NO. 9 FORECAST OF U. S. PRECIPITATION — ABBOT 57 



entered at the bottom of the columns of A in table 14. These averages 

 will be needful to the use of table 15 which is to follow. 



Assuming that the degree of success attained in the forecasts from 

 1950 through 1958 will be attained from 1959 through 1967, I have 

 prepared table 15 from which the probable sizes and numbers of 

 discrepancies between forecasts and events in 4-month mean values 

 over the entire interval of 9 years, 1959-1967, may be estimated. 

 Selected from table 14, four groups of cities, 25 in all, are tabulated 

 in table 15. The first group of 11 cities have average 4-month mean 

 discrepancies, 1950-1958, of about 20 percent between forecasts and 

 events. 



The second group of six cities have mean 4-month discrepancies 

 of about 26 percent, the third group of five cities, 30 percent, and the 

 fourth group of three cities, 40 percent. All the percentages relate to 

 normal precipitation given in table 9, with the scale correetions from 

 table 9a used in table 14. 



The six columns of table 15 give, respectively, the numbers of cases 

 in table 14 when the discrepancies between forecast and event, 1950- 

 1958, are (a) less than one-fourth, (b) one-fourth to one-half, (c) 

 one-half to one times, (d) one to one and one-half, (e) one and one- 

 half to two, and (f) over two times the average discrepancy of the 

 group. 



If the same degree of success is reached 1959- 1967 as was reached 

 1950-1958, the interested person of a city in Group 1 would expect 

 the numbers of discrepancies (O-F) among the 4-month means stated 

 in the mean values at the bottom of the columns of table 15 to occur 

 in the entire interval of 9 years with magnitudes in percent of the 

 normals as stated at the top of the columns of the first group. If he 

 were located at a city of Group 4, the percentages would be twice 

 as large, because the numbers heading Group 4 are twice those head- 

 ing Group 1. But the numbers of cases would be the same. 



Stated numerically, a person residing where the mean departure of 

 forecast from observation, given in table 14 for 4-month intervals 

 from 1950 through 1958, was about 20 percent of normal precipita- 

 tion, may expect the following numbers and magnitudes of departure 

 from the forecast of 4-month means during the entire 9 years, 1959- 

 1967, given in table 14. 



Numbers of departures 4.6 4.5 6.1 6.0 2.8 3.0 



Magnitudes in percent 0-5 6-10 11-20 21-30 3 T_ 40 >40 



If he resided where the mean departure given in table 14 was greater, 

 the numbers of departures as just given would be unchanged, but f 



f Text continued on page 67. 



