Using Environmental Tracers and Age Distributions to Estimate Non-Fickian Transport Behaviors

Abstract:

For many practical problems, non-Fickian transport can be described using a convolution of an input function and the underlying travel time distribution (TTD) between two points. In some cases, the TTD for dissolved solutes can also be expressed in terms of the groundwater age distribution and it has been shown that highly accurate estimates of transport behavior can be found when the required age distributions are known. However, identifying the TTD typically requires a field tracer test or numerical modeling, both of which may be difficult to implement realistically and both may contain significant uncertainty or error. Considering these limitations, it has yet to be shown how accurate transfer function based models that use an estimated TTD will be given an inexact approximation of the age distribution, that is one derived directly from measured concentrations of environmental tracers. This work considers the accuracy of the transfer function approach for simulating non-Fickian transport using several different approaches for constraining the age distributions. Fickian and non-Fickian models for age are estimated from simulated tracer data using simple 1-D models and streamtube ensembles. The errors in the estimated TTD propagate into the transport simulations and significant alterations in shape of the breakthrough curve can result. Despite these errors, the estimated transport behaviors are often very reasonable approximations of the known solutions, particularly in light of the limited data used to generate the estimates.