Estimating sampling variance and local environmental heterogeneity for both known and estimated analytical variance

Dubious data can yield incorrect interpretations. Therefore, environmental monitoring results should be unambiguously confirmed to be scientifically credible before evaluation. This paper illustrates this concept via a data-quality assessment of two water chemistry sites on the St. Mary River, British Columbia, Canada. This assessment required proof that the analytical variance is less than 4% of the observed variance in the concentrations of selected analytes, and that the sum of the sampling variance plus analytical variance is less than 20% of the observed variance. Verification, or lack thereof, in the above data-quality requirements is described along with a procedure to calculate sampling variance and local environmental variance when the analytical variance is either known or indirectly estimated. Two methods are described to estimate the analytical variance when it is not known. The data-assessment process and the procedures used appear to be effective tools for validating data and optimizing the costs of environmental monitoring programs.