Scaling of groundwater flow subject to managed aquifer recharge using injection boreholes: Physical experiments and upscaled numerical models

Abstract

Managed aquifer recharge (MAR) holds great promise for solving the issue of groundwater resource depletion and quality deterioration in urban areas, but the scaling of groundwater flow patterns under MAR influence from controlled laboratory experiments to complex field scale conditions has not been fully resolved. This study applied physical and numerical groundwater flow experiments at the laboratory scale (100 m) along with ten numerical models at increasing spatial scales (100–103 m) for porous media representative of unconsolidated sand aquifers. We considered several spatial and temporal features that characterize groundwater flow patterns, such as the water table elevation, the thickness of the capillary fringe, the MAR lens geometry (defined as formed by the plume of artificially recharged water), the locations of stagnation points (defined as the points of convergence between ambient groundwater flow and injected water flow), and the groundwater travel times, which were examined at relative scales for comparison. The analysis showed that small-scale sandbox experiments overall underestimate the magnitude of the effects of MAR on groundwater flow pathways. The trends with increasing scales for these parameters were non-linear and could be fitted to power law relationships. These trends were more prominent at scales <20 times the size of the laboratory sand tank. Importantly, the heterogeneity of the sand material, as reflected by non-uniform spatial distributions of hydraulic conductivities (K) also had a major effect on the scaling of groundwater flow patterns, even for the relatively low degree of heterogeneity considered. Specifically, an increase in the average K and its standard deviation further exacerbates the scale dependency. This study provides insights into the appropriate scaling techniques to be used when applying small scale experimental results (e.g., from laboratory) to predict MAR-induced groundwater flow dynamics at larger field scales in moderately heterogeneous, unconsolidated sands.