THE HYDROLOGIC CYCLE AND RIVER FORECASTING 
various durations and patterns of rainfall distribution 
on each basin. The forecaster then selects the one most 
nearly comparable to the conditions of the current 
storm. By multiplying its ordinates by the predicted 
volume of runoff for the storm, he determines the prob- 
able outflow hydrograph. For storms extending over 
several forecast periods, the predicted hydrographs for 
each period may be added to arrive at the total flow. 
UNIT HYDROGRAPHS 
60 
FRENCH BROAD RIVER AT 
DANDRIDGE, TENN. 
40 
DRAINAGE AREA = 4446 SQ. MI. 
20 
wn O 
u 
568 
5 SUSQUEHANNA RIVER AT 
S TOWANDA, PA. 
fo) 
= 6 
2 DRAINAGE AREA = 7797 SQ. MI. 
© 
Q4 
< 
a5 
oO 
wo 
a2 
TI i ae T 
RAPPAHANNOCK RIVER NEAR 
FREDRICKSBURG, VA. 
20 
DRAINAGE AREA= 1599 SQ, MI. 
(0) | 2 3 4 5S 6 7 8 
TIME IN DAYS 
Fie. 4.—Typical unit hydrographs. 
The problem of predicting the basin-outflow hydro- 
graph is quite similar in its theoretical aspects to that 
of routing a flood wave downriver. The ordinary rout- 
ing procedure makes it necessary to adopt routing 
periods so short that the computations for a headwater 
basin would be too time-consuming for forecasting pur- 
poses. However, the electronic routing machine [6] 
makes this approach a practical reality. In fact, the 
use of headwater forecasting techniques may be ex- 
tended to areas much larger than can now be success- 
fully treated with the unit hydrograph. 
Snow-Melt Forecasting. Forecasting the runoff from 
melting snow poses many problems which have not 
yet been solved satisfactorily. In level terrain, the rate 
of melting of snow has been forecast with reasonable 
accuracy by use of factors expressing melt per degree 
day above 32F. If melting is concurrent with rainfall, 
the water equivalent of the snow may be added to the 
rainfall and a relation such as Fig. 3 used to compute 
runoff. 
In mountainous terrain the situation is much more 
difficult. Here the snow pack accumulates in the form 
of a wedge with its greatest depths at high elevations, 
1053 
tapering to zero depth at intermediate and low levels. 
Owing to the variation of temperature with elevation, 
melting normally occurs over a fairly narrow range of 
elevation immediately above the snow line. The term 
“snow line” is used to describe the lower limit of the 
snow pack and does not imply a contour of constant 
elevation. As this melting zone moves upslope with the 
annual rise of temperature, the distance to the outlet 
of the basin increases and the effect of a unit of melt 
on the outflow changes. 
Although it is believed that our understanding of 
the problem is sufficiently complete to permit a solution 
by the use of the multiple graphical correlation so effec- 
tively used in other phases of forecasting, records of 
snow-line elevation or area covered by snow have never 
been kept on a systematic basis and it has been ex- 
ceedingly difficult to procure data which are adequate 
for such analysis. 
Water-Supply Forecasting. The time lag of several 
months between the accumulation of snow in moun- 
tain areas and its subsequent melting has led to the 
development of methods for forecasting the volume of 
stream flow to be expected from a winter’s accumula- 
tion of snow. Such forecasts make possible the sched- 
uling of hydroelectric operations and the planning of 
agricultural work in terms of the water which will be 
available for irrigation. 
Two different methods have been used in the prep- 
aration of such forecasts. One method uses relations 
derived by correlation of accumulated. winter precipi- 
tation with runoff. Both precipitation and stream-flow 
data must be adjusted to eliminate time trends, and 
weights are assigned to the various precipitation sta- 
tions and months of the year in proportion to their 
significance in the correlation [5]. The other technique 
makes use of a direct relation between the water equiva- 
lent of snow on the ground at the end of the accumula- 
tion season and subsequent runoff [2]. Both approaches 
seem to have about the same level of accuracy, with 
some advantage in reliability going to the first because” 
of the longer record lengths which permit more careful 
statistical analysis. In all probability, a technique com- 
bining the two types of basic data would yield the 
most accurate forecasts. 
Closely related to the problem of long-range volume 
forecasts is that of long-range estimates of the seasonal 
peak flow. Such forecasts have been attempted for the 
larger basins with moderate success. However, mete- 
orological conditions during the late spring and early 
summer months play a far greater role in determining 
the peak flow than in determining the total volume. 
With the same amount of snow available, continued 
high temperatures for two or three weeks may cause a 
flood, while the same temperatures interspersed with 
short periods of cool weather will result in only mod- 
erate flows. Hence, large errors in long-range peak- 
flow forecasts must be expected in some years. 
The Role of Meteorology in River Forecasting. Both 
short-range flood forecasts and long-range estimates of 
water-supply volume or seasonal peak flow may be 
seriously in error if weather conditions differ materially 
