PNNL

Abstract

Uncertainty is naturally inherent in any environmental measurement. Contributions to uncertainty can come from a variety of sources. When measurements are derived from analysis of field samples, uncertainties in the results can be due to large-scale spatial site variations, small scale local in-homogeneity, sampling methods, sample handling, sample preparation, sub-sampling, and analytical variations. Each of these components of variation can be broken into additional sub-components that all combine to affect the total uncertainty. Visual Sample Plan (VSP) is a tool for determining the required number and placement of samples to ensure that the resulting data can support a sufficiently confident decision. It is important that users of VSP understand how the uncertainty estimates used in VSP represent the various components of variation described above. This paper will show how VSP can be used to explore the relative contributions of sampling and analytical uncertainties to the total uncertainty. Using VSP, one can evaluate whether it is better to reduce sampling variations by obtaining more samples or improving the sampling technique verses conducting replicate analyses or using a more precise analytical technique. The Measurement Quality Objectives (MQO) option will be demonstrated and discussed.