252 RADIO WAVE PROPAGATION EXPERIMENTS 
the time of year, geographical location, ete. 
3. Estimation, from (1) and (2) above, of the 
amount of trapping to be expected for specified radio 
frequencies and specified elevations of sites. 
Rercions CHOSEN For STUDY 
The area chosen for investigation was that bounded 
approximately by 10° and 50° N latitude and by 120° 
and 150° E longitude. These regions include the Japa- 
nese islands, the coasts of Korea, Manchuria, and 
China, northern Philippines, the Marianas, the 
Bonins, and the Ryukyu Islands. The computations 
were carried out for representatively selected 5x5 degree 
sectors or “squares.” The results for several regions 
in this area are summarized and are given below. 
Sources oF CuimaToLogicaL Data 
The charts and atlases used in compiling the data 
can be found in references 11 to 15. 
METHOD OF COMPUTATION 
The procedures described on pp. 246-251 were ap- 
plied as follows: 
1. A determination of the monthly mean tempera- 
ture excess was made for each 5x5 degree square for 
each month (1) by taking the difference between the 
mean temperature of the air over the sea and the mean 
sea surface temperature, and (2) by taking the differ- 
ence between the mean temperature of the air at a land 
station (when the given square was near a coast) and 
the mean sea surface temperature. 
2. A determination of the monthly mean M deficit* 
was made for each square for each month using the 
nomograms and taking the data for (1) the mean 
temperature and wet bulb depression of the air over 
the sea and the mean sea temperature, and (2) the 
mean temperature and relative humidity of the air 
at a land station (when the given square was near a 
coast) and the mean sea temperature. 
3. For each square for each month note was made 
of the various surface wind velocity ranges that oc- 
curred with each wind direction (eight points of the 
compass) and the percentage of time the wind lay 
within each velocity range. 
4. In terms of the mean temperature excess and 
mean M deficit, the corresponding duct width was 
computed for each wind velocity range, and, knowing 
kThese calculations do not represent exactly the correct 
mean value of the M deficit, since M is not a linear function 
of the temperature and humidity, and so the value of M com- 
puted from mean temperature and humidity data is not quite 
the same as the mean of all M’s computed from individual 
temperatures and humidities. A cursory evaluation of this 
error has indicated that the values computed are if anything 
conservative, i.e., that the actual mean M deficits are prob- 
ably larger than those computed. 
!This led to a slight error, inasmuch as the wind at 1,000 ft 
should have been used in place of the surface wind (data 
were available only for the latter). This error in most cases 
resulted in calculated duet widths of slightly less magnitude 
than would have been obtained if the 1,000-ft wind had 
been used. 
the percentage of time that winds of each magnitude 
occurred, it was possible to compute the percentage of 
time that ducts of various widths would occur, both 
for each wind direction (on an eight point compass) 
and for the overall picture regardless of wind direction. 
RESULTS OF COMPUTATIONS 
The results were summarized by lumping individual 
squares showing similar characteristics into nine re- 
gions within the whole area. In each region data for 
the individual months were lumped together on the 
basis of similarity to divide the year into three or four 
parts (these varying according to region). Then for 
each group of months in each region were listed those 
results which were considered to be the most impor- 
tant, namely, 
1. The range in duct widths, giving an indication 
of the variability at any one place at any one time 
of year. 
2. The percentage of time that ducts characterized 
by widths greater than 40 ft occur and, following this, 
the most prevalent wind direction associated with 
ducts of these widths, as well as the minimum wind 
velocity necessary to establish them. 
3. The percentage of time that ducts characterized 
by widths from 20 to 40 ft occur, similarly followed 
by the associated prevailing wind direction and the 
minimum required wind velocity. 
4. The percentage of time that ducts do not occur 
or have a width less than 20 ft. 
Figure 20 contains these summarized results. The 
numerical listings in the figure make no claim toward 
being exact, as is evident in view of the remarks made 
on pp. 249-251 on the applicability and limitations 
of the method, in addition to the slightly erroneous 
computations of the monthly mean M deficit and the 
use of the surface winds instead of those at 1,000 ft. 
(The effect of the latter two errors is mainly to cause 
the calculated duct widths to tend toward the conserva- 
tive side.) In Figure 20, for purposes of consistency 
throughout the area, only the calculations based on 
air temperature and humidity records over the sea 
were used [i.e., only (1) under “Method of Compu- 
tation”]. The difference between these calculations 
and those based on land station data is primarily that 
temperature excesses and M deficits are larger in the 
latter case (this again tends to make the tabulated 
results conservative). 
Other deficiencies of the method which might be 
cited are the neglect to take into account the occur- 
rence of fog or rain (which tends to create substandard 
or standard conditions) and the omission of the in- 
fluence of local effects such as the topography along 
coast lines (see pp- 254-256). In some cases the 
period of record was fairly short, so that the data 
used in the calculations were not always completely 
representative. Lastly, the ranges of duct widths were 
calculated on the assumption that there was a varia- 
