Abstract:

Pumping-test interpretation in unconfined aquifers is often based on the Neuman solution, derived for homo- geneous aquifers, whose performance under heterogeneity and vertical anisotropy is unclear. We investigate how the piezometer location, normalized by spatial correlation length, affects hydraulic-parameter inference when applying the Neuman method to heterogeneous unconfined media, with emphasis on the effects of anisotropy. Synthetic heterogeneous 3D aquifers were generated and populated with stochastic random fields to account for differences in the hydraulic parameters (Kh, Kv, S, SY). Drawdown data for constant-rate pumping tests were generated at selected observation points, and hydraulic parameters were estimated from single-well transients using both the Single Parameter Value and Transient Parameter Value approaches based on the Neuman formulation. Results show that three location-dependent regimes exist as a function of two representative dis- tances – the vertical representative extent (VRE ≈ 15Lv) and the horizontal representative extent (HRE ≈ 15Lh), where Lh and Lv indicate horizontal and vertical correlation length scales, respectively. For “nearby” piezometers the parameter estimates exhibit strong, realization-dependent variability and converge to non-unique values, for “intermediate” piezometers they display transient variability, and for “distant” piezometers the influence of anisotropy largely vanishes and the estimates become realization-independent (converging to the geometric mean for Kh and Kv and to the arithmetic mean for S and SY). We translate these findings into practical strategies – use the transient approach to identify the HRE, prefer observation points at or beyond this extent, and combine near- and far-well data to bracket variability when only limited monitoring is available.