WMOD00583

• Contract/Permit #: 1-07-81-V0175
• Title: Some Laboratory Tests of the Accuracy of the Belfort Universal Recording Raingage - Technical Report
• Date: 3/1/1985
• Weather Modification - Doc Type: Report

To estimate the accuracy of the gages in measuring each
of the 1- to 3-hour-duration increments, the control flow
rates before and after each test flow period were averaged.
The average rate was used to estimate the amount of
precipitation which should have fallen into the gage for that
particular test flow period. This estimate was compared with
the chart total from readings at the beginning and end of
each test flow period. This procedure was applied to all 58
available test flow periods. The bulk of the data (72
percent) consisted of I-hour control samples, while the 1.5-,
2.0-, and 3.0-hour samples comprised 17, 7, and 4 percent of
the data, respectively. Inspection of the data (presented in
Appendix 2) shows that in most instances control flow rates
were quite consistent from one period to the next. In
approximately 78 percent of the cases, the flow rate deviated
by less than 10 percent from control flow period to the
next. A few cases exhibited substantial changes because
control flow rates were deliberately altered by 100 percent.
However, the flow system was never tampered with or adjusted
during a test flow period. At any rate, it is noteworthy
that even in the instances of altered control flow, the
departures between the precipitation estimated from the
averagedcont~Qlflows and that indicated on the chart was not
obviously larger than those observed when flow rates
were very consistent. This suggests that any changes
in flow rate were well represented by averaging.

Results, shown in Figure 3, indicate that in almost 40
percent of the cases the chart response was within -0.003 to
-0.007 inch of the estimated input. The estimated I-hour
control input amounts ranged from 0.009 to 0.109 inch
with a mean of 0.051 inches. The 1.5-hour data (9 cases)
ranged from 0.042 to 0.067 inches with a mean of 0.060
inches. Four 2.0-hour blocks had a mean of 0.070 inches
with a range of 0.038 to 0.101 inches while the two 3-hour
cases were 0.041 and 0.120 inches.

,
I
1

(

A tabulation of the data contained in Appendix 2
indicated that in 54 percent of the cases the chart response
was within ~ 10 percent of the input as estimated from
the control flow data. In 91 percent of the cases, the
chart response was within ~ 30 percent of the estimated
input, and in 97 percent of the cases, it was within
~ 50 percent. This is remarkable performance considering
the simulated precipitation amounts ranged from about
0.01 to 0.12 inches, that is, considering that only small
storm amounts were'simulated.

12