A sampling plan for conduit-flow karst springs: minimizing sampling cost and maximizing statistical utility

摘要

Analytical data for nitrate and triazines from 566samples collected over a 3-year period at Pleasant Grove Spring,Logan County, KY, were statistically analyzed to determine theminimum data set needed to calculate meaningful yearly averagesfor a conduit-flow karst spring. Results indicate that a biweeklysampling scheduld augmented with bihourly samples from high-flow events will provide meaningful suspended-constituent anddissolved-constituent statistics. Unless collected over anextensive period of time, daily samples may not be representativeand may also be autocorrelated. All high-flow events resulting ina significant deflection of a constituent from base-lineconcentrations should be sampled. Either the geometric mean orthe flow-weighted average of the suspended constituents shouldbe used. If automatic samplers are used, then they may beprogrammed to collect storm samples as frequently as every fewminutes to provide details on the arrival time of constituents ofinterest. However, only samples collected bihourly should beused to calculate averages. By adopting a biweekly samplingschedule augmented with high-flow samples, the need tocontinuously monitor discharge, or the search for and analyzeexisting data to develop a statistically valid monitoring plan, is lessened.