| Summary: | Thesis (Ph. D.)--University of Washington, 1993 Low-frequency variability in the Northern Hemisphere wintertime 500 hPa height field has been investigated. The Pacific/North American pattern and the North Atlantic Oscillation stand out above the background continuum of variability in rotated EOF analysis based on the temporal covariance matrix. It is shown that these spatial patterns are much less sensitive to sampling fluctuations than the corresponding ones derived from the temporal correlation matrix and the rotated principal components are less sensitive than the EOFs on which they are based.Four distinctive clusters are identified based on a series of cluster analyses on perturbed data sets created by withholding from the 44 winter dataset the maps belonging to each individual winter: one cluster is marked by a closed anticyclone over the southern tip of Greenland, the others by ridges over the Gulf of Alaska, the Rockies, and western Europe, respectively. The primary anomaly centers of these clusters coincide with the the regions of large variance and strong positive skewness of the 500 hPa height field. The first three clusters can be reconstructed remarkably well by linear combinations of the two leading EOFs of the 500 hPa height field. They are related to the prominent features in the probability density function derived from the coefficients of these EOFs.Retrogression of blocking signatures appears in evolution of the clusters with positive primary anomaly centers located in high latitudes whereas the other clusters are characterized by in situ growth and decay of anomaly centers without significant phase propagation. In contrast to previous studies, we find that the hemispherically averaged root-mean-squared height tendencies associated with these clusters are not particularly small and there are no significant preferred transitions between these clusters.The interaction between the troposphere and the lower stratosphere during wintertime is documented using compositing based on the clusters in ...
|
|---|