|
<br />N
<br />.......
<br />N
<br />W
<br />
<br />
<br />However, for this site, the accuracy of the model may not be any more ques-
<br />tionable than the accuracy of the mass-balance estimate. The San Rafael River
<br />alternately has periods of no flow and flash flooding. Studies of other
<br />intermittent streams in the Upper Colorado River Basin have indicated that
<br />dissolved-solids concentrations fluctuate considerably during storms and from
<br />one storm to the next (Shen and others, 1981; Riley and others, 1982a, 1982b).
<br />When concentration fluctuates, estimation of long-term dissolved-solids
<br />discharge from periodic samples can be inaccurate. Therefore, the large
<br />difference between the statistical model and the mass-balance estimates at
<br />this site did not necessarily indicate an inaccurate model. Also, the abso-
<br />lute difference was only 1 percent of the mass-balance estimate at the next
<br />downstream site, number 16, Colorado River at Lees Ferry, Ariz., so the error
<br />at site 13 was inconsequential in contributing to any error in the intervening
<br />reach.
<br />
<br />Monthly mean natural dissolved-solids concentrations were computed as the
<br />quotient of the discharge values reported in table 9 and the monthly mean
<br />natural streamflows provided by the U,S. Bureau of Reclamation (written
<br />cornmun, , 1983). These concentrations then were plotted (figs. 4-8) to qual-
<br />itatively determine whether the model results were reasonable. The true
<br />natural dissolved-solids concentrations were assumed to plot as a fairly
<br />smooth curve having a minimum between May and July during snowmelt, and a
<br />maximum between November and March during winter baseflow. In addition, there
<br />should not be a pronounced peak in the late summer or early fall, Such a peak
<br />has been reported as characteristic of sites affected by irrigation return
<br />flows (Moody and Mueller, 1984), and its occurrence in estimates of natural
<br />dissolved-solids concentrations would make the estimates questionable.
<br />
<br />The curves were similar within subbasins and showed the expected changes
<br />in dissolved-solids concentration downstream. Along the Colorado River (fig,
<br />4), concentration increased from site 1 to site 2 (between Glenwood Springs
<br />and Cameo, Colo.), which is reasonable because saline springs contribute
<br />dissolved solids in this reach, Concentration then decreased between site 2
<br />and site 5 (from Cameo, Colo., to Cisco, Utah), which could be accounted for
<br />by the dilution effect of Gunnison River inflow, Concentration continued to
<br />decrease between site 5 and site 16 (Cisco, Utah, to Lees Ferry, Ariz,) from
<br />September through March, but increased from April through July, This effect
<br />could be caused by mixing of the Green River and San Juan River inflows, for
<br />which concentration does not vary as much as it does for the Colorado River
<br />(figs. 5 and 6),
<br />
<br />The curves for the Green River (fig. 5) generally indicated a steady
<br />increase in dissolved-solids concentration downstream. This increase was
<br />expected because the tributary concentrations coming into the Green River were
<br />relatively large; therefore, there was no dilution due to inflow. Baseflow
<br />concentration at the headwater site 6, below Fontenelle Reservoir, Wyo., was
<br />substantially smaller than that for site 1, Colorado River near Glenwood
<br />Springs, Colo. (fig. 4). This difference can be attributed to the large
<br />saline springs that occur upstream from site 1. Dissolved-solids concentra-
<br />tions were similar at the two sites during snowmelt when the effect of the
<br />springs was diluted, The concentration curve for site 8, Green River near
<br />Greendale, Utah, is somewhat anomalous, It seems too small for the fall and
<br />
<br />34
<br />
|