A Comparison Between Principal Component and Self-Organizing Map Analyses
| Johnson, Nathaniel | The Pennsylvania State University |
Category: Atmospheric State and Surface
This study evaluates the effectiveness of principal component and self-organizing map analyses as faithful representations of the corresponding input sea-level pressure patterns. Both analyses examine the same 51 winters of daily NCEP-NCAR sea-level pressure data centered over the North Atlantic. Input data are composited by score so that high-score observations are compared to the corresponding analysis pattern through the use of pattern correlations. The results of this evaluation reveal stronger correspondence for the self-organizing map analysis than for the principal component analysis. Because of these results, future work will utilize self-organizing maps rather than principal component analysis in order to link weather regimes to ARM data.
This poster will be displayed at the ARM Science Team Meeting.


