The Connected-Network Paradigm, Observation Bias and Non-Fickian Transport

Abstract:

Quantitative geologic modeling and percolation theory indicate that the upper 13- to 20% of the hydraulic conductivity (K) distribution in 3D clastic deposits tends to interconnect or percolate in all three dimensions, except when geological unconformities are present. This type of heterogeneous structure has been referred to as the connected-network paradigm (Fogg et al., 2000). In addition, quite commonly the total K distribution is comprised predominantly or substantially of aquitard materials (e.g., silts and clays, or, in the case of gravelly deposits, fine sands and silts). Because there are many more observations in the aquifers than the non-aquifer materials, a number of observation biases ensue, including underestimating the variance of K, lack of detection of the fastest moving solutes in connected high K channels, lack of detection of solute mass in non-aquifer materials. In turn, especially when the systems are modeled with Fickian assumptions, the observation biases lead to model biases, including underprediction of solute first breakthrough and of time scales for contaminant remediation or retention by the aquifer system. Further, environmental tracers are frequently used to calibrate regional transport models, but these groundwater age estimation tools are biased in that most of them sample only the youngest (<~50 yr) fraction of a mixture of groundwater ages per sample, neglecting the commonly important intermediate age waters (>50 to ~3,000 yr). This groundwater age estimation problem is yet another observation bias that may be confounding attempts to build more realistic, regional transport models. Importantly, even non-Fickian models will not capture the above effects if the observation biases, are not recognized and corrected. Examples of the various observation and resultant model biases will be presented and discussed along with the proposition that the connected-network paradigm, and hence non-Fickian transport, is the rule rather than the exception. Lastly, the resulting need for more subsurface transport observatories will be asserted.