In this paper the author treats theoretically how the essential feature of the atmospheric diffusion appears to be remarkably dependent upon the duration of time T* during which the phenomenon is observed. T* is called the averaging time. This problem is considered in view of extending the author's previous study, in which only the case of T*=∞ was discussed. The most important difference between the present and previous considerations is that in the present one, the eddies smaller than those of a certain scale related to the averaging time only are considered to be relatively effective for the turbulent diffusion, while in the latter all eddies were considered to be so irrespective of their scale.
Considering that many atmospheric phenomena can well be regarded to be nearly two-dimensional, the author theoretically deals with the two-dimensionally isotropic turbulent field. The structure of the field is investigated in close connection with that of the three-dimensionally isotropic turbulent field. Taylor-Kármán's correlation coefficient for the former is found to be approximately the same with that for the latter in their numerical values, though their functional forms are considerably different from each other. The calculated numerical values of the coffiicient agree well with those actually observed in the atmosphere.
In this paper, the author analyzed the 10-day period of air pressure often found in Kanto District and its neighborhood from April to June, and from September to October, and the following results were obtained: 1. Periodic change most develops in the middle decade of July and the last decade of September. (Fig. 1) 2. Spatial distribution of periodic change is given. (Fig. 2) 3. Statistically, the center of periodic change situates in Kanto District. 4. The above change is considered to be a stationary oscillation and accompanies the temperature change as shown in Fig. 5.