Practical Guidance for Improving Pump-Test Based Inference of Hydraulic Parameters in Heterogeneous Unconfined Aquifers

Hydrogeological modeling relies on scale estimates for relevant hydraulic parameters. To obtain these, pump test data are typically interpreted using methods derived for idealized conditions (such as a spatially homogeneous medium). For unconfined aquifers, interpretation is made challenging by the time-evolving phreatic surface that imposes strong non-linearities in the equations. This study uses numerical simulation to examine the nature of parameter estimates obtained for heterogeneous unconfined aquifers using Neuman-based solutions. Results show that the representative horizontal extent of the cone of depression stabilizes when its radius approaches/exceeds $\mathit{\sim }15$ times the correlation length scale of aquifer heterogeneity. In practice, long pumping durations are needed to achieve this when length scales are large, presenting difficulties due to limited horizontal cone expansion caused by vertical leakage when vertical conductivity and specific yield are high. Further, the interpretation should properly account for observation well location. Estimates using data from wells beyond the horizontal representative extent align with arithmetic and geometric means, while those using wells from within the vertical representative extent deviate due to vertical and horizontal heterogeneities. Meanwhile, results obtained using wells from the “transition” zone are influenced by horizontal heterogeneity. The practical implications are that (a) Neuman-based estimates will only be reliable when the cone of depression approaches/exceeds the horizontal representative size, and (b) Observation wells should be located at appropriately representative distances. We recommend using a minimum of three strategically placed observation wells to ensure some degree of reliability when interpreting the results of pump tests in heterogeneous unconfined aquifers.