Principal sequence pattern analysis: a new approach to classifying the evolution of atmospheric systems

Abstract

A new eigentechnique approach, Principal Sequence Pattern Analysis (PSPA), is introduced for the analysis of spatial pattern sequence, as an extension of the traditional Principal Component Analysis set in the T-Mode. In this setting, the variables are sequences of k spatial fields of a given meteorological variable. PSPA is described and applied to a sample of 256 consecutive daily 1000 hPa geopotential height fields. The results of the application of the technique to 5-day sequences demonstrate the advantages of this procedure in identifying field pattern sequences, thereby allowing the determination of the evolution and development of the systems, together with cyclogenesis and anticyclogenesis processes.In order to complete the study, the more traditional Extended Empirical Orthogonal Function (EEOF) analysis, which is the S-mode equivalent of the PSPA, was applied to the same dataset. For EEOF, it was not possible to identify any real sequences that could correspond to the sequences of patterns yielded by the EEOF. Furthermore, the explained variance distribution in the EEOF was significantly different from that obtained with PSPA. Conversely, the PSPA approach allowed for the identification of the sequences corresponding to those sequences observed in the data. Using diagrams of the squares of the component loadings values, as a function of time, the study of the times of occurrence of dominant field characteristics was also possible. In other words, successful determination of periods where the basic flow was dominant and times when strongly perturbed transient events with a significant meridional component occurred, was facilitated by PSPA.

Spatio‐temporal rainfall patterns in Southern South America

“…Principal components and singular spectrum analysis Spatial and temporal patterns of climatic variability were examined using PCA of selected climatic series. The eigenvector analysis has been widely used to identify dominant patterns of climate variability and to reduce the dimensionality of climate data (Lorenz, 1956;Richman, 1986;White et al, 1991;Compagnucci et al, 2001). This analysis extracts orthogonal linear combinations of the series capturing the maximum proportion of initial variance in the data set.…”

Analysis of atmospheric conditions associated to CHACO events of the Low Level Jet East of the Andes and their implications for regional transport

“…A similar approach was applied by Compagnucci and Salles (1997), Salles et al (2001), Compagnucci et al (2001) and Compagnucci and Richman (2008). In this paper, the domain selected to apply the PCA stretches from 10 ∘ N to 60 ∘ S in latitude and from 1.25 ∘ W to 100 ∘ W in longitude and includes 5016 grid points.…”

Principal sequence pattern analysis of episodes of excess mortality due to heat in the Barcelona metropolitan area

“…The classification is applied to sea level pressure (SLP) anomalies, temperature anomalies at 850 hPa (T850) and geopotential anomalies at 500 hPa (Z500), for the JJA period. The data are shown with the grid points in the rows and the sequences in the columns (see Table 1), with a matrix configuration in T-Mode (Compagnucci et al 2001). The sequence, delimited for an n-day length, is identified by the sequence key (D), which is the last day in the sequence and is determined from those days when the daily mortality in the Barcelona metropolitan area exceeded the 95th percentile.…”

“…Analyses of the synoptic history can be undertaken using a multivariable classification of the synoptic sequences related to the main atmospheric parameters, with an hourly or daily resolution. Methodologically, this classification is supported as a variant of principal component analysis (PCA) and is known as principal sequence pattern analysis (PSPA: Compagnucci et al 2001;Escobar et al 2004;Jacobeit et al 2006;Esteban 2008;Philipp 2009;Aran et al 2011a).…”