9521

• UCREFRP Catalog Number: 9521
• Author: Liebermann, T. D., D. K. Mueller, J. E. Kircher and A. F. Choquette.
• Title: Characteristics and Trends of Streamflow and Dissolved Solids in the Upper Colorado River Basin, Arizona, Colorado, New Mexico, Utah, and Wyoming.
• USFW Year: 1989.
• USFW - Doc Type: Denver.
• Copyright Material: NO

sites had been sampled at least once for dissolved-solids con-
centration, and 214 streamflow-gaging stations had at least
25 dissolved-solids analyses.
Selection of Streamflow-Gaging Stations
Seventy streamflow-gaging stations (hereinafter re-
ferred to as sites) were selected for analysis in this report
(table 3, pl. 1). Minimum criteria for selection were that a
site have a sustained period of analyses for dissolved solids
and a concurrent record of daily values of streamflow. Other
selection criteria included the length and completeness of
record, the availability of daily values of specific conduct-
ance, and a geographic location along a major stream or in
an area of special interest. The period of record for several
sites was increased by combining the records of adjacent
sites. For this study, about 1,500 site-years of dissolved-
solids data were compiled.
Estimation of Dissolved Solids
Annual and monthly dissolved-solids concentrations
and loads and mass fractions of major ionic constituents were
estimated using the computerized method described by
Liebermann and others (1987). Data were retrieved from the
U.S. Geological Survey's data base (Hutchison, 1975). The
data available in WATSTORE included mean daily stream-
flow, mean daily specific conductance, and periodic chemical
analyses. These data were evaluated to locate potential errors
that, if uncorrected, might degrade the accuracy of subse-
quent regression analyses and lead to erroneous results.
The corrected data were used to estimate the daily loads
of dissolved solids and selected ionic constituents. Loads
were computed using the linear regression
In(L)=by+bt In(Q)+b2 In (SC) (1)
where
In (L) =the estimated natural logarithm of load;
In(Q)=the natural logarithm of streamflow;
In (SC) = the natural logarithm of specific conductance;
and
bo, bl, 62 =regression coefficients.
If specific conductance was not available, load was computed
as a function of streamflow only:
In(L)=bo+bt in (Q) (2)
Equation 2 may not be appropriate for sites immediately
downstream from a large reservoir. In order to ensure the
applicability of equation 1 for such sites, missing values of
specific conductance were estimated by linear interpolation
between the last observation preceding the missing record
and the first observation following the missing record. Miss-
ing values of specific conductance were estimated for 3 of
the 70 sites evaluated in the Upper Colorado River Basin:
site 34, Green River near Greendale, Utah (downstream from
Flaming Gorge Reservoir); site 61, San Juan River near
Archuleta, N. Mex. (downstream from Navajo Reservoir);
and site 69, Colorado River at Lees Ferry, Ariz. (down-
stream from Lake Powell). For each site, missing specific-
conductance values were estimated only for the period when
streamflow was regulated.
The observed dissolved-solids and constituent loads
used to calibrate the regression models (eqs 1 and 2) were
computed as the product of streamflow and the dissolved-
solids or constituent concentration. For specific sampling
dates, constituent loads were calculated for dissolved
calcium, magnesium, sodium plus potassium, the carbonate
equivalent of alkalinity, chloride, and sulfate. For each site
the regression models were evaluated on three-year groups
of data. The calibrated models then were used to estimate
daily loads for the central year of the group. The daily loads
were summed to produce the monthly values used in subse-
quent analyses. Daily streamflows also were summed to
produce monthly values. The flow-weighted mean monthly
concentrations of dissolved solids and ionic constituents then
could be computed by division of the monthly load by the
monthly streamflow. The monthly load and streamflow were
summed to produce annual values, from which flow-weighted
mean annual concentrations were computed. A complete
tabulation of the monthly and annual time series of stream-
flow, load, and concentration at the 70 sites is included in
a separate data report (Liebermann and Nordlund, 1988).
Flow-Adjusted Concentration
For most sites in the Upper Colorado River Basin, con-
centration of dissolved constituents is related to streamflow.
As streamflow increases, concentration decreases; as stream-
flow decreases, concentration increases. This relation can
affect subsequent trend analyses of the data, because a signifi-
cant trend in concentration may be entirely the result of a
corresponding trend in streamflow. To distinguish a trend
in concentration caused by changing supply rates or sources,
the effect of streamflow first must be removed. The resulting
flow-adjusted concentrations then may be analyzed for trends
over time.
Residuals from regression of dissolved-solids concen-
tration as a function of streamflow commonly are used as
flow-adjusted concentrations. In this report, the regression
model assumed to relate concentration to streamflow for all
the sites was:
In (C) = bo + bt In (Q) (3)
Methods of Data Analysis 15