Abstract
As the first step in a series of statistical downscaling studies, an objective classification of the sea level pressure (SLP) distribution pattern in the East Asian region was conducted. Patterns classified by this method for the cold half of the year were characterized by referring to their group mean distribution patterns, prevailing seasons, and those obtained by traditional weather classification methods. On the basis of the time series of patterns obtained, the continuity of each pattern and transition between patterns were also investigated quantitatively depending on the probability. This method is based on multivariate analysis, in which a large number of parameters from the SLP field (distribution pattern) are integrated into a limited number of objective indices, and the distribution patterns are classified depending on the indices. First, SLP data at 00UTC for a period of 30 years (1979–2008) from the NCEP/NCAR dataset were analyzed using principal component analysis (PCA). Then, 10,958 days (distribution patterns) were classified into groups by cluster analysis in the six-dimensional space with the axes of the first six component scores of the PCA. Winter SLP distribution patterns in the East Asian region were classified into 12 groups (patterns). Each pattern was named according to the feature clarified. These patterns were mainly characterized as a strong winter pattern (three groups: Siberian high type, Aleutian low type, mixed type), weak winter pattern (three groups), cyclone pattern, anticyclone/cyclone pattern (two groups), anticyclone pattern (two groups), and others. Our method classified the strong winter pattern into three groups from the meteorological point of view. Through the use of the additionally calculated six-hourly time series of the patterns, the temporal continuity of each group was investigated. The group to which specific groups tend to shift was also clarified quantitatively. In the interannual variation of the frequency of each group, no increasing or decreasing trend was found, especially in those of the strong winter patterns. Our method enables the objective classification of the SLP distribution patterns according to the quantitative value. This will be helpful in evaluating climate change quantitatively from the frequency and intensity of each pattern when our method is applied to future SLP distribution patterns in the globally warming world.
