Abstract
In response to the need to examine the large amount of historic water-quality data
comprehensively across North Dakota and evaluate the efficiency of the State-wide sampling
programs, a study was done by the U.S. Geological Survey in cooperation with the North Dakota
State Water Commission and the North Dakota Department of Health to describe the water-quality data collected for the various programs and determine an efficient State-wide sampling
design for monitoring future water-quality conditions. Although data collected for the North
Dakota State Water Commission High-Low Sampling Program, the North Dakota Department
of Health Ambient Water-Quality Network, and other projects and programs provide valuable
information on the quality of water in streams in North Dakota, the objectives vary among the
programs, some of the programs overlap spatially and temporally, and the various sampling
designs may not be the most efficient or relevant to the objectives of the individual programs as
they have changed through time.
One objective of a State-wide sampling program was to evaluate ways to describe the spatial
variability of water-quality conditions across the State in the most efficient manner. Weighted
least-squares regression analysis was used to relate the average absolute difference between
paired downstream and upstream concentrations, expressed as a percent of the average
downstream concentration, to the average absolute difference in daily flow between the
downstream and upstream pairs, expressed as a percent of the average downstream flow.
The analysis showed that a reasonable spatial network would consist of including the most
downstream sites in large basins first, followed by the next upstream site(s) that roughly bisect
the downstream flows at the first sites, followed by the next upstream site(s) that roughly bisect
flows for the second sites. Sampling sites to be included in a potential State-wide network were
prioritized into 3 design levels: level 1 (highest priority), level 2 (second priority), and level 3
(third priority).
Given the spatial distribution and priority designation (levels 1–3) of sites in the potential spatial
network, the next consideration was to determine the appropriate temporal sampling frequency
to use for monitoring future water-quality conditions. The time-series model used to detect
concentration trends for this report also was used to evaluate sampling designs to monitor future
water-quality trends. Sampling designs were evaluated with regard to their sensitivity to detect
seasonal trends that occurred during three 4-month seasons—March through June, July through
October, and November through February.
For the 34 level-1 sites, samples would be collected for major ions, trace metals, nutrients,
bacteria, and sediment eight times per year, with samples in January, April (2 samples),May,
June, July, August, and October. For the 21 level-2 sites, samples would be collected for major
ions, trace metals, and nutrients six times per year (January, April, May, June, August, and
October), and for the 26 level-3 sites, samples would be collected for these constituents four
times per year (April, June, August, and October).