|
<br />Correlation between concentrations of metsls
<br />and streamflow, specific conductance, and pH were
<br />evaluated to determine whether field measurements
<br />of streamflow, specific conductance, or pH could
<br />potentially be used to predict metal concentrations.
<br />At site AR34.5, correlations ofthe logarithm of metal
<br />concentrations with streamflow and specific conduc-
<br />tance generally were not significant or had correlation
<br />coefficients less than 0.6 (table 3), indicating that
<br />neither streamflow nor specific conductance would be
<br />a reliable predictor of metal concentrations for this
<br />site. At site AR3l.0, correlations of metal concentra-
<br />tions with streamflow and specific conductance
<br />generally were not significant (table 4).
<br />At both sites, the concentrations of several
<br />metals were strongly or moderately negatively
<br />correlated to pH. At site AR34.5, pH was strongly
<br />correlated to concentrations of dissolved aluminum,
<br />dissolved iron, dissolved copper, dissolved and total
<br />cadmium, and total zinc. Correlation coefficients
<br />ranged from -0.874 to -0.912 (table 3), indicating
<br />that if a linear relation existed between pH and the
<br />logarithms of the concentrations of metals, pH could
<br />explain between 76 and 83 percent of the variation in
<br />metal concentration. Additionally, pH was moderately
<br />correlated to total copper, dissolved and total
<br />manganese, and dissolved zinc; correlation coeffi-
<br />cients ranged from -0.806 to -0.862 (table 3),
<br />indicating that pH could explain between 65 and
<br />74 percent of the variation in metal concentration.
<br />The pH was weakly correlated to the logarithms of
<br />concentrations of total aluminum and total iron
<br />(correlation coefficient of -0.551 and -0.512, respec-
<br />tively). Because pH was strongly correlated with
<br />concentrations of several metals, scatterplots of pH
<br />and the logarithms of the concentrations of metals
<br />were made to further evaluste whether a linear relation.
<br />appeared to exist between pH and the logarithms
<br />of the concentrations of selected metals. Upstream
<br />from Terrace Reservoir at site AR34.5, a visual
<br />observation of the plots indicated that a reasonably
<br />linear relation existed between pH and the logarithms
<br />of concentrations of dissolved aluminum, dissolved
<br />iron, dissolved and total copper, and dissolved and
<br />total zinc. Linear relations also existed between
<br />correlations of pH to the logarithms of the concentra-
<br />tions of metals that were moderately correlated
<br />with pH. Downstream from Terrace Reservoir at
<br />site AR3l.0, correlations of pH to the logarithms
<br />of concentrations of dissolved aluminum, dissolved
<br />and total copper, and total cadmium were strongly
<br />correlated (table 4). Additionally, correlations of
<br />
<br />pH to the logarithms of concentrations of dissolved
<br />iron, dissolved cadmium, and dissolved zinc (table 4)
<br />were moderately correlated. However, an evaluation
<br />of scatterplots of pH and the logarithms of the concen-
<br />trations of these metals indicated that, with the
<br />exception of dissolved aluminum, a linear relation
<br />did not exist between pH and the logarithms of the
<br />concentrations of these metals.
<br />Correlations among concentrations of metals
<br />indicate that, during the study, significant correlations
<br />occurred between the concentrations of all the metals
<br />of concern. At both sites, concentrations of most of
<br />the metals were moderately to strongly correlated with
<br />one another (tables 3 and 4). Strong cross correlations
<br />between concentrations of metals indicate that the
<br />predominant source of metals was the same or that
<br />similar processes affect the metal concentrations,
<br />or both.
<br />
<br />METAL LOADS INTO AND OUT
<br />OF TERRACE RESERVOIR
<br />
<br />Metal loads were computed to estimate the
<br />quantity of metals that was transported into and
<br />out of Terrace Reservoir between April 1994
<br />through March 1995. Metal loads are a function
<br />of metal concentration and streamflow and represent
<br />the amount of a metal transported past a river cross
<br />section during a specific time interval. The load
<br />estimates presented in this report are limited to
<br />dissolved and total aluminum, iron, copper, cadmium,
<br />manganese, and zinc.
<br />There are numerous methods for computing
<br />metal loads. The simplest method of estimating loads
<br />is the time-interval method (Scheider and others,
<br />1979), in which the data record for each site is divided
<br />into discrete intervals, generally at the midpoint
<br />between sampling points. The metal load then is
<br />estimated as the product of the concentration for that
<br />discrete period and the sum of the streamflow for that
<br />period. The primary disadvantage of this method is
<br />that the statistical uncertainty of the loads cannot be
<br />determined. Another method used for estimating
<br />loads is with regression equations between two
<br />constituents that are strongly related. Correlations
<br />between the logarithms of metal concentrations and
<br />streamflow and the logarithms of metal concentrations
<br />and specific conductance were weak (tables 3 and 4).
<br />Therefore, streamflow and specific conductance were
<br />not used to estimate concentrations for calculating
<br />metal loads. Correlations between the logarithm of
<br />
<br />METAL LOADS INTO AND OUT OF TERRACE RESERVOIR 25
<br />
<br />003182
<br />
|