A NEW METHOD OF ESTIMATING STREAM-FLOW 
207 
spring flood peaks has already been noted. The comparison of total computed 
with total observed spring flood for each year and for the whole 5-year period is 
shown in Table 58. 
Tablk 58 — Comparison of total computed and observed spring floods 
Calendar 
year 
Percentage error in total 
computed spring flood 
1911 
Stream A 
Stream Ii 
+ 31 
-31 
-15 
+ 13 
+ 12 
+ 48 
-38 
+ 7 
+ 9 
+ 10 
1912 
1913 
1914 
1915 
The aggregate computed flood-flow for these spring-flood periods for all five 
years was 1 per cent less on Stream A and 0.3 per cent less on Stream B than the 
observed, respectively. 
COMPARISON BY YEARS 
How close is the agreement between the aggregate stream-flow as computed 
for each year and that observed for the year? This is shown in Table 59a. 
Table 59a — Comparison of annual aggregate computed with observed stream-flow 
Calendar 
year 
Percentage error in total 
computed annual flow 
1911 
1912 
1913 
1914 
1915 
Stream A 
Stream B 
+ 20 
-26 
-12 
+ 16 
+ 10 
+26 
-31 
+ 5 
+ 8 
+ 5 
For the whole 58 months on each stream treated as a single group, the aggregate 
computed flow is exactly equal to the observed flow. 
Note that the agreement on both streams for 1914-15 is better than for the 
years 1911-13, from which the constants were computed. 
RAINFALL, NET MELTING, EVAPORATION AND RUN-OFF ON WATERSHED B 
In Table 596 are shown the rainfall, net melting, evaporation and run-off on 
Watershed B in units of 0.01 inch of depth for four years beginning October 1, 1911. 
The net melting was computed from equation (36) in which were substituted 
values of C, F, M and T" shown in Table 49 for Solution X, Stream B. It will be 
noted that the resulting specific formula differs but little from the final adopted 
one, equation (92). 
The computed evaporation from water was computed by equation (23), in 
which, as already mentioned on page 155, the wind term entered very rarely 
because of the low altitude and sheltered locations of the anemometers. 
