312 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1956 



vary from one-half to 2 knots per day for the size storms that occur 

 most frequently in the Atlantic. He further estimated that for the 

 very large typhoons that sometimes occur in the Pacific the magnitude 

 of this acceleration might be as much as 8y 2 knots per day. Mr. Ando 

 in Japan made a statistical study and found that, on the average, 

 intense typhoons move faster to the north than weaker typhoons. 



Dr. Tu-Cheng Yeh [25] has described another internal mechanism 

 that results from the fact that superposition of a vortex on a steering 

 current is nonlinear. This fact becomes manifest through small oscil- 

 lations with amplitude of 0.5° to 2° of latitude and with a period of 

 12 hours to 2 days for the normal range of the observed winds, ac- 

 cording to Dr. Yeh. Yeh has developed formulas for computing both 

 the amplitude and the period of oscillation, and both of these quan- 

 tities vary with the wind speed, the size and intensity of the storm. 

 Post-storm analyses of many hurricane and typhoon tracks indicate 

 that there is some oscillation similar to the one which Yeh described. 

 Insufficient data have been available for fully testing the theories of 

 Dr. Yeh, and most of the use that has been made of his ideas in fore- 

 casting have been by extrapolating a suspected oscillation into the 

 future. 



COMMONLY USED FORECAST METHODS 



Statistical approach. — C. L. Mitchell [12] and I. E. Tannehill [23] 

 have prepared summaries of hurricane tracks by months which have 

 been published. Jose Colon [3] has extended the work started by these 

 two and has computed for each 5° square of latitude and longitude the 

 percent frequency storms have moved along each of the 16 principal 

 directions. He also calculated the probability of a displacement along 

 the modal direction and the average speed of movement (fig. 6) . When 

 data are sparse, frequently the best forecast that can be made is one 

 based on charts similar to those prepared by Colon. Fortunately, in 

 the tropical Atlantic area where we ordinarily have very little data, 

 the statistical approach seems to work very well. Even when data 

 are available the statistics can be used as a rough guide to what a hurri- 

 cane is most likely to do. 



Persistence of past movement. — Probably the most commonly used 

 method of forecasting hurricane movement is the so-called "path 

 method." This consists of extrapolating a hurricane movement into 

 the future on the basis of what has happened in the past 6, 12, 

 or 24 hours. Although extrapolation need not consist of projecting the 

 hurricane path forward along a straight line, that is the way it is used 

 most commonly. The "path method" integrates the effect of both 

 the internal and the external forces. If one may assume that the same 

 steering and internal forces will be acting in the next period, the "path 

 method" should give very good results. In this statistical work, Mr. 



