HYDROMETEOROLOGY IN THE UNITED STATES 
possess this characteristic, the assumption has come to 
be made that all of the maximum possible DDA values 
will not necessarily fall in one great storm but, rather, 
that each of several great storms will yield a limited 
number of maximum DDA values. Until meteorological 
theory can demonstrate otherwise, such an assumption 
—based upon data collected for maximum storms of 
record—appears to be logical. 
In practice, the space-time rainfall distributions ob- 
served in the greatest storms transposable to a given 
basin are used to estimate the distribution of maximum 
possible precipitation for that basin. If major-storm 
experience for the basin is large, a good selection of 
distributions becomes available. Regardless of the num- 
ber of such distributions which have been observed, 
however, it is apparent that the most critical distri- 
bution possible may not have occurred. Accordingly, 
the dependability of the method necessarily hinges upon 
the number of data which has been collected during a 
relatively short number of years. While the method 
is objective and reasonable within limitations, investi- 
gations should be conducted to develop other methods 
based upon assumptions physically more acceptable. 
Studies of intense, short-duration and small-area rain- 
fall such as occurs within a severe thunderstorm have 
disclosed a very wide variety of isohyetal patterns. For 
working purposes it is therefore assumed that there is no 
limit to the kinds of patterns which can result from such 
localized storms (very roughly, for areas up to one 
hundred or two hundred square miles and durations up 
to one or two days). It is possible, nevertheless, that 
detailed analyses of large numbers of thunderstorm- 
type isohyetal patterns may reveal certain wide limits 
as to space and time distributions of rainfall. 
Where production of rain is primarily controlled by 
orographic features, the problem becomes simpler. In an 
analysis of rainfall over a small drainage basin in 
Venezuela, forexample, Fletcher [5] founda high correla- 
tion between patterns of major storms and between the 
patterns of a major storm and of the mean annual 
precipitation. Here, meteorological reasoning strongly 
suggests that the most effective rain-bearing winds 
come only from one direction, thus that the maximum 
possible storm would closely resemble the mean annual 
precipitation as far as the pattern in space is concerned. 
For other orographic regions, therefore, the analysis in 
Venezuela suggests that design patterns may be deter- 
mined for each wind direction. 
The assumption is sometimes made that the total 
rainfall im a storm occurring in a mountainous region 
may be separated into two mutually independent parts, 
the orographic and nonorographic components. The 
orographic rain is assumed to result only from oro- 
graphic lifting; the nonorographic component is as- 
sumed to be that which is due only to the passing 
meteorological storm. If the occurrences of nonoro- 
graphic rain are random, the mean seasonal precipita- 
tion pattern can be assumed to be produced purely by 
the topographic configurations of the region. What is 
known as the isopercental method of storm transpo- 
sition is based on these assumptions. For a major 
1043 
recorded storm, the ratio (expressed in per cent) of the 
observed precipitation at each station to the corre- 
sponding mean seasonal value is plotted on a map. The 
resulting pattern of isopercentals is transposed from the 
region of storm occurrence to the project basin, and the 
percentages, applied to the mean seasonal values of the 
basin, produce the transposed isohyetal pattern and the 
magnitude of each isohyet. There is considerable logic 
to this method. However, there are certain objections 
which indicate that research should be conducted to 
estimate the degree to which it may be in error. For 
instance, there are probably significant inaccuracies in 
the assumption that the orographic and nonorographic 
components are mutually independent. It is conceivable 
that many storms occur in mountainous regions simply 
because the mountains are there; if the ground had 
been level, the synoptic situation might have existed as 
a nonproducer of rain. Again, the mean seasonal pat- 
tern may deviate significantly from a pattern derived 
from only the major rainstorms of record. In the many 
ordinary rainstorms which are so important in forming 
the mean seasonal pattern, variation of the condensa- 
tion level alone could produce a pattern appreciably 
different from that produced by storms with a uniform 
condensation level. Furthermore, effects of orography 
upon rain-bearing winds from various directions can 
produce profound changes in the isohyetal pattern, 
such that the orographic-component pattern with one 
wind direction might only remotely resemble that with 
another. 
Recurrence Interval. In some drainage basins of 
large area, it is a matter of days for runoff from the 
upper reaches to arrive at proposed construction sites. 
It becomes important, for a basin of this sort, to prepare 
an estimate of the minimum number of rainless days 
that can occur between two storms of either maximum 
possible or near maximum magnitudes. Fundamentally, 
this is the same kind of problem brought up in the 
previous paragraphs, since it is concerned with space- 
time distributions of rainfall: The customary DDA 
analysis of rainfall on the basis of individual storms 
requires, however, adoption of somewhat different 
approaches. 
The technique usually followed relies upon precedent 
established in the approximately half a century of 
United States weather-map analysis. The synoptic type 
of the storm yielding the critical rainfall depths is first 
determined. The weather maps of record are then 
searched for recurrences of storms of the critical type, 
or types, and the minimum duration between the occur- 
rences is chosen as the recurrence interval. 
Application of the form of equation (9) to durations 
of several days can provide answers for recurrence- 
interval problems, although it must first be established 
that this equation is applicable in specific regions. The 
relation has been developed, basically, from highest 
available depths of record from storms occurring 
throughout the world. 
Seasonal Distribution of Rainfall. Design and oper- 
ational problems often require estimates of the way in 
which maximum possible precipitation varies with 
