Abstract
I. One of the fundamental problems in climatology, the typical patterns of air-temperature was investigated applying the method of factor analysis as follows:
_??_
where the common fa, , tor is Fs (s=1, 2, …, m) and the unique factor Uj. Fs and Uj submit to N(0, 1) and are independent of eanh other. As data for the analysis, the mean monthly temperatures of January, April, August, and October for the period 1901 to 1940, at twenty-five weather stations in Japan were utilized. Initially, factor Fl was calculated and secondly, the factor F2, thereafter, finding the remaining correlation coefficients insigni-ficant, further calculations were abandoned. The coefficients of F1 and F2 for the four months are shown in Figures 1 to. 4. Generally, the coef-ficients of F1 are very large throughout Japan, showing values ranging between 0.7 and 0.9 while the coefficients of F2 are smaller and vary a great deal by districts. Analyzing the variances in the data by the three way lay-out, Table 1 was derived. The variance of coefficienss by lands (L) and by factors (F) are significant while that by seasons (S) is insignificant. The interaction term of L×F is significant, but those of LxS and F×S are insig-nificant.
II.hogically, the next step is a consideration of the kinds of common factors controlling air temperatures in the months under study. The first factor is one which has a great influence over a vast region, Judging from the fact that the coefficients of the first factor commonly show large values throughout Japan. It may be considered to be one as vital, as the general circulation of the atmosphere. As the second factor, may be considered a function of latitude, such as solar radiation, judging from the fact that the coefficients. of the second factor are arranged in order of latitude from north to south. The correlation coefficients between the coefficients of F2 and the solar radiation in three of the months are very large and significant, showing 0.834 in January, 0.923 in August, and 0.870 in October, while that for April shows only 0.362. From this fact., it appears that solar radiation is a primary factor in the distribution of air temperature.
III. From the results of the factor analysis of air temperature in April, Japan may be divided into two large areas, the north and the south (Figure V). In order to further divide these areas, factor analysis of air temperature was again attempted. Two subdivisions for each area were obtained and are shown in the figure as Areas ff 1 and If 2 in southern Japan and 11 and 12 in northern Japan. Tius Japan may be divided into four climatic divisions by the two factors F1 and F2 when it is considered that two common factors control the distribution of air temperature.
