| Water-Quality Trend Analysis and Sampling Design for the Devils Lake Basin, North Dakota, January 1965 through September 2003
Master's Candidate, Department of Statistics
Colorado State University
Friday, Oct 6, 2006
This report presents the results of a study conducted by the U.S. Geological Survey, in cooperation with the North Dakota
State Water Commission, the Devils Lake Basin Joint Water Resource Board, and the Red River Joint Water Resource District, to
analyze historical water-quality trends in three dissolved major ions, three nutrients, and one dissolved trace element for eight stations
in the Devils Lake Basin in North Dakota and to develop an efficient sampling design to monitor the future trends.
A multiple-regression model was used to detect and remove streamflow-related variability in constituent concentrations. To
separate the natural variability in concentration as a result of variability in streamflow from the variability in concentration as a
result of other factors, the base-10 logarithm of daily streamflow was divided into four components—a 5-year streamflow anomaly,
an annual streamflow anomaly, a seasonal streamflow anomaly, and a daily streamflow anomaly. The constituent concentrations
then were adjusted for streamflow-related variability by removing the 5-year, annual, seasonal, and daily variability. Constituents
used for the water-quality trend analysis were evaluated for a step trend to examine the effect of Channel A on water quality in the
basin and a linear trend to detect gradual changes with time from January 1980 through September 2003.
The fitted upward linear trends for dissolved calcium concentrations during 1980-2003 for two stations were significant. The
fitted step trends for dissolved sulfate concentrations for three stations were positive and similar in magnitude. Of the three upward
trends, one was significant. The fitted step trends for dissolved chloride concentrations were positive but insignificant. The fitted
linear trends for the upstream stations were small and insignificant, but three of the downward trends that occurred during 1980-
2003 for the remaining stations were significant. The fitted upward linear trends for dissolved nitrite plus nitrate as nitrogen concentrations
during 1987-2003 for two stations were significant. However, concentrations during recent years appear to be lower
than those for the 1970s and early 1980s but higher than those for the late 1980s and early 1990s. The fitted downward linear trend
for dissolved ammonia concentrations for one station was significant. The fitted linear trends for total phosphorus concentrations
for two stations were significant. Upward trends occurred from the late 1980s to 2003 for most stations, but a small and insignificant
downward trend occurred for one station. Continued monitoring will be needed to determine if the recent trend toward higher
dissolved nitrite plus nitrate as nitrogen and total phosphorus concentrations continues in the future.
For continued monitoring of water-quality trends in the upper Devils Lake Basin, an efficient sampling design consists of five
major-ion, nutrient, and trace-element samples per year at three existing stream stations and at three existing lake stations. This
sampling design requires the collection of 15 stream samples and 15 lake samples per year rather than 16 stream samples and 20
lake samples per year as in the 1992-2003 program. Thus, the design would result in a program that is less costly and more efficient
than the 1992-2003 program but that still would provide the data needed to monitor water-quality trends in the Devil Lake Basin.