480 
polynomial interpolation, and to calculate the initial 
data at the points of a predetermined grid. Given the 
mechanical means for the accomplishment of this pro- 
130__ 11090 
rae, 
aN 
SS 
DYNAMICS OF THE ATMOSPHERE 
applies only indifferently, if at all, to many of the small- 
scale but meteorologically significant motions. We have 
merely indicated two obstacles that stand in the way 
Z 
ay 
al 
C 
Fie. 3.—Forecast of January 30, 1949, 0300 GMT: (a) observed z (heavy lines) and ¢ + f (light lines) at t = 0; (6) observed 
zand ¢ + f at ¢ = 24 hr; (c) observed (continuous lines) and computed (broken lines) 24-hr height change; (d) computed z 
and ¢ +f att = 24hr. The height unit is 100 ft and the unit of vorticity is ; X 10-* sec. 
eram, there would, of course, remain other problems 
to be solved, not the least of which would be the de- 
vising of an objective technique for the location and 
elimination of errors in the raw data. 
Use of the Primitive Equations 
The discussion so far has dealt exclusively with the 
quasi-geostrophic equations as the basis for numerical 
forecasting. Yet there has been no intention to exclude 
the possibility that the primitive Eulerian equations 
can also be used for this purpose. The outlook for 
numerical forecasting would indeed be dismal if the 
quasi-geostrophic approximation represented the upper 
limit of attainable accuracy, for it is known that it 
of the application of the primitive equations: First, 
there is the difficulty raised by Richardson that the 
horizontal divergence cannot be measured with suffi- 
cient accuracy; moreover, the horizontal divergence is 
only one of a class of meteorological unobservables 
which also includes the horizontal acceleration. And 
second, if the primitive Hulerian equations are em- 
ployed, a stringent and seemingly artificial bound is 
imposed on the size of the time interval for the finite- 
difference equations. The first obstacle is the more 
formidable, for the second only means that the in- 
tegration must proceed in steps of the order of fifteen 
minutes rather than two hours. Yet the first does not 
