| Abstract: | ABSTRACT: This paper presents an approach to the evaluation of water quality sampling locations for their potential use for long term monitoring. This approach was applied to four sites on the Yukon River near the Canada-United States boundary. At three of these sites it was difficult to obtain representative samples due to the presence of extensive lateral heterogeneities. These heterogeneities occur because of a lack of mixing between the Yukon River and the major tributaries upstream from Dawson. Only one of the sampling locations is spatially homogeneous enough to provide representative samples. Concentration variations over the annual cycle are very large, often as much as two orders of magnitude. Estimates are made of the frequency and density with which samples must be collected to be able to detect a 10 percent different between annual mean concentrations. The estimated frequencies are so large that such an undertaking would be impractical. More importantly, the assumptions of this analysis are invalid, and time series analysis of fixed frequency samples is proposed as an alternative that is statistically rigorous and requires fewer samples. |
