Dependence of Hydraulic Conductivity on Test Scale: Artifact or Something More?

Friday, 19 December 2014: 9:00 AM
James J Butler Jr, University of Kansas, Kansas Geological Survey, Lawrence, KS, United States
Over the years, there have been frequent reports of a dependence of hydraulic conductivity (K) estimates on test volume (scale). This dependence is commonly demonstrated using results from laboratory permeameter tests and from slug and pumping tests. Typically, K estimates from pumping tests are reported to be considerably larger than those from slug tests, which, in turn, are considerably larger than those from permeameter tests. Despite the large body of evidence, a key question remains unresolved – is this relationship produced by methodological differences or a natural underlying scale dependence in K? The most effective means to resolve this question is to use a single technique that can be configured to span orders of magnitude in test scale. The constant-rate pumping test is just such a technique. In the first few minutes to tens of minutes of pumping, the affected volume is relatively small. In contrast, after days of pumping, this volume can be orders of magnitude larger. If there is an underlying scale dependence, it should be revealed by K estimates determined from different durations of pumping. We have been investigating this issue using pumping tests in unconsolidated aquifers in central and western Kansas. In western Kansas, irrigation wells are pumped nearly continuously for the entire growing season. If nearby observation wells are available, we can use this near-continuous pumping as a long-duration, albeit imperfectly controlled, pumping test to increase the scale of the investigation. An initial assessment of these formal and ad-hoc pumping tests has found no evidence of a scale dependence in K; these tests are typically characterized by homogeneous aquifer behavior at small to moderate times of pumping after which boundary effects become dominant. Thus, if there is an underlying scale dependence, it is below the scale of these observations, which would have to be considered rather serendipitous. We can assess the relationship at smaller scales by placing observation wells immediately adjacent to the pumping well and by relating slug tests to pumping tests in the same well using deconvolution approaches. Although further work is needed to finally resolve this long-standing conundrum, we have seen no indication that the reported dependence is a reflection of an underlying scale dependence in K.