A quantitative comparison of simulated and measured heads showed the best overall fit, as indicated by the lowest absolute mean residual and normalized root-mean squared error, corresponded to a mid-stand of water table. The best head correlation, as indicated by the highest correlation coefficient, corresponded to a high water table. The worst overall fit to the model corresponded to a low water table. The less model accuracy during high and low water tables suggested that the model is desensitized by specified head boundaries near the recirculation trenches.
A comparison of simulated flow lines with those drawn from measured heads showed the best overall fit corresponding to a low water table, despite this condition also corresponding to the worst quantitative fit of heads. The projected flow lines showed a reasonable match to mid-stand of water table. The worst match between flowlines corresponded to a high water table, although this condition also corresponded to the best statistical correlation of heads.
This study illustrates the importance of: 1) post-auditing to confirm model accuracy; 2) avoiding use of head boundaries near flow boundaries (e.g., pumping and infiltration); and 3) evaluating the post-audit in terms most relevant to the purpose of the simulations (e.g., flow lines and targeted treatment areas rather than heads). Based on an accurate groundwater model, a full-scale remedial design can be prepared with a higher degree of confidence in the actual effects of the selected remedy.
See more of: Groundwater Remediation
See more of: Topical Sessions