|
<br />001924
<br />
<br />644
<br />
<br />], ENVIRON, QUAL., VOL. 27, MAY-JUNE 1998
<br />
<br />
<br />load calculations, with the order of preference being CDS,
<br />SUM, and ROE. Sites with data retrieved from STORET
<br />have dissolved-solids concentrations based on ROE,
<br />Monthly and annual dissolved-solids loads are based on
<br />estimates provided by the computer program SLOAD (Lieb-
<br />ermann et aI., 1987). In SLOAD, periodic water-quality data
<br />were used to calculate 3-yr moving regressions of dissolved-
<br />solids loads based on the daily stream discharge record and,
<br />if available, daily specific conductance. The accuracy of dis-
<br />solved-solids loads estimations commonly are improved when
<br />daily specific-conductance values are available, Daily dis-
<br />solved-solids loads were estimated by applying the regression
<br />coefficients to the daily-values data of discharge and specific
<br />conductance, and then the daily loads were summed by month
<br />and year to get aggregate monthly and yearly (calendar year
<br />or water year) dissolved-solids loads. Although monthly dis-
<br />solved-solids loads were determined for each site and then
<br />used in the calculation of annual dissolved-solids loads, tabula-
<br />tion of monthly loads for each site is too lengthy to include
<br />in this article. In using moving regressions to calculate loads,
<br />the dissolved-solids loads for the beginning and end years of
<br />data-1970, 1980, or 1986 and 1993, respectively-were less
<br />accurate than for the other years of data. To correct for this,
<br />dissolved-solids loads for the time periods 1970 to 1993, 1980
<br />to 1993, and 1986 to 1993 were estimated based on data for
<br />1969 to 1994, 1979 to 1994, and 1985 to 1994, respectively,
<br />
<br />Streamflow Adjustment
<br />
<br />Prior to analysis for trends, dissolved-solids concentrations
<br />and loads for 1970 to 1993, 1980 to 1993, and 1986 to 1993
<br />were adjusted for streamflow. Without this adjustment, the
<br />
<br />900,000
<br />
<br />800,000
<br />
<br />700 ,000
<br />
<br />C' 600,000
<br />~
<br />e
<br />J! 500,000
<br />~
<br />~
<br />.c
<br />0
<br />'" 400,000
<br />is
<br />! 300,000
<br />en
<br /> 200 ,000
<br /> 100,000
<br />
<br />o
<br />{\~
<br />~
<br />
<br />~"""
<br />
<br />{\'><
<br />~
<br />
<br />{\b
<br />~
<br />
<br />{\'b
<br />~
<br />
<br />~~
<br />
<br />fluctuation in dissolved-solids concentrations and loads due
<br />to natural or human-induced causes may have been masked
<br />by fluctuations in streamflow. Periodic dissolved-solids con-
<br />centrations were adjusted for streamflow by a multiple regres-
<br />sion of the natural log of concentration to the natural log of
<br />streamflow and the square of the natural log of streamflow, the
<br />residuals of which were called flow-adjusted concentrations
<br />(Smith et aI., 1982; Liebermann et aI., 1989; Schertz et aI.,
<br />1991), To adjust monthly and annual loads for streamflow, a
<br />relation between loads and streamflow was determined using
<br />hyperbolic regression models (Smith et aI., 1982), The residu-
<br />als, predicted values of the regression models minus the ob-
<br />served values, were called the flow-adjusted loads,
<br />
<br />Trends Testing
<br />
<br />The flow-adjusted concentrations and flow-adjusted loads
<br />were then tested for trends by using the seasonal Kendall test
<br />for dissolved-solids concentrations and monthly loads and by
<br />using linear regression for annual dissolved-solids loads. In
<br />the seasonal Kendall test, variations in periodic concentration
<br />data and monthly loads due to seasonality were accounted for
<br />by comparing concentration or load values measured during
<br />the same time periods or seasons. For example, January dis-
<br />solved-solids load data of I yr were compared only with J anu-
<br />ary dissolved-solids load data of other years. The seasonal
<br />Kendall test determines whether the seasonal Kendall tau
<br />value, which measures the strength of the monotonic relation-
<br />ship between two variables, is significantly different from zero.
<br />As tau deviates more from zero, with zero meaning no trend,
<br />the greater is the likelihood of a trend, In the seasonal Kendall
<br />test, the trend slopes are the median slopes of all possible
<br />
<br />300 ,000
<br />
<br />250,000
<br />
<br />200,000 ~
<br />g
<br />'-'
<br />
<br />~
<br />
<br />'"
<br />150,000 ]
<br />o
<br />en
<br />-0
<br />o
<br />>
<br />Sl
<br />'"
<br />100,000 is
<br />
<br />-Stream Discharge
<br />- - - Dissolved-Solids Load
<br />
<br />50,000
<br />
<br />o
<br />
<br />OJ'b''''
<br />....
<br />
<br /><>,~
<br />'\:
<br />
<br />!bb
<br />,~
<br />
<br />!bIb
<br />~
<br />
<br />Ojcf'
<br />'\:
<br />
<br />Ojo,'V
<br />'\:
<br />
<br />Water Year
<br />Fig.2. Annual stream discharge and dissolved-solids loads for Eagle River at Gypsum (5), water years 1970-1993. Number after site name refers
<br />to site number in Fig. 1.
<br />
|