WSP04999

• File Number: 8054
• Description: Water Salvage
• State: CO
• Basin: Statewide
• Date: 6/1/1975
• Author: Utah Water Research
• Title: Water Salvage Potentials in Utah - Volume II - Evaporation Suppression by Reservoir Destratification
• Water Supply Pro - Doc Type: Report/Study

1'lI"J?"q
.I..J...J ....._

all converted to hourly readings. None of the
temperatures were recorded from the thermistors
situated close to the coils; these temperatures
served as a check on the accuracy of the surface
temperature and also showed the mixing efficiency.

Suppression was compuled from the differ-
ence in evaporation arising between the cooled
pans and the normal pan. This difference was
expressed as a fraction of the total evaporation
from the normal pan. Suppression averaged 29
percent in the north pan and 33 percent in the
center pan.

In the preliminary analysis, suppression was
used as the dependent variable in a stepwise
multiple regression program; the independent
variables were water temperature. change in water
temperature, evaporation, dew point. and wind.
The change in water temperature was the most
significant variable followed by evaporation, wind,
temperature, and dew point. The coefficient of
determination (R) was found to vary through a
range of .67 to .71. Since the additional variables
had little effect on the correlation. suppression was
determined as a function of change in temperature
only in a least squares linear regression.

Scatter among the data increased at the end of
the evaporation season. When data from the last 9
days. October 8 to October 16 were deleted. the
correlation coefficient of .88 was achieved. The
results are shown in Figure 7. This decrease in
error is to be expected since the error involved in
filling a pan increases substantially as evaporation
decreases. For this reason, the data gathered on
cold days (the last 9 days) were deleted from the
analysis. This pan mling error is discussed in detail
in Appendix A.

Confidence intervals were estahlished at the 9S
percent level on the intercept value and the 98
percent level on the slope.

Running averages were calculated to rid the
data of the day to day variability common in
meteorological observations. Using the smoothed
data for to-day averages for both suppression and
change in temperature, linear regression showed a
substantial decrease in scatter (Figure 8). The
correlation coefficient was increased to .95.
Confidence intervals were established at the same
percentage levels used for the raw data analysis.

The last 8 days of data taken at the station
from October 17 until October 28 were used to
check the evaporation values of the pans with no
cooling effects. The first 4 days were without
cooling, and the last 4 days without cooling or
stirring. During this time there was no difference in

evaporation among the three pans. This indicates
that: the cooJing coils were not influencing any
other aspect of the evaporation pans aside from
changing the temperature of the water and the
stirrer had a uniform effecl on all three pans.

Modelsuppresslou functlou

The suppression function discussed previously
will be referred to as model suppression in order to
ditl'erentiate this calculation from the empirical
results measured at the experimental pans. The
basic model function is:

Ec eswc fIT c)
Supp. = 1--= 1--= I--

En eswc f(T n)

The suppression values computed from this
equation indicate that suppression does vary
slightly with water temperature (Tn) and this effect
is included in the detailed model to be discussed
later. However. the correlation with Tn is so small
that it will be ignored in the discussion that follows
so that the simplest possible (2 parameter)
comparison between model and empirical results
can be made. The model suppression values are
shown in Figure 9. The data fit a cubic equation
precisely but are close enough to linear for the
range of interest that the linear function coeffi-
cients are presented. The vertical distance between
the data points shows the minor effect of a range of
Tn from 70 to 300C.

Model and emplrlcal suppresslou
comparloou

Empirical suppression values were consistently
higher than the calculated model suppression
values. For the north pan the average value over the
period of record was a change in temperature of
3.90C with a corresponding average suppression
value of 29 percent. The mode' value of
suppression would be 23 percent for this tempera-
ture change. For the center pan the average
temperature change was 4.40C. and the average
suppression was 33 percent. A 4.40C temperature
change would have a model suppression of 2S
percent. A comparison between the 10 day
running average empirical suppression and the
model suppression is given in Figure 10. The figure
shows that the conservative nature of the model
which was anticipated in the previous chapter is
varified by this empirical phase of the research.

The model assumes that all parameters aside
from the vapor pressure difference are constant.
This approach does not take into account the
interactions among changing values. One impor~

17