1044 
season. Preliminary estimates of the variations may be 
determined by the processes of storm transposition and 
adjustment, month by month. Unfortunately, such an 
approach suffers from a decimation of the number of 
storms available for processing. Other difficulties with 
storm samplings arise from the fact that storms are 
most frequently chosen for DDA analysis because of 
unusually large rainfall depths which have been ob- 
served, and that some seasons are characterized by 
maximum observed amounts much lower than those of 
other seasons. The method will become more useful as 
more storms are analyzed for their DDA values, and as 
the sampling becomes more nearly equal, season by 
season. 
The estimates will become more reliable when physi- 
cal bases are established for seasonal variations in the 
factors appearing in theoretical rainfall equations. It 
must be emphasized, however, that extreme rainfall 
depths are bemg considered, and thus that seasonal 
variations of extreme rather than average values of 
meteorological parameters must be developed. Hydro- 
meteorology necessarily deals with envelopes rather 
than with means. 
Snow Melt. When estimates of maximum possible 
precipitation are made for drainage basins which are 
located in the more northerly latitudes of the United 
States, or parts of which lie at high elevations, con- 
sideration must be given to the contribution of snow 
melt to the total runoff. In the spring of 1948, the 
Columbia River Basin experienced an excellent example 
of the conversion of a deep blanket of snow, in combina- 
tion with heavy rains, into a disastrous flood [25]. 
Basically, determination of potential runoff is amen- 
able to analysis by means of thermodynamic and turbu- 
lent-exchange theories. In a detailed treatment of the 
problem, Light [12] has defined the ‘“‘effective snow 
melt” as a combination of (1) melt due to the direct heat 
exchange which exists when the air is warmer than the 
snow mantle, (2) melt resulting from the latent heat 
released when atmospheric water vapor is condensed 
on the snow surface, and (3) the condensation of such 
atmospheric water vapor. He gives the simple relation- 
ship for the effective snow melt D in centimeters depth 
per second: 
D = Q-+ 600F)/80 + F = Q+ 680F)/80, (10) 
where Q is the heat transfer by convection in calories 
per square centimeter per second and F is the water 
transfer in cubic centimeters per second. Here, the 
latent heat of condensation amounts to 600 cal per cc 
of water deposited, and the latent heat of fusion is 80 
cal per ce. It can be seen that the moisture condensed 
on the snow surface melts 7.5 times its own weight of 
snow. 
The quantities Q and F’, inthe form given by Sverdrup 
[23], are expressible in terms of surface wind velocity, 
temperature difference between air and snow surface, 
air density and pressure, the surface roughness 
parameter, and the elevations of the anemometer and 
HYDROMETEOROLOGY 
hygrothermograph above the snow surface. These ex- 
pressions may be placed in equation (10) for the deriva- 
tion of Light’s snow-melt formula: 
D = U,,(0.00184(T — 32)10-0-%156 
+ 0.00578(e — 6.11)], 
where D is the effective snow melt in inches per six 
hours, Um is the average wind speed in miles per hour, 
T is air temperature in degrees Fahrenheit, ¢ is vapor 
pressure in millibars, and h is station elevation in 
thousands of feet above sea level. The formula assumes 
a snow-surface temperature of 32F, and reference eleva- 
tions of 50 and 100 ft for the anemometer and hy- 
grothermograph, respectively. The roughness parameter 
is assumed to be 0.25 cm. 
The snow-melt formula can be refined to take air 
stability and wind shear into account more accurately 
through the use of meteorological measurements at 
more than one level. The net melting effect of incoming 
and outgoing radiation may also be estimated by a 
method given by Wilson [83]. Furthermore, the rela- 
tively unimportant melting effect of rain may be esti- 
mated when assumptions are made as to raindrop 
temperatures. 
The theoretical formula represents the rate of melt 
of a smooth snow surface over which the wind blows 
without appreciable retardation due to such obstructions 
as trees. Project basins usually have sizable tree-covered 
areas, however, and sometimes are located in moun- 
tainous terrain. To account for errors in the assumption 
of surface roughness, an empirical factor of proportion- 
ality—the basin factor—is introduced into the formula. 
Comparisons of observed with computed values of melt 
are the bases for determination of the basin factor. The 
latter is normally found to be less than unity since, in 
most basins, forests exist to such an extent that the 
turbulent heat exchange occurring at a well-exposed 
index anemometer station is greater than the average 
heat exchange over the basin as a whole. 
When the basin factor has been determined, synoptic 
meteorological studies lead to conclusions regarding 
magnitudes of snowstorms antecedent to the maximum 
possible rainstorms, and the antecedent temperature 
regime and its effect upon the antecedent snow cover. 
The space and time distributions of wind and dew point 
immediately preceding and during the maximum rain- 
storm then form the bases for the determination of the 
wind and temperature parameters appearing in the 
snow-melt formula. It must be established that a certain 
critical depth of snow can exist at the onset of the rain- 
storm. This depth is such that the winds and temper- 
atures of the rain period will melt all of the snow cover. 
A snow cover which is greater than that which can be 
melted during the rain period will store some of the snow 
melt and act to reduce the runoff. Thus it is evident 
that the critical snow depth is not necessarily the 
maximum possible. 
The snow-melt problem is being systematically at- 
tacked by the collection and processing of data at three 
(11) 
