Quantification of network model confidence using tracer test data

Tracer tests can generate a large amount of data associated with the residence time and travel path of water within a distribution system. Typical approaches for using tracer test data for validating or calibrating a network model rely on quali- tative approaches, often using simple visualization to estimate goodness-of-fit. This study presents three individual metrics to quantify different aspects of network model confidence: residence time, path mixing, and signal attenuation. Data from a recent field-scale tracer test is used to illustrate the three metrics, which are then applied to interpret a sub-set of monitoring locations from the tracer test data set. In general, the three metrics reasonably describe the ability of the existing network model to represent the observed tracer data. Additional discussion focuses on developing an overall single metric that could be used in estimating model confidence.