712 
measured continuously in time at the surface, are not 
even transmitted over the present communication sys- 
tem. 
Improved Techniques 
Some new techniques and station models which were 
designed to eliminate most of the shortcomings listed 
above have been described [1, 2, 3]. Following is a 
discussion of the basic concepts of these new techniques 
as they are related to the previous considerations. 
It is apparent that some means of automatically 
plotting the observations in final form is required to 
eliminate large plotting staffs. From an equipment- 
design point of view it is in general more convenient to 
represent the observations by plotting graphs auto- 
matically, rather than by printing numbers, on the 
appropriate maps or charts. ‘ 
Facsimile communication equipment is already in 
use for automatically plotting some of the observations 
(a few thermodynamic diagrams) in graphical form. 
However, some serious limitations to the transmission 
of observations by facsimile are: 
1. High carrier frequencies are required, thus limit- 
ing the usable types of transmission lines and compli- 
cating or eliminating the processes of storing, editing, 
and routing in the communication system. 
2. All observations must be collected, probably by 
means other than facsimile, plotted in a standardized 
form, and retransmitted. 
3. All automatic plotting is limited to a standardized 
form for transmission so that the individual analyst’ 
has little or no possibility of selecting those charts, 
parameters, or particular values which are best suited 
for his individual purposes. 
A direct-writer type of communication system seems 
to be ideally suited to the transmission of synoptic 
observations. In this system, signals representing the 
coordinates of any desired graph are transmitted. These 
signals cause a pen to draw the graph, or, if desired, 
digital numbers corresponding to the desired observa- 
tion at discrete points can be printed. Such a system 
has none of the limitations listed for the facsimile system, 
and it could also be used at least as efficiently as fac- 
simile for the transmission of analysis of weather condi- 
tions. No such direet-writer system is as yet sufficiently 
develéped for universal use, but there are no apparent 
serious technical difficulties to be overcome in its de- 
velopment. 
In order to reduce the number of sheets of paper 
required for complete representations, station models 
which use the least space should be adopted. In this 
respect it is advantageous to use graphs, rather than 
digital numbers, since in a graph but one point is re- 
quired to represent a given observation. This economy 
of space is best realized by choosing parameters for the 
description of the observations so that a minimum of 
grid, or reference, lines are required for their interpre- 
tation. This elimimation of grid lines prevents confusion 
because of their overlapping when graphs for different 
stations or times are plotted close together. 
Three-dimensional representations of the observa- 
OBSERVATIONS AND ANALYSIS 
tions can be obtained with graphs arranged according 
to isometric drawing principles. For example, the height 
scales of all graphs can be drawn parallel to each other 
with their origins at points on a map corresponding to 
the positions of the observing stations. They can then 
be viewed as, say, telephone poles sticking up into the 
air. The values of observations at any given height 
can be represented by lines drawn from the graphs of 
the observations to the height scale. A three-dimen- 
sional illusion is then obtained by considering the lines 
to be crossbars on the telephone poles, with lengths 
proportional to the values of the observations. 
This isometric technique can also be used to provide 
useful representations of the continuous time variations 
of surface observations throughout the region con- 
sidered. Parallel time scales, with one particular time 
at the geographical positions of the stations, can be 
visualized as lying along the ground. 
By placing both the time and height graphs on the 
same map, all the observations of a particular param- 
eter or group of parameters made in a given region 
and time interval can be plotted on one piece of paper. 
Examples of this technique [8] indicate that it can 
materially aid an analyst in ascertaining the three- 
dimensional distribution of conditions as well as the 
surface time variations of these conditions. These ex- 
amples also indicate that much analysis in the form of 
the drawing of isopleths can be eliminated when only a 
general concept of the spatial distribution of a given 
parameter is desired. 
Graphs drawn according to these isometric principles 
have the useful property that the relative geographical 
positions of the observations are precisely maintained. 
This is seen from the fact that a given value of an 
observation at a given time or height is represented by a 
point which is displaced from the origin by the same 
amount at each station. This property permits con- 
venient analysis of almost any desired conditions. For 
example, the height of and conditions on any desired 
constant-pressure surface, isentropic surface, version, 
front, etc., are readily available. Similarly, the condi- 
tions in any desired vertical cross section are repre- 
sented on the isometric maps. 
Additional advantages can be obtained by using 
special parameters for describing the observations in 
terms of dynamic models. Some parameters of this 
type are described below. 
The Hydrostatic Equation. Useful continuous repre- 
sentations of the pressure-height relationship in the 
vertical can be obtained by using parameters defined 
in terms of deviations from arbitrary standard condi- 
tions [1]. These definitions amount to defining a stand- 
ard column of air for use as a barometer. The parameter 
for describing the intensity of pressure then becomes 
the height z, at which that pressure occurs in the 
standard atmosphere. Heights at which given pressures 
occur in the actual atmosphere can then be described 
either by the mean sea-level height z, or by the devia- 
tion from the standard height D, defined by 
(1) 
Di = 2 — zp. 
