INVESTIGADORES
BETTOLLI Maria Laura
artículos
Título:
ASPECTOS METODOLÓGICOS DEL USO DEL ANÁLISIS DE COMPONENTES PRINCIPALES EN CAMPOS DE ANOMALIAS DE ALTURA GEOPOTENCIAL EN EL SUR DE SUDAMERICA
Autor/es:
BETTOLLI, MARIA LAURA; VARGAS, WALTER M; PENALBA, OLGA C
Revista:
REVISTA BRASILEIRA DE METEOROLOGIA
Editorial:
Sociedad Brasileira de Meteorologia
Referencias:
Año: 2007 vol. 22 p. 273 - 285
ISSN:
0102-7786
Resumen:
RESUMEN
En este trabajo se discutieron aspectos metodológicos del uso del Análisis de Componentes Principales
(ACP) a fin de verificar la eficacia del método para el conjunto de campos diarios de anomalías de
altura geopotencial de 1000 y 500 hPa de los reanálisis 2 del NCEP en el sur de Sudamérica de los
años 1979-2001 en el período comprendido entre octubre y mayo. Este análisis metodológico se realizó
con el objetivo de obtener una base representativa de la información para clasificaciones sinópticas
futuras. Se evaluó la necesidad de realizar rotaciones ortogonales y su efectividad, encontrando que
éstas no mejoran la redistribución de la varianza explicada. A partir de una submuestra de 100 días
elegidos al azar se analizó la in. uencia del efecto de la persistencia de los campos diarios consecutivos.
Dicho efecto no modificó sustancialmente los resultados del ACP y no introdujo otras componentes
principales significativas. El conjunto de datos de anomalías de altura geopotencial puede reducirse
efectivamente realizando un ACP como técnica de síntesis. Este resultado se corroboró utilizando el
test de esfericidad de Bartlett y una prueba empírica modificando la matriz de correlación. Finalmente,
para representar un campo de anomalía real y, de esta manera, facilitarle un sentido físico es necesaria
la combinación de dos o más CP.
Palabras claves: patrones sinópticos; altura geopotencial; sur de Sudamérica; análisis de componentes
principales
principales
principales
principales
patrones sinópticos; altura geopotencial; sur de Sudamérica; análisis de componentes
principales
ABSTRACT: METHODOLOGICAL ASPECTS OF THE USE OF THE PRINCIPAL COMPONENT
ANALYSIS OF 1000 AND 500 hPa ANOMALIES OF GEOPOTENTIAL HEIGHT FIELDS IN
SOUTHERN SOUTH AMERICA.
This paper presents a discussion of methodological aspects of the use of the principal component
analysis (PCA) to verify its eficacy synthesis for the daily 1000 and 500 hPa anomalies of geopotential
height fields of the October to May 1979-2001 NCEP reanalyses II in southern South America. This
analysis was performed in order to obtain a representative data base for future synoptic classifications.
The need of orthogonal rotations and their eficacy was analyzed showing that the explained variance
redistribution was not improved. A random sub-sample of 100 days was used to study the influence of
the persistence effect of consecutive daily fields. This effect did not significantly modify the PCA results
and did not introduce other significant PCs. The set of geopotential height data may be considerably
reduced using PCA as a synthesis technique. This result was corroborated using Bartletts sphericity
test and an empirical test on the modified correlation matrix. Finally, to represent a real field and thus
facilitate a physical sense, it is found that the combination of two or more PCs is necessary.
ANALYSIS OF 1000 AND 500 hPa ANOMALIES OF GEOPOTENTIAL HEIGHT FIELDS IN
SOUTHERN SOUTH AMERICA.
This paper presents a discussion of methodological aspects of the use of the principal component
analysis (PCA) to verify its eficacy synthesis for the daily 1000 and 500 hPa anomalies of geopotential
height fields of the October to May 1979-2001 NCEP reanalyses II in southern South America. This
analysis was performed in order to obtain a representative data base for future synoptic classifications.
The need of orthogonal rotations and their eficacy was analyzed showing that the explained variance
redistribution was not improved. A random sub-sample of 100 days was used to study the influence of
the persistence effect of consecutive daily fields. This effect did not significantly modify the PCA results
and did not introduce other significant PCs. The set of geopotential height data may be considerably
reduced using PCA as a synthesis technique. This result was corroborated using Bartletts sphericity
test and an empirical test on the modified correlation matrix. Finally, to represent a real field and thus
facilitate a physical sense, it is found that the combination of two or more PCs is necessary.
ANALYSIS OF 1000 AND 500 hPa ANOMALIES OF GEOPOTENTIAL HEIGHT FIELDS IN
SOUTHERN SOUTH AMERICA.
This paper presents a discussion of methodological aspects of the use of the principal component
analysis (PCA) to verify its eficacy synthesis for the daily 1000 and 500 hPa anomalies of geopotential
height fields of the October to May 1979-2001 NCEP reanalyses II in southern South America. This
analysis was performed in order to obtain a representative data base for future synoptic classifications.
The need of orthogonal rotations and their eficacy was analyzed showing that the explained variance
redistribution was not improved. A random sub-sample of 100 days was used to study the influence of
the persistence effect of consecutive daily fields. This effect did not significantly modify the PCA results
and did not introduce other significant PCs. The set of geopotential height data may be considerably
reduced using PCA as a synthesis technique. This result was corroborated using Bartletts sphericity
test and an empirical test on the modified correlation matrix. Finally, to represent a real field and thus
facilitate a physical sense, it is found that the combination of two or more PCs is necessary.
ANALYSIS OF 1000 AND 500 hPa ANOMALIES OF GEOPOTENTIAL HEIGHT FIELDS IN
SOUTHERN SOUTH AMERICA.
This paper presents a discussion of methodological aspects of the use of the principal component
analysis (PCA) to verify its eficacy synthesis for the daily 1000 and 500 hPa anomalies of geopotential
height fields of the October to May 1979-2001 NCEP reanalyses II in southern South America. This
analysis was performed in order to obtain a representative data base for future synoptic classifications.
The need of orthogonal rotations and their eficacy was analyzed showing that the explained variance
redistribution was not improved. A random sub-sample of 100 days was used to study the influence of
the persistence effect of consecutive daily fields. This effect did not significantly modify the PCA results
and did not introduce other significant PCs. The set of geopotential height data may be considerably
reduced using PCA as a synthesis technique. This result was corroborated using Bartletts sphericity
test and an empirical test on the modified correlation matrix. Finally, to represent a real field and thus
facilitate a physical sense, it is found that the combination of two or more PCs is necessary.
METHODOLOGICAL ASPECTS OF THE USE OF THE PRINCIPAL COMPONENT
ANALYSIS OF 1000 AND 500 hPa ANOMALIES OF GEOPOTENTIAL HEIGHT FIELDS IN
SOUTHERN SOUTH AMERICA.
This paper presents a discussion of methodological aspects of the use of the principal component
analysis (PCA) to verify its eficacy synthesis for the daily 1000 and 500 hPa anomalies of geopotential
height fields of the October to May 1979-2001 NCEP reanalyses II in southern South America. This
analysis was performed in order to obtain a representative data base for future synoptic classifications.
The need of orthogonal rotations and their eficacy was analyzed showing that the explained variance
redistribution was not improved. A random sub-sample of 100 days was used to study the influence of
the persistence effect of consecutive daily fields. This effect did not significantly modify the PCA results
and did not introduce other significant PCs. The set of geopotential height data may be considerably
reduced using PCA as a synthesis technique. This result was corroborated using Bartletts sphericity
test and an empirical test on the modified correlation matrix. Finally, to represent a real field and thus
facilitate a physical sense, it is found that the combination of two or more PCs is necessary.
Keywords: synoptic pattern; geopotential height; southern South America; principal component
analysis.
analysis.
analysis.
analysis.
synoptic pattern; geopotential height; southern South America; principal component
analysis.