Comparison of principal component and cluster analysis for classifying circulation pattern sequences for the European domain

The study examines two methods for classification of daily pressure-pattern sequences (extended PCA and extended cluster analysis) and evaluates whether sequence classification is more suitable to describe surface air-temperature conditions in Europe. For this purpose, sequences of daily sea-level pressure fields are classified for each month of the year using a wide range of class numbers (2–40) and sequence lengths from 1 to 6 days. In each case, the classification is used to reconstruct temperature variability in 5°×5° grid-boxes throughout Europe for the period 1850–2003 in order to compare the circulation type based reconstruction to monthly observations for skill estimation. Results show that extended PCA leads to more dynamic pattern sequences for subordinated classes, but also that it is limited in skill, because classes of very low frequencies are generated. In contrast, extended cluster analysis offers higher skill levels for temperature downscaling. However, both methods can benefit from using sequences instead of single days for classification, especially for continental regions.

[1]  Radan Huth,et al.  AN INTERCOMPARISON OF COMPUTER‐ASSISTED CIRCULATION CLASSIFICATION METHODS , 1996 .

[2]  A. Sterl,et al.  The ERA‐40 re‐analysis , 2005 .

[3]  Manola Brunet,et al.  Daily Mean Sea Level Pressure Reconstructions for the European–North Atlantic Region for the Period 1850–2003 , 2006 .

[4]  R. Compagnucci,et al.  Principal sequence pattern analysis: a new approach to classifying the evolution of atmospheric systems , 2001 .

[5]  Radan Huth,et al.  Properties of the circulation classification scheme based on the rotated principal component analysis , 1996 .

[6]  H. Wanner,et al.  Summer heat waves over western Europe 1880–2003, their relationship to large-scale forcings and predictability , 2007 .

[7]  P. Jones,et al.  Uncertainty estimates in regional and global observed temperature changes: A new data set from 1850 , 2006 .

[8]  Jucundus Jacobeit,et al.  Frequency and within‐type variations of large‐scale circulation types and their effects on low‐frequency climate variability in central europe since 1780 , 2007 .

[9]  P. Jones,et al.  Long-Term Variability of Daily North Atlantic–European Pressure Patterns since 1850 Classified by Simulated Annealing Clustering , 2007 .

[10]  W. Enke,et al.  A novel scheme to derive optimized circulation pattern classifications for downscaling and forecast purposes , 2005 .

[11]  R. Preisendorfer,et al.  Principal Component Analysis in Meteorology and Oceanography , 1988 .

[12]  L. Kalkstein,et al.  An Evaluation of Three Clustering Procedures for Use in Synoptic Climatological Classification , 1987 .