PRESENTATION OF BASIC DATA 



95 



Figure 4.4. Mean No, August 0200 local time. 



No. The advantage of adopting No is illustrated by the reduction in 

 range from 190 A^ units for iV^ to 115 A^ units for No, thus diminishing 

 the number of contours of the resulting maps. 



It would appear that by removing the influence of station elevation it 

 would be more efficient to estimate A''^ from A^^o charts rather than from 

 charts of A''^. As a test of this hypothesis, N s and A^'o contour charts were 

 prepared for both summer and winter from only 42 of the 62 U.S. Weather 

 Bureau stations for which 8-year means of N s are available. The remain- 

 ing 20 stations, distributed at random about the country, were used as a 

 test sample by estimating their 8-year mean value of N s from the A'^o and 

 N s contours. Summertime examples of these charts are given by figures 

 4.6 and 4.7. Note that due to the reduced range of N, the A^'o charts are 

 drawn every 5 A'^ units as compared to the 10-A^-unit contours of the N s 

 charts. The individual deviations of the values obtained from the con- 

 tour maps with the actual 8-year means are listed in table 4.2. By com- 

 paring the root mean square (rms) deviations of 10.7 A'^ units in winter 

 and 13.0 N units in summer obtained by estimating N s from the N s con- 

 tours with the 2.7-A^-unit rms of estimating N s from A'^o contours, one 

 concludes that it is at least 4 times more accurate to estimate N s from A''o 

 contours than from those of A^^. An inspection of the individual devia- 

 tions in table 4.2 indicates that the A^o contour method is particularly 

 efficient at elevations in excess of 1,200 m or where the terrain is changing 



