An Exploration on the Applicability of Similarity Parameter in Similarity Forecasting

According to the previous various forecasting verification results concerning similarity measurements and starting from formulas,the fact that there exist the same features between similarity parameter and Hamming distance is analyzed and then proved.Meanwhile,the limitations of similarity parameter in similarity forecasting are discussed.The data from 96 different stations at 850 hPa height fields in East Asia during 1-30 May 2010 are utilized to make selection experiments among several frequently used similarity measurements and a new one is proposed by the authors in this paper.The results indicate that,(1) similarity parameter and Hamming distance are very much similar,with over 80%selected samples being the same;(2) similarity parameter has the biggest difference from similarity coefficient,with over 70%selected samples being different;and(3) the new analog quantity is more similar to the correlation coefficient, with 60%being the same.Similarity parameter and Hamming distance reflect how much the distances of the samples are similar to each other,while correlation coefficient and the new analog quantity reflect how much the shapes of the selected samples are similar to each other.