CUMULATIVE UNCERTAINTY IN MEASURED STREAMFLOW AND WATER QUALITY DATA FOR SMALL WATERSHEDS

Abstract

The scientific community has not established an adequate understanding of the uncertainty inherent in measuredwater quality data, which is introduced by four procedural categories: streamflow measurement, sample collection, samplepreservation/storage, and laboratory analysis. Although previous research has produced valuable information on relativedifferences in procedures within these categories, little information is available that compares the procedural categories orpresents the cumulative uncertainty in resulting water quality data. As a result, quality control emphasis is often misdirected,and data uncertainty is typically either ignored or accounted for with an arbitrary margin of safety. Faced with the need forscientifically defensible estimates of data uncertainty to support water resource management, the objectives of this researchwere to: (1) compile selected published information on uncertainty related to measured streamflow and water quality datafor small watersheds, (2) use a root mean square error propagation method to compare the uncertainty introduced by eachprocedural category, and (3) use the error propagation method to determine the cumulative probable uncertainty in measuredstreamflow, sediment, and nutrient data. Best case, typical, and worst case data quality scenarios were examined. Averagedacross all constituents, the calculated cumulative probable uncertainty (%) contributed under typical scenarios rangedfrom 6% to 19% for streamflow measurement, from 4% to 48% for sample collection, from 2% to 16% for samplepreservation/storage, and from 5% to 21% for laboratory analysis. Under typical conditions, errors in storm loads rangedfrom 8% to 104% for dissolved nutrients, from 8% to 110% for total N and P, and from 7% to 53% for TSS. Results indicatedthat uncertainty can increase substantially under poor measurement conditions and limited quality control effort. Thisresearch provides introductory scientific estimates of uncertainty in measured water quality data. The results and procedurespresented should also assist modelers in quantifying the quality of calibration and evaluation data sets, determining modelaccuracy goals, and evaluating model performance.