A field scale study of pumping-induced drainage and recovery in an unconfined aquifer

Abstract

An aquifer test involving seven days of pumping and five days of recovery was performed in the unconfined aquifer at Canadian Forces Base Borden in Ontario, Canada. Hydraulic head levels were intensively monitored in eleven observation wells, and moisture content profiles were frequently logged using a neutron moisture probe at six different radial distances from the pumping well. Capillary fringe thickness variations were inferred by comparing these hydraulic and moisture content measurements.

Allowing for the effects of lithologic variations, the response of the moisture content profile is predominately characterized by a downward translation during pumping-induced drainage. Our observations indicate that there is minimal excess storage above the capillary fringe at late times. The capillary fringe extended significantly with pumping in comparison to its static thickness; this extension progressively increased during the entire test. The capillary fringe extension decreased with increasing radial distance and was inferred to have eventually extended out to between 20 and 25 m. This extension represented a significant volume of excess capillary fringe storage that persisted into late times. A direct relationship between the vertical gradients below the water table and the excess capillary fringe storage is not apparent from these experimental data.

Analogous to the pumping test observations, the response of the moisture content profile during recovery is essentially an upward translation. While the hydraulic head levels recovered nearly to their pre-pumping elevations, the moisture content profiles stabilized approximately 0.20 m below their pre-pumping positions. This recovery differential leads to a capillary fringe compression that converted excess storage during pumping into storage deficit during recovery. In contrast to the significant radial variation of the capillary fringe extension during pumping, the capillary fringe compression is relatively uniform out to at least 15 m during the late recovery stage. The recovery deficit and entrapped residual air observed below the upward translating transition zone are strong indications that significant hysteretic processes are occurring at the field scale.

Introduction

Pumping and recovery tests are commonly used to determine the properties of unconfined aquifers (i.e. transmissivity, storativity and specific yield). The analyses of the data obtained from these aquifer tests are based on mathematical models for unconfined aquifer behavior. Because of the complexity of unconfined aquifer systems, these models contain approximations and assumptions that make their solutions more tractable. A number of authors (e.g., Neuman, 1979, Narasimhan and Zhu, 1993) have investigated the applicability of the assumptions. In almost all case, the conclusions of these studies are based on numerical modeling. While these studies are useful, additional insight can be derived from the interpretation of data collected during actual hydraulic testing of unconfined aquifers.

One critical parameter obtained from aquifer pumping test analyses is specific yield. Accurate values of this parameter are important to properly quantify the potentially extractable water volume from an unconfined aquifer; the main source of this water is drainage from above the declining water table. While it is possible to obtain reasonable estimates of specific yield (Moench, 1994), unrealistically low values of this parameter are often obtained in practice (Kruseman and de Ridder, 2000).

It has been proposed that poor specific yield estimates are a result of inadequately describing drainage phenomena in the analytical solutions used for pumping test analyses (Nwankwor et al., 1984, Nwankwor et al., 1992). This problem is partially due to the fact that the drainage processes occurring during a pumping test in an unconfined aquifer are poorly understood. To improve the understanding of these processes at the field-scale, detailed observations of the relationship between transient hydraulic head and moisture content distribution during pumping tests are required.

Existing field-scale data on pumping-induced drainage are limited. Nwankwor et al. (1992) performed a 24-h pumping test in the Borden aquifer during which transient hydraulic head and moisture content data were acquired, concentrating on earlier time response. Hydraulic head data were obtained in both the saturated and unsaturated zones and provided information on the temporal behaviour of vertical hydraulic gradients. In-situ moisture content data were acquired using a gamma attenuation technique and gravimetric water content analyses performed on soil cores obtained during pumping. While these data provided useful information on pumping-induced drainage, the tests were of too short a duration to observe the longer-term relationships between drainage and hydraulic head drawdown.

Recovery tests also provide valuable aquifer information. Usually, the late-time hydraulic head data, when elastic storage effects have dissipated, are analyzed using the Theis recovery method (e.g. Kruseman and de Ridder, 2000). The early- and intermediate-time hydraulic head data during recovery are not commonly examined in detail. Further, the behavior of the moisture content profile at the field-scale during recovery and its relationship to hydraulic head below the water table during this process are not well understood. To date, observations of the moisture content profile during recovery have not been published. These observations are important for understanding the imbibition process and potential hysteretic effects during recovery.

Geophysical imaging techniques have recently been applied to monitoring aquifer tests (e.g. Endres et al., 2000, Bevan et al., 2003). In general, these methods respond to variations in subsurface water content and have the potential to provide independent information about the drainage and imbibition. If these measurements are to be incorporated with standard hydraulic head data in aquifer test analyses, it is necessary to establish the relationship between moisture content distribution and hydraulic response of aquifers during testing.

This study presents the results of an extensively monitored pumping and recovery test in the unconfined Borden aquifer. This test was designed to improve the understanding of the moisture content profile response and its relationship to hydraulic head measurements during both pumping and subsequent recovery. Our experimental design expanded on the previous work by Nwankwor et al. (1992) in a number of aspects. First, the pumping test was performed for a seven-day duration to observe the late-time aquifer response. Second, the water content profile and hydraulic head was observed for five days during recovery. Third, the moisture content profile was monitored at denser radial and vertical sampling intervals in this experiment. While our observations support some concepts proposed by Nwankwor et al. (1992) regarding drainage during pumping, other aspects of observed aquifer behaviour differed significantly from those reported in that earlier work.