“類似示数”による型の分類と予想天気図の検証

Abstract

The analogy index proposed in this paper experesses objectively the degree of similarity in shape and value between two kinds of distribution patterns of a physical quantity. We define this index by the following equation;s I=f(γ, E)=γ(1-E/R) where I is the analogy index, γ and E are the correlation coefficient and the root mean square difference between two kinds of patterns respectively, and R is the range of the physical quantity which forms the patterns and is a proper constant value the dimension of which is the same as the root mean square difference.R is defined by R=nδ, δ=1/2(δx+δy), where δ× and δy are respectively the standard deviation of each pattern. As a result of the consideration, the most appropriate value of R is 4δ for giving the analogy index the characteristics of the correlation coefficient r and the root mean square difference E.The verifications of the analogy index are being carried out for the surface pressure patterns and the diurnal variations of visibility at Tokyo International Airport in the warm seasons of 1966 and 1967. The analogy index is far better than the root mean square difference and somewhat better than the correlation coefficient in the case of the surface pressure patterns. On the other hand, in the case of diurnal variations of visibility, the analogy index is far better than the correlation coefficient and has almost the same accuracy as the root mean square difference.We make practical applications of the analogy index as follows. In accordance with the method used by IA. Lund (1963) and K. Abe and E. Suzuki (1969), we classify the surface pressure patterns and get the same result as obtained through synopic meteorology. Then, after verification of the 24-hour prognostic surface map, we gain the result that the analogy index is more advantageous than the correlation coefficient in showing the similarity of the pressure values.