9343

• UCREFRP Catalog Number: 9343
• Author: Greve, A. I., N. E. Spahr, P. C. Van Metre and J. T. Wilson.
• Title: Identification of Water-Quality Trends Using Sediment Cores from Dillon Reservoir, Summit County, Colorado.
• USFW Year: 2001.
• USFW - Doc Type: Denver.
• Copyright Material: NO

Collecting water-quality samples from a
cableway downstream from Dillon Reservoir
on the Blue River.

were used for the reservoir samples. Concentrations of
dissolved aluminum, barium, copper, iron, manganese,
molybdenum, and zinc were determined both for
the stream- and the reservoir-water samples. Total
concentrations were determined for copper, iron,
manganese, molybdenum, and zinc. "Dissolved"
indicates that the water was passed through a
0.45-micrometer filter in the field prior to laboratory
analysis; "total" indicates that the water was not
filtered. Reservoir-water analysis data are available
in the USGS NWIS data base.

Data Analysis

Trend analysis was performed on the data
from the sediment core collected near the dam by
using Kendall's tau (Helsel and Hirsch, 1993). This
analysis measures the monotonic relation between
two variables, x and y. Because the analysis is rank-
based, it is resistant to the effects of small sample
sizes, censored data, and non-normal population distri-
butions. A positive correlation is identified if the y
variable increases more often than it decreases as x
increases. In this study, a trend was determined to be
present when the p-value of the statistical test was less

than 0.05. A p-value expresses the probability of
detecting a trend when no trend is present. A detailed
discussion of Kendall's tau and its application is
included in Helsel and Hirsch (1993). Because
Kendall's tau depends upon the rank of data rather
than a numerical value, an analysis using the core
depth or the estimated year of deposition would yield
the same results. In most cases, core depth (x variable)
was paired with constituent concentration (y variable);
the results were used to make inferences about trends
in water quality within the reservoir over time.
Mass loads of trace elements were determined
for water entering and leaving the reservoir during
water years 1997 and 1998. The loads were calculated
as a sum of daily loads, using daily mean flow from
gaging stations located near the sampling sites in a
regression based on the log of the instantaneous
concentration relative to the log of instantaneous flow.
The smearing estimator (Duan, 1983) was used to
correct for retransformation bias for log-transformed
data. In cases where there was no relationship between
the log of streamflow and the log of concentration,
the mean concentration was used to estimate daily
loads. This case was most often observed in the data
obtained from the Blue River downstream from Dillon
Reservoir. Censored data were replaced with one-half
the detection limit. Robertson and Roerish (1999) esti-
mated for small streams, where load estimates are less
accurate than those of larger streams, that errors are
commonly about 30 percent, assuming 2 years of
data and small sample sizes.

Quality-Control Samples

Quality-control samples were collected with
the sediment-core and water-quality samples. Quality-
control samples allow the bias and variability associ-
ated with sample collection, processing, and analysis
to be evaluated.

Sediment Cores

Field-blank quality-control samples were not
run as part of the pesticide or PCB analysis. However,
as part of the standard laboratory methods for pesti-
cide analysis, a replicate sample, surrogate recoveries,
and spike recoveries were determined and deemed to
be within acceptable limits by the laboratory with each
batch of 12 environmental samples. These methods are
described by Foreman and others (1995). A method

8 Identification of Water-Quality Trends Using Sediment Cores from Dillon Reservoir, Summit County, Colorado