718 
Several diverse techniques of analysis exist, differing 
primarily in their method of attaining a verisimilar 
upper-level baric pattern over regions of sparse reports. 
Im such areas upper-level contours are not uniquely 
determined by the wind and height data now available. 
Thus some sort of systematic extrapolation is necessary. 
One such system is the technique of differential 
analysis. Approximately ten years ago the basic princi- 
ples underlying this analytical method were formulated 
in the United States by Rossby and Starr and put into 
practice in analysis by Willett and Namias in extended- 
range forecasting. During World War II Petterssen [16] 
and the meteorologists of the United Nations who 
collaborated with him in England fully developed this 
technique into a powerful analytical tool in conjunction 
with forecasting high-level winds for bombing oper- 
ations. Another form of differential analysis had previ- 
ously evolved in Germany under the leadership of 
Scherhag [20]. 
The German method of differential analysis differs 
from the others in that it is based upon the partition of 
the atmosphere into layers chosen in such a way that 
the thickness of any one layer is approximately equal 
to that of adjacent strata. Below 500 mb, Scherhag’s 
technique is virtually identical with Namias’ method, 
which will be treated in detail below. For the very high 
atmosphere, Scherhag has compiled statistical tables 
relating the thickness of each of his selected strata with 
the height above sea level of the lower limit of the 
layer. By noting the difference between the statistical 
values for the thickness of the various layers and the 
observed thicknesses, where radiosondes are operated, 
Scherhag is able to extend the application of differential 
analysis up to the high stratosphere. Herein lies the 
significance of Scherhag’s development. In this age of 
stratospheric aviation with its concomitant demands 
for wind forecasts at very high levels, Scherhag’s method 
constitutes a systematic technique for extrapolating 
data throughout the stratosphere to obtain a logical, 
internally consistent, baric pattern. 
For mid-tropospheric levels the best method devised 
for extrapolation of upper-air data is a combination of 
vertical extrapolation from individual reports, plus a 
horizontal distortion of normal thickness lines. This 
distortion must conform with all available reports at 
the level in question, the surface frontal configuration, 
the previously observed thickness pattern, the thermal 
winds, and the major atmospheric circulations at sea 
level and aloft. Such a method of differential analysis, 
described by Namias [12], is now used by the Weather 
Bureau for drawing the Northern Hemisphere daily 
charts for 700 mb and appears equally applicable to all 
levels below the tropopause. This technique makes use 
of monthly normal thickness maps, drawn at biweekly 
intervals on a transparent plastic material, which depict 
the normal thickness between the 1000- and 700-mb 
levels. 
The actual procedure as developed by Namias is to 
superimpose upon the completed surface map the ap- 
propriate normal thickness chart. On top of both is laid 
the 700-mb chart, plotted on a tissue paper base, since 
OBSERVATIONS AND ANALYSIS 
the translucent tissue paper allows the analyst to see 
the lines on the two charts beneath it. Finally, the 
700-mb chart is covered by a transparent plastic sheet 
on which the thickness lines for the current map are to 
be drawn. 
Before the analysis begins, the following data are 
plotted on the 700-mb chart: 
1. At all radiosonde stations—the 700-mb temper- 
ature, wind, and height, and the thickness from 1000 to 
700 mb; 
2. All available winds at 700 mb (lower levels if 
none are available for 700 mb); 
3. At ships located in a region where the meteorologi- 
cal situation permits an accurate estimation of the 
lapse rate—the estimated thickness between 1000 and 
700 mb; 
4. In regions of sparse data—the thermal winds be- 
tween gradient level and 700 mb. 
With these data available, the analysis begins. First, 
the analyst draws lines of constant thickness in the 
regions where data are scanty. (Such regions now 
include the greater part of the oceanic areas of the 
Northern Hemisphere, Siberia, and most of southern 
Asia.) This is done by distorting the normal thickness 
lines to fit all the upper-air data, the thermal winds, and 
the circulation patterns of the surface chart, with due 
regard to the continuity of the mean temperature field. 
An example of an observed thickness pattern and the 
associated surface frontal and pressure fields is pre- 
sented in Fig. 1. After the thickness lines are completed, 
the height of the 700-mb surface is determined by 
adding the thickness to the 1000-mb height. The latter 
is obtained by converting the sea-level pressure to the 
1000-mb height. In actual practice the analyst computes 
the height of the 700-mb surface at the intersections of 
latitude and longitude lines, using a ten-degree spacing 
of both longitude and latitude. Thus a grid of 700-mb 
heights is obtained. 
With this grid in the regions of infrequent reports, 
and with the radiosonde data elsewhere, the analyst 
can, with assurance, draw in the contours on the 700-mb 
chart. This method has the advantage of insuring 
horizontal as well as vertical consistency and combines 
expedition of analysis with maximum accuracy over 
land as well as over oceanic areas. 
At present, however, this method has one serious 
shortcoming: lack of sufficient data precludes the con- 
struction of normal thickness charts in the upper layers 
of the troposphere and in the stratosphere. There is thus 
an empirical ceiling on the efficacious use of this method. 
Above 500 mb, Scherhag’s system of extrapolating data 
for the construction of baric charts still is applicable, 
as is another method, based upon a study conducted by 
Riehl and LaSeur [19] of high tropospheric lapse rates. 
The general technique is little different from the long- 
standing practice of assuming a lapse rate at a pomt and 
thereby determining the pressures aloft at that point. 
For instance, the height of the 700-mb surface can be 
attained with reasonable accuracy by assuming a moist- 
adiabatic lapse rate up to 700 mb above a ship reporting 
a shower, but the choice of an appropriate lapse rate 
rh se mt 
