Determination of tracer mass for effective groundwater tracer tests

Defining protection areas around wells and springs used for water supply currently relies almost exclusively on model-derived groundwater velocities. Tracer testing provides actual groundwater velocities but is seldom undertaken. The reticence to undertake tracing can partly be attributed to the dilemma in selecting a tracer mass that can be unambiguously identified without unacceptable contamination of the water. A large number of ad hoc equations have been proposed to evaluate the mass of tracer required for tracer tests between sinking streams and springs or between wells but the accuracy of such equations has not been assessed. Here, a meta-analysis is undertaken of 211 natural-gradient sinking stream to spring tests and 44 forced-gradient well to well tests in carbonate aquifers, mostly using fluorescent dyes. The pertinent variables are mass (M), discharge (Q), linear distance (L), peak concentration (c) and travel time (t). Simple linear regression of log-transformed combinations of trace variables has been used to generate bivariate power relations. Regression analysis shows that the two equations first proposed by Martel and by Dole had the highest correlations, with the best fit for sink to spring tests being M/c = 23(LQ)(0.97) and M/c = 0.76(tQ)(0.99), respectively, using base SI units (m, s, g). For the well to well tests M/c = 3,100(LQ)(0.97) and M/c = 4.1(tQ)(1.02). Equations using travel time are marginally better fits, but are impractical for most applications. Well to well tracer tests exhibit much greater variability than sink to spring tests reflecting variations in preferential flow.