4. World Maps of aN 



4.1. Development 



Twelve maps of monthly mean AN were pre- 

 pared from the 5-year mean values obtained by 

 interpolation of mandatory pressure level radio- 

 sonde data for the 268 stations listed in table 

 B-l and located in figure B-l in appendix B. 

 These maps, contoured in the same manner as 

 the N(z) maps (sec. 3.3), are figures B-2 

 through B-13. 



Previous work has shown that a good corre- 

 lation may exist between aN and surface N on 

 a monthly mean basis [Bean and Thayer, 1959 ; 

 Bean and Cahoon, 1961b; CCIR, 1965]. In fact, 

 in many areas of the world this seasonal corre- 

 lation is very high, and in such areas a regres- 

 sion line might provide better estimates of 

 monthly mean AN than the maps in appendix B 

 (if the mean value of surface N were available 

 for that particular month of that year for the 

 desired location). This regression line could 

 also be used with the N s distribution data from 

 table A-l to provide estimates of the aN distri- 

 bution. Correlations were therefore calculated 

 for the 12 monthly mean values of AN and sur- 

 face N for each of the stations in the sample. 

 The equations resulting from these calculations 

 were put into the form of deviations from the 

 annual means : 



JN=b(N s -N s ) + W ±S.E., (5) 



where N s is the surface value of N, the single 

 bar represents a monthly mean value, the dou- 

 ble bar represents the annual mean, b is the 

 slope of the regression line, and S.E. is the 

 standard error of estimate. The equations were 

 put in this form because the intercept of the 



equations in the ordinary form is equal to AN - 



bN s , which is too unwieldy for contouring on 

 maps. Maps, which appear in appendix B, were 

 prepared of the slope (b) , the annual means 



(aA t and N ) , and the standard error of pre- 

 diction and correlation coefficient of the regres- 

 sion lines. 5 



4.2. Discussion of Contours and Reliability 

 of A/V~ Maps 



The world aN maps of this atlas do not show 

 as much detail as may be found in other publi- 

 cations which consider only specific areas [Bean 



5 N is an approximate sea-level value of N a , defined by the equa- 

 tion No = N s e"- lz , where z is the elevation above sea level in km. 



et al., 1960b; du Castel, 1961; Rydgren, 1963; 

 CCIR, 1963]. It was necessary in this study to 

 omit some of the available radiosonde data in 

 areas with relatively dense weather networks 

 (e.g., the U.S.A. and Europe) in order to obtain 

 a more nearly uniform worldwide coverage. 

 Even with this coverage, the map scale size pre- 

 cluded the contouring detail which would be 

 necessary if localized terrain effects were to be 

 included ; for example, mountainous locations 

 (higher than 1 km) probably have lower values 

 of AN than those indicated in figures B-2 

 through B-13. Some dissimilarity in the con- 

 tour patterns between the maps in appendix B 

 and other aN maps may also be found because 

 of the differences in the time period used in the 

 samples; such disagreements emphasize the 

 fact that 5 years of data are not adequate for 

 reliable means in many areas. 



The map contours of worldwide AN indicate 

 that: 



(1) Low values of aA t are characteristic of: 



(a) large desert and steppe regions such as 

 the Sahara, the Australian interior, the south- 

 western U.S.A., and the Asian region southeast 

 of the Caspian Sea all year ; 



(b) high plateau areas during all seasons 

 except winter. 



(2) High values of AN are found in: 



(a) all areas where large masses of subsid- 

 ing air prevent the normal diffusion of water 

 vapor, creating an unusually large A'-gradient 

 between the moist surface air and the very dry 

 air at 1 km. Specific examples are : 



(1) continental west coasts at latitudes 

 20°- 35° in the summer hemispher e an d 10°- 

 25° in the winter. In fact, the true aN may be 

 higher than indicated on the maps at locations 

 such as Dakar, Senegal, where a very thick 

 (~250 m) surface or near-surface ducting 

 layer occurs much of the time ; 



(2) tropical ocean areas where a trade- 

 wind inversion leads to a persistent elevated 

 ducting layer below 1 km. [Note : In a few cases 

 where the entire thickness of an elevated layer 

 lies between 1 km and the height of the 850-mb 

 level, the interpolation method gives a map 

 value which may be 5 to 10 A T -units too high.] 



(b) southeast Arabia and the Gulf of Per- 

 sia during July and August, when orographic 

 subsidence traps moisture from the southwest 

 monsoon in the gulf and lowlands between the 

 mountains. 



10 



