PNNL

Abstract

Visual Sample Plan (VSP) is a Data Quality Objectives (DQO) based software tool that helps determine the required number and optimal placement of samples and performs statistical tests on the resulting data to support confident decisions. A new module is being added to better address long-term monitoring needs. This module is focused on trend detection and estimation. The objective of many long-term monitoring programs is to ensure that sampling is adequate to ensure a timely and likely detection of a significant trend. VSP allows the user to specify the size of the change that must be detected, the tolerable decision error rates, and the rate of sampling, then determines the number of samples required at each sampling interval. The VSP trend detection module also permits analysis of the data with and without seasonal adjustments. When seasonality effects are present, the Seasonal Kendall trend test is performed whereas if no seasonality effects exist, the Mann-Kendall test is performed. Least squares linear regression fits, nonparametric linear fits of the trends, and normality tests and descriptive statistics on the data and residuals are also provided in VSP. This paper will illustrate these methods and as time permits, new 3-dimensional visualization methods in VSP will be presented.