of Meteorology, University of Maryland, College Park, Maryland

Search for other papers by Ferdinand Baer in
Current site
Google Scholar
PubMed
Close
and
Yuejian Zhu General Sciences Corporation, Laurel, Maryland

Search for other papers by Yuejian Zhu in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The National Center for Atmospheric Research Community Climate Model 1 was used as an experimental prediction model to assess the value of reassigning model levels in the vertical based on an optimizing hypothesis. The model was considered for T31 horizontal truncation and 12 vertical levels. The levels were relocated in a model called test, and the model with the conventional levels was denoted standard. Both models were integrated for 5 days with six independent initial states, and the results were composited. Analyses of the composites for both models were compared to actual observations. The results of the experiments indicate that the barotropic component of the flow was predicted with equal quality by both models but that the baroclinic component was predicted better by the test model. This observation may be explained by the increased fidelity of the vertical structure in the test model, since it has more resolution in the stratosphere.

Additional analyses were performed using a hypothesized three-dimensional scale index that relates the vertical to the horizontal truncation. The results of those analyses were sufficiently suggestive to encourage further studies to find optimum truncation in all three dimensions simultaneously.

Abstract

The National Center for Atmospheric Research Community Climate Model 1 was used as an experimental prediction model to assess the value of reassigning model levels in the vertical based on an optimizing hypothesis. The model was considered for T31 horizontal truncation and 12 vertical levels. The levels were relocated in a model called test, and the model with the conventional levels was denoted standard. Both models were integrated for 5 days with six independent initial states, and the results were composited. Analyses of the composites for both models were compared to actual observations. The results of the experiments indicate that the barotropic component of the flow was predicted with equal quality by both models but that the baroclinic component was predicted better by the test model. This observation may be explained by the increased fidelity of the vertical structure in the test model, since it has more resolution in the stratosphere.

Additional analyses were performed using a hypothesized three-dimensional scale index that relates the vertical to the horizontal truncation. The results of those analyses were sufficiently suggestive to encourage further studies to find optimum truncation in all three dimensions simultaneously.

Save