## 1. Introduction

[2] The estimation of groundwater velocity is a fundamental requirement in contaminant hydrogeology. Typically, this estimation is achieved through Darcy's law, corrected for porosity,

where *K* is hydraulic conductivity (L/T), *v* is average linear groundwater velocity (L/T), *n* is porosity, *h* = *ψ* + *z* (L), *z* is elevation head (L), *ψ* is pressure head (L), and *l* is the distance over which the hydraulic head is observed to change (L). Units given are generalized units with L = length, T = time. This calculation is sometimes applied to obtain a site-wide velocity estimate, using simple hand calculations based on hydraulic head and an estimate of bulk hydraulic conductivity. Often, more sophisticated calculations are performed by software that solves the equations for three-dimensional flow. Numerical models of this kind can account for hydraulic conductivity variations in space, as well as anisotropy, making it possible to compute groundwater velocity fields that are of great value in predicting contaminant movement.

[3] Despite the strengths of the Darcy's law approach, situations occur in which alternative methods of velocity estimation offer advantages. For example, it sometimes arises that the gradient is inherently low and difficult to measure accurately, even over distances of tens of meters [*Devlin and McElwee*, 2007]. In other cases, the area of concern is too small to permit accurate hydraulic gradients to be determined. With the advent of permeable reactive barriers, the need for velocity estimations over short distances is acute. Groundwater velocities determine residence times, and these are intimately connected to barrier performance. In these cases, and in general, uncertainty in the value(s) of *K* used in the calculation is quite large [*Sudicky*, 1986; *Mas-Pla et al.*, 1997; *Gierczak et al.*, 2006], and this uncertainty transfers to the velocity estimate.

[4] Several techniques have been developed to measure groundwater velocities directly. An obvious basis for these kinds of measurements involves the use of tracers. The environmental tracers, including ^{2}H, ^{3}H, ^{18}O, ^{14}C and ^{36}Cl, Freon compounds, and heat, have been used to study flow and transport at a variety of scales from meters to hundreds of kilometers [*Robertson and Cherry*, 1989; *Clark and Fritz*, 1997; *Bartolino and Cole*, 2002; *Kalbus et al.*, 2006].

[5] The use of injected tracers to investigate natural gradient groundwater flow goes back to the turn of the twentieth century [*Schlichter*, 1905]. Controlled tests of these kinds, called natural gradient tracer tests, have been used to investigate groundwater movement and transport processes on relatively small scales, up to a few hundred meters [*Mackay et al.*, 1986; *LeBlanc et al.*, 1991]. Unintentional spills have sometimes led to longer plumes [e.g., *Perlmutter and Lieber*, 1970; *van der Kamp et al.*, 1994], but lack of control of the history of these plumes often limits what can be learned from them. The injected tracers used for groundwater velocity estimation have included such substances as chloride and bromide [*Mackay et al.*, 1986], and fluorescent dyes [*Kasnavia et al.*, 1999].

[6] Because of the time and cost of performing natural gradient tracer tests a variety of instruments have been developed to measure groundwater velocity at the scale of a single well. These include point (borehole) dilution devices [*Pitrak et al.*, 2007], the Geoflo meter® [*Kerfoot and Massard*, 1985], the In Situ Permeable Flow Sensor [*Ballard*, 1996; *Alden and Munster*, 1997], the Colloidal Borescope [*Kearl*, 1997] and the Laser Doppler Velocimeter [*Momii et al.*, 1993]. Most of these techniques were included in the work described here, and so are discussed in more detail below. Some of the methods require a well, while others depend on the instruments coming into direct contact with the aquifer material. Among the instruments in the latter group is the recently introduced point velocity probe (PVP) [*Labaky et al.*, 2007]. A PVP consists of a cylinder outfitted with a tracer release and detection system on its surface. By timing the arrival of the tracer at 2 or more detectors on the cylinder surface, at different distances from the injection port, both magnitude and direction of the average linear velocity vector can be determined (Figure 1). In addition, unlike other methods, PVPs provide velocity estimates relevant to the centimeter scale, a scale comparable to that of multilevel sampling, and a scale at which geochemical and hydrogeological effects of microbial activity can be observed [*Devlin and Barker*, 1996; *Schillig*, 2008]. The viability of the PVP method was demonstrated in laboratory tanks and with numerical modeling [*Labaky et al.*, 2007]. It remains to be demonstrated that the probes can be installed and used to advantage in a field setting.

[7] The purpose of this work was to test PVP performance in the field against several other established techniques for groundwater velocity measurement. The methods chosen for comparison included bulk estimates of velocity from known discharge rates, the Geoflo® meter [*Kerfoot and Massard*, 1985; *Guthrie*, 1986], borehole dilution from drive point and standard wells [*Drost et al.*, 1968], and the colloidal borescope [*Kearl*, 1997]. Details of the materials and procedures for all methods are given below. *Labaky et al.* [2007] noted that a challenge for the PVP method, in field applications, is for the instrument to be installed in a fashion that promotes good contact between the aquifer and the probe surface, with a minimal disturbed zone, i.e., minimization of skin effects. In this work, we begin to address the challenge by performing the comparisons presented here in a noncohesive sand aquifer with a further goal of comparing methods of installing PVPs, including driving by vibrating hammer and jetting. Additional work and evaluations will be needed in other aquifer settings.