Regionality and periodicity observed in rainfall variations of the baiu season over Japan

Abstract

AbstractThe regionality of rainfall variations of the Baiu, a rainy season over Japan, is investigated with special reference to periodicity. Using the Varimax rotated principal component analysis (PCA), Japan is divided into eight regions based on natural‐logarithm‐transformed rainfall data of June. In order to find significant rainfall variations from a sequence of rainfall data in each region, non‐rotated PCA is applied to the rainfall data for each region. Its first component score is defined as the rainfall variation index (RVI). Four regions show statistically significant periodicities of 10, 8, 4, or 2·2 years in rainfall variations. In addition, the rainfall variations reveal a similarity in the general shapes of spectral density for each region. The temporal variations of each periodicity show non‐stationary changes. This feature is especially distinct for periodicities at 4 and 2·2 years. From the results of squared coherence, it is pointed out that these periodicities are closely related to variations of the North Pacific High and the Okhotsk High. In the discussion, the validity of the regional division proposed in the present study is confirmed in comparison with other studies. Seasonal representativity of monthly rainfall variations is also discussed.