Extended pilot test of a cross-injection in situ denitrification system for pre-emptive treatment of municipal well water.

Elevated groundwater nitrate concentrations have been linked to deleterious health and environmental effects. A significant source of the nitrate is nitrogen fertilizers applied to agricultural landscapes. Beneficial Management Practices (BMPs), including the optimization of fertilizer use and selective crop rotations, have proven to be effective in some cases. The city of Woodstock in southern Ontario relies on public wells for all of its municipal supply. Several of the wells have experienced chronic increases in nitrate concentrations exceeding the maximum allowable limit of 10 mg/L N-NO3-. While BMPs are established, an interim reduction plan based on enhanced in situ denitrification (Cross Injection System, CIS) in a 15 m thick zone of high nitrate mass flux within the aquifer zone was evaluated. Based in the results of preliminary acetate injection experiments, a C:N ratio of 2.35, (approximately 260 mg acetate/L), was selected to optimize the denitrification reaction. Injections were performed for six hours a day every day for a period of approximately two months. Dissolved oxygen (DO) and nitrate concentrations recorded over time indicated that reduction of both commenced within a few days of the beginning of the acetate injections and reduced levels were maintained for the remainder of the two-month injection period. Denitrification occurred throughout the profile although nitrate reduction was the highest in the lower groundwater velocity zones. An overall reduction of nitrate of 50% was achieved through the treated section of the aquifer. It is estimated that an upscaled treatment system utilizing a treatment width of only 70 m would be sufficient to reduce the nitrate concentrations to below the drinking water limit demonstrating the potential for the CIS method to functions as an interim groundwater nitrate reduction strategy.

Design, testing, and implementation of a real-time system for monitoring flow in horizontal wells.

The Horizontal Reactive Media Treatment Well (HRX Well®) is a technology capable of collecting and treating groundwater passively. To monitor the internal flow rate, ensuring it remains at an acceptable level and maintains the desired capture zone size in the aquifer, Point Velocity Probes (PVPs) were adapted for the task. The modified PVP was assessed for its performance in a mock-HRX Well setting, which consisted of a laboratory sand column, similar in diameter to the HRX Well cartridges that would house the PVP in the field, with flow directed along the longitudinal axis of the probe. Experiments were conducted to assess 1) the effects of friction between the fluid and the probe surface, which could bias a velocity measurement low, and 2) the effect of gas in the porous medium, potentially generated by some reactive media, on PVP signals and measurements of velocity. It was determined that PVP length exerted no discernable effect on the quality of the PVP performance. However, the effects of gas in the porous medium were varied. Compared to velocity estimates assuming a fully saturated porous medium, the bias to velocity measurements was negative if gas was limited to locations near a PVP's injection-detection array, such as might occur by inadvertent bubble injection during the tracer pulses. The bias was positive if the gas was generated uniformly throughout the porous medium, such as might occur in a reactive porous medium like granular iron, or one featuring vigorous biological activity. A field trial of the HRX Well was subsequently undertaken with internal PVPs. The PVPs identified a missing seal that was later retrofit, and documented flow in the HRX Well at velocities in the range of 1.3 to 4.0 m/d, which compared well with expectations based on a site model that predicted velocities of 1.3 to 2.3 m/d. The results of this study demonstrate that HRX Wells perform hydraulically as designed, and that PVPs are effective devices for tracking flow rates within them in near real-time.

Bonding of monocarboxylic acids, monophenols and nonpolar compounds onto goethite.

Adsorption of a diverse set of chemicals onto goethite was evaluated by column chromatography. The pH of the effluents was 4.7-5.2. Van der Waals forces dominate the exothermic adsorption of 8 nonpolar compounds (e.g., PAHs and chlorobenzenes). H-bonding is responsible for the adsorption of 32 monocarboxylic acids (i.e., benzoic acids, naphthoic acids and acidic pharmaceuticals) and their adsorption tends to be endothermic. Steric effects significantly decreased the bonding of monocarboxylic acids with ortho-substitutions. Exothermic adsorption of 10 monophenols is controlled by weak H-bonding. Bonding of these 50 solutes onto goethite is totally reversible. In contrast, inner-sphere complexation of phthalic acid and chlortetracycline with goethite occurred according to their low desorption ratio (1.1%-54.4%). Polyparameter linear free energy relationship (PP-LFER) models were established to provide acceptable fitting results of the goethite-solute distribution coefficients (RMSE = 0.32 and 0.30 at 25 °C and 5 °C, respectively). It is worthy to note that steric effects must be considered to get a better prediction for compounds with ortho-substitutions.

Application of new point measurement device to quantify groundwater-surface water interactions.

The streambed point velocity probe (SBPVP) measures in situ groundwater velocities at the groundwater-surface water interface without reliance on hydraulic conductivity, porosity, or hydraulic gradient information. The tool operates on the basis of a mini-tracer test that occurs on the probe surface. The SBPVP was used in a meander of the Grindsted Š(stream), Denmark, to determine the distribution of flow through the streambed. These data were used to calculate the contaminant mass discharge of chlorinated ethenes into the stream. SBPVP data were compared with velocities estimated from hydraulic head and temperature gradient data collected at similar scales. Spatial relationships of water flow through the streambed were found to be similar by all three methods, and indicated a heterogeneous pattern of groundwater-surface water exchange. The magnitudes of estimated flow varied to a greater degree. It was found that pollutants enter the stream in localized regions of high flow which do not always correspond to the locations of highest pollutant concentration. The results show the combined influence of flow and concentration on contaminant discharge and illustrate the advantages of adopting a flux-based approach to risk assessment at the groundwater-surface water interface. Chlorinated ethene mass discharges, expressed in PCE equivalents, were determined to be up to 444 kg/yr (with SBPVP data) which compared well with independent estimates of mass discharge up to 438 kg/yr (with mini-piezometer data from the streambed) and up to 372 kg/yr crossing a control plane on the streambank (as determined in a previous, independent study).

Surface carbon influences on the reductive transformation of TCE in the presence of granular iron.

To gain insight into the processes of transformations in zero-valent iron systems, electrolytic iron (EI) has been used as a surrogate for the commercial products actually used in barriers. This substitution facilitates mechanistic studies, but may not be fully representative of all the relevant processes at work in groundwater remediation. To address this concern, the kinetic iron model (KIM) was used to investigate sorption and reactivity differences between EI and Connelly brand GI, using TCE as a probe compound. It was observed that retardation factors (Rapp) for GI varied non-linearly with influent concentrations to the columns (Co), and declined significantly as GI aged. In contrast, Rapp values for EI were small and insensitive to Co, and changed minimally with iron aging. Moreover, although declines in the rate constants (k) and increases in the sorption coefficients were observed for both iron types, they were most pronounced in the case of EI. SEM scans of the EI surface before and after aging (90 days) established the appearance of carbon on the older surface. This work provides evidence that iron with a higher surface carbon content outperforms pure iron, suggesting that the carbon is actively involved in promoting TCE reduction.

Laboratory validation of a point velocity probe for measuring horizontal flow from any direction.

The point-velocity probe (PVP) quantifies groundwater speed and flow direction, i.e., velocity, at the centimeter scale. The first probe designs required that the flow direction be known a priori, within about 100° in order to position the probe during installation. This study introduces and assesses a '360° PVP' that measures flow from any direction without foreknowledge of the groundwater velocity. In tests conducted in a Nested Storage Tank (NeST) aquifer simulator packed with sand, PVP-measured velocities matched expected velocities within ±9° in direction and ±15% in magnitude, on average, consistent with previously reported PVP performances in laboratory studies. In tests involving 17 repacked NeSTs, the measured and expected velocities were within ±30° and ±30% on average, illustrating the sensitivity of flow to porous medium packing, and the probes' ability to sense these changes; the porosity was found to vary considerably between packings i.e., n=0.34±0.2. For flow directions between 0° and 80° of an injection port, the experimental error on velocity magnitude was within the ranges reported above. At higher flow angles, experimental sources of error contributed to greater uncertainties. Fortunately, in these cases there were always alternative injection ports (with lower angles to flow) that could be used to circumvent any biases. At low experimental flow angles (<10°) the calculated values tended to overestimate the actual flow angles. Fortunately, these cases were identifiable by the detection of tracer at detectors on either side of the active injection port. In several tests designed with an expected flow direction of 0°, averaging the calculated directions from each side of the injection port resulted in improved matches to the expected flow direction.

Validation of a new device to quantify groundwater-surface water exchange.

Distributions of flow across the groundwater-surface water interface should be expected to be as complex as the geologic deposits associated with stream or lake beds and their underlying aquifers. In these environments, the conventional Darcy-based method of characterizing flow systems (near streams) has significant limitations, including reliance on parameters with high uncertainties (e.g., hydraulic conductivity), the common use of drilled wells in the case of streambank investigations, and potentially lengthy measurement times for aquifer characterization and water level measurements. Less logistically demanding tools for quantifying exchanges across streambeds have been developed and include drive-point mini-piezometers, seepage meters, and temperature profiling tools. This project adds to that toolbox by introducing the Streambed Point Velocity Probe (SBPVP), a reusable tool designed to quantify groundwater-surface water interactions (GWSWI) at the interface with high density sampling, which can effectively, rapidly, and accurately complement conventional methods. The SBPVP is a direct push device that measures in situ water velocities at the GWSWI with a small-scale tracer test on the probe surface. Tracer tests do not rely on hydraulic conductivity or gradient information, nor do they require long equilibration times. Laboratory testing indicated that the SBPVP has an average accuracy of ±3% and an average precision of ±2%. Preliminary field testing, conducted in the Grindsted Å in Jutland, Denmark, yielded promising agreement between groundwater fluxes determined by conventional methods and those estimated from the SBPVP tests executed at similar scales. These results suggest the SBPVP is a viable tool to quantify groundwater-surface water interactions in high definition in sandy streambeds.

Sensitivity analyses of the theoretical equations used in point velocity probe (PVP) data interpretation.

Point velocity probes (PVPs) are dedicated, relatively low-cost instruments for measuring groundwater speed and direction in non-cohesive, unconsolidated porous media aquifers. They have been used to evaluate groundwater velocity in groundwater treatment zones, glacial outwash aquifers, and within streambanks to assist with the assessment of groundwater-surfaced water exchanges. Empirical evidence of acceptable levels of uncertainty for these applications has come from both laboratory and field trials. This work extends previous assessments of the method by examining the inherent uncertainties arising from the equations used to interpret PVP datasets. PVPs operate by sensing tracer movement on the probe surface, producing apparent velocities from two detectors. Sensitivity equations were developed for the estimation of groundwater speed, v∞, and flow direction, α, as a function of the apparent velocities of water on the probe surface and the α angle itself. The resulting estimations of measurement uncertainty, which are inherent limitations of the method, apply to idealized, homogeneous porous media, which on the local scale of a PVP measurement may be approached. This work does not address experimental sources of error that may arise from the presence of cohesive sediments that prevent collapse around the probe, the effects of centimeter-scale aquifer heterogeneities, or other complications related to borehole integrity or operator error, which could greatly exceed the inherent sources of error. However, the findings reported here have been shown to be in agreement with the previous empirical work. On this basis, properly installed and functioning PVPs should be expected to produce estimates of groundwater speed with uncertainties less than ±15%, with the most accurate values of groundwater speed expected when horizontal flow is incident on the probe surface at about 50° from the active injection port. Directions can be measured with uncertainties less than 15° with the most accurate measurements occurring when the flow angles are relatively low - on the order of 20°. At still lower flow angles, quantitation may suffer due to experimental limitations related to tracer delivery. However, useful qualitative assessments of α may still be possible under these conditions.

Upscaling Point Velocity Measurements to Characterize a Glacial Outwash Aquifer.

Small-scale point velocity probe (PVP)-derived velocities were compared to conventional large-scale velocity estimates from Darcy calculations and tracer tests, and the possibility of upscaling PVP data to match the other velocity estimates was evaluated. Hydraulic conductivity was estimated from grain-size data derived from cores, and single-well response testing or slug tests of onsite wells. Horizontal hydraulic gradients were calculated using 3-point estimators from all of the wells within an extensive monitoring network, as well as by representing the water table as a single best fit plane through the entire network. Velocities determined from PVP testing were generally consistent in magnitude with those from depth specific data collected from multilevel monitoring locations in the tracer test, and similar in horizontal flow direction to the average hydraulic gradient. However, scaling up velocity estimates based on PVP measurements for comparison with site-wide Darcy-based velocities revealed issues that challenge the use of Darcy calculations as a generally applicable standard for comparison. The Darcy calculations were shown to underestimate the groundwater velocities determined both by the PVPs and large-scale tracer testing, in a depth-specific sense and as a site-wide average. Some of this discrepancy is attributable to the selective placement of the PVPs in the aquifer. Nevertheless, this result has important implications for the design of in situ treatment systems. It is concluded that Darcy estimations of velocity should be supplemented with independent assessments for these kinds of applications.

Stimulating in situ denitrification in an aerobic, highly permeable municipal drinking water aquifer.

A preliminary trial of a cross-injection system (CIS) was designed to stimulate in situ denitrification in an aquifer servicing an urban community in southern Ontario. It was hypothesized that this remedial strategy could be used to reduce groundwater nitrate in the aquifer such that it could remain in use as a municipal supply until the beneficial effects of local reduced nutrient loadings lead to long-term water quality improvement at the wellfield. The CIS application involved injecting a carbon source (acetate) into the subsurface using an injection-extraction well pair positioned perpendicular to the regional flow direction, up-gradient of the water supply wells, with the objective of stimulating native denitrifying bacteria. The pilot remedial strategy was targeted in a high nitrate flux zone within an aerobic and heterogeneous section of the glacial sand and gravel aquifer. Acetate injections were performed at intervals ranging from daily to bi-daily. The carbon additions led to general declines in dissolved oxygen concentrations; decreases in nitrate concentration were localized in aquifer layers where velocities were estimated to be less than 0.5m/day. NO3-(15)N and NO3-(18)O isotope data indicated the nitrate losses were due to denitrification. Relatively little nitrate was removed from groundwater in the more permeable strata, where velocities were estimated to be on the order of 18 m/day or greater. Overall, about 11 percent of the nitrate mass passing through the treatment zone was removed. This work demonstrates that stimulating in situ denitrification in an aerobic, highly conductive aquifer is challenging but achievable. Further work is needed to increase rates of denitrification in the most permeable units of the aquifer.

Consideration of grain packing in granular iron treatability studies.

Commercial granular iron (GI) is light steel that is used in Permeable Reactive Barriers (PRBs). Investigations into the reactivity of GI have focused on its chemical nature and relatively little direct work has been done to account for the effects of grain shape and packing. Both of these factors are expected to influence available grain surface area, which is known to correlate to reactivity. Commercial granular iron grains are platy and therefore pack in preferential orientations that could affect solution access to the surface. Three packing variations were investigated using Connelly Iron and trichloroethylene (TCE). Experimental kinetic data showed reaction rates 2-4 times higher when grains were packed with long axes preferentially parallel to flow (VP) compared to packings with long axes preferentially perpendicular to flow (HP) or randomly arranged (RP). The variations were found to be explainable by variations in reactive sorption capacities, i.e., sorption to sites where chemical transformations took place. The possibility that the different reactive sorption capacities were related to physical pore-scale differences was assessed by conducting an image analysis of the pore structure of sectioned columns. The analyses suggested that pore-scale factors - in particular the grain surface availability, reflected in the sorption capacity terms of the kinetic model used - could only account for a fraction of the observed reactivity differences between packing types. It is concluded that packing does affect observable reaction rates but that micro-scale features on the grain surfaces, rather than the pore scale characteristics, account for most of the apparent reactivity differences. This result suggests that treatability tests should consider the packing of columns carefully if they are to mimic field performance of PRBs to the greatest extent possible.

Applications and implications of direct groundwater velocity measurement at the centimetre scale.

Three projects involving point velocity probes (PVPs) illustrate the advantages of direct groundwater velocity measurements. In the first, a glacial till and outwash aquifer was characterized using conventional methods and multilevel PVPs for designing a bioremediation program. The PVPs revealed a highly conductive zone that dominated the transport of injected substances. These findings were later confirmed with a natural gradient tracer test. In the second, PVPs were used to map a groundwater velocity field around a dipole recirculation well. The PVPs showed higher than expected velocities near the well, assuming homogeneity in the aquifer, leading to improved representations of the aquifer heterogeneity in a 3D flow model, and an improved match between the modelled and experimental tracer breakthrough curves. In the third study, PVPs detected subtle changes in aquifer permeability downgradient of a biostimulation experiment. The changes were apparently reversible once the oxygen source was depleted, but in locations where the oxygen source lingered, velocities remained low. PVPs can be a useful addition to the hydrogeologist's toolbox, because they can be constructed inexpensively, they provide data in support of models, and they can provide information on flow in unprecedented detail.

Transient heterogeneity in an aquifer undergoing bioremediation of hydrocarbons.

Localized, transient heterogeneity was studied in a sand aquifer undergoing benzene, toluene, ethylbenzene, and xylene bioremediation using a novel array of multilevel, in situ point velocity probes (PVPs). The experiment was conducted within a sheet-pile alleyway to maintain a constant average flow direction through time. The PVPs measured changes in groundwater velocity direction and magnitude at the centimeter scale, making them ideal to monitor small-scale changes in hydraulic conductivity (K). Velocities were shown to vary nonuniformly by up to a factor of 3 when a source of oxygen was established down-gradient of the petroleum spill. In spite of these local variations, the average groundwater velocity within the 7 m × 20 m sheet-piled test area only varied within ± 25%. The nonuniform nature of the velocity variations across the gate indicated that the changes were not due solely to seasonal hydraulic gradient fluctuations. At the conclusion of the experiment, microbial biomass levels in the aquifer sediments was approximately 1 order of magnitude higher in the oxygen-amended portion of the aquifer than at the edge of the plume or in locations up-gradient of the source. These data suggest that the transient velocities resulted, at least in part, from enhanced biological activity that caused transient heterogeneities in the porous medium.

Transport and kinetic studies to characterize reactive and nonreactive sites on granular iron.

To better understand controls on observed trichloroethene (TCE) reaction rates in granular iron permeable reactive barriers, column experiments were conducted with different iron loadings. Using a reactive transport model and the Kinetic Iron Model (KIM), unique estimates of Langmuir sorption parameters for both the nonreactive and reactive sites, and a rate constant for TCE reduction on the iron surface provides new insights into the character of granular iron that determines overall reactivity. Nonreactive isotherms found in this work compared well with literature isotherms, and it was also found that for TCE and Connelly iron, only about 2% of sorption occurred to reactive sites, in agreement with earlier work by others. Thus, the KIM parameters were found to be relevant and useful under the conditions of these experiments.

Effect of mixed anions (HCO3(-) -SO4(2-)-ClO4 on granular iron (Fe(0)) reactivity.

Batch experiments were conducted with granular iron (Fe(0)) in pH 10 solutions of 4-chloronitrobenzene (4ClNB) and mixed anions (ClO4-, SO(2-), and HCO3-). In pure solutions, SO4(2-) is known to enhance Fe(0) reactivity, whereas HCO3- has been variously reported to depress Fe(0) reactivity or enhance it ClO4- has been found to be minimally reactive with Fe(0). It was hypothesized that the effects of the anions on reactivity were mutually independent, and the combined effects could be predicted from simple mixing lines. In concentrated carbonate solutions (> 25% of the bicarbonate salt content in 8 mM ionic strength solutions), the hypothesis was supported. In mixtures where the aqueous carbonate species concentrations were low (< 25% of the salt content in 8 mM ionic strength solutions) an anomalous reactivity enhancement was noted. Geochemical modeling using PHREEQC suggested that precipitation of Fe(OH)2(a) in preference to FeCO3(s) in weak carbonate solutions freed CO3(2-) to corrode the iron, causing the deviation from the mixing line prediction. SEM analysis confirmed higher carbon presence on iron that had contacted carbonate rich solutions compared to iron that had not.

Development and assessment of a rate equation for chemical transformations in reactive porous media.

With increasing interest in reactive porous media for groundwater remediation, such as granular iron, kinetic rate equations based on the Langmuir-Hinshelwood (L-H) assumptions have proven useful. Three parameters describe L-H kinetics: the two Langmuir sorption parameters, J and Cmax, and the first order rate constant, k. Unfortunately, the Cmax and k are lumped in the L-H equation, making it impossible to estimate their individual magnitudes. A re-examination of the theory underlying the L-H rate equation showed that L-H kinetics are not necessarily appropriate for packed reactive porous media experiments in columns or in the field. A more general rate equation was derived by accounting for changes in sorbed concentrations over time. The equation contains the Langmuir sorption parameter Cmax not lumped with the reaction rate constant, k, so it is possible to obtain unique estimates of J and Cmax and the rate constant, k. A sensitivity analysis suggested that this separation of variables can be achieved over a finite range of conditions applicable to granular iron media. The equation was demonstrated to be applicable, and the separation of variables possible, using the reduction of 4-chloronitrobenzene with Connelly granular iron as a test case.

Effects of measurement error on horizontal hydraulic gradient estimates.

During the design of a natural gradient tracer experiment, it was noticed that the hydraulic gradient was too small to measure reliably on an approximately 500-m(2) site. Additional wells were installed to increase the monitored area to 26,500 m(2), and wells were instrumented with pressure transducers. The resulting monitoring system was capable of measuring heads with a precision of +/-1.3 x 10(-2) m. This measurement error was incorporated into Monte Carlo calculations, in which only hydraulic head values were varied between realizations. The standard deviation in the estimated gradient and the flow direction angle from the x-axis (east direction) were calculated. The data yielded an average hydraulic gradient of 4.5 x 10(-4)+/-25% with a flow direction of 56 degrees southeast +/-18 degrees, with the variations representing 1 standard deviation. Further Monte Carlo calculations investigated the effects of number of wells, aspect ratio of the monitored area, and the size of the monitored area on the previously mentioned uncertainties. The exercise showed that monitored areas must exceed a size determined by the magnitude of the measurement error if meaningful gradient estimates and flow directions are to be obtained. The aspect ratio of the monitored zone should be as close to 1 as possible, although departures as great as 0.5 to 2 did not degrade the quality of the data unduly. Numbers of wells beyond three to five provided little advantage. These conclusions were supported for the general case with a preliminary theoretical analysis.

Field test of a cross-injection scheme for stimulating in situ denitrification near a municipal water supply well.

A pilot-scale test of an in situ denitrification scheme was undertaken to assess an adaptation of the nutrient injection wall (NIW) technology for treating a deep (30-40 m) nitrate contamination problem (N-NO(-)(3) ~ 10-12 mg/L). The adaptation is called the Cross-Injection Scheme (CIS). It duplicates the NIW method without a wall; wells are installed and operated directly in the aquifer and high-flux zones of the aquifer are preferentially targeted for treatment. The test was conducted on the site of a municipal water supply well field, with the supply well pumping between 15-80 m(3)/h. Acetate was periodically injected into the aquifer between an injection-extraction well pair positioned across the normal direction of flow. The injected pulses were then permitted to move with the water toward the municipal wells, providing a carbon supply to drive the desired denitrification. The fate of nitrate, nitrite, acetate and sulphate were monitored at multilevel wells located between the injection location and the municipal wells. The acetate pulsing interval was approximately weekly (9 h injections), so that the system was operating passively 95% of the time. Previous work on the site has established that the highest solute fluxes were associated with a 1-3 m thick zone about 35 m below surface. This zone was found to respond to the acetate additions as a function of the municipal pumping rate and the carbon-to-nitrogen ratio (i.e., determined by the injected acetate concentration). Initially, acetate was injected just below the theoretical stoichiometric requirement for complete denitrification and nitrate disappearance was accompanied by nitrite production. Increasing the C:N ratio (doubling the acetate injection concentration) increased the removal of nitrate and diminished the occurrence of nitrite. Slowing the municipal pumping rate, with a C:N ratio of 1.2-1.6, resulted in complete nitrate attenuation with no nitrite production and no sulfate reduction. The experiment demonstrated that the CIS injection scheme is a viable option for the treatment of nitrate contamination in situ near high-capacity wells.

Probe for measuring groundwater velocity at the centimeter scale.

A novel method of measuring small-scale groundwater velocities in unconsolidated noncohesive media uses the travel time of a tracer pulse between an injection port and two detectors located on the surface of a cylindrical probe, called a point-velocity probe (PVP), as the basis for velocity estimation. The direction and magnitude of the water velocity vector were determined to within +/- 9% of magnitude and +/- 8 in direction, on average, in ten laboratory tank tests conducted with the PVP, when the velocities were between 5 and 98 cm/ day. Numerical simulations supported the accuracy of the underlying theory for interpretation of the PVP data and indicated that the technology is capable of measuring velocity at a very fine scale (0.5 cm around the circumference). The benchtop and modeling investigations indicated that the probe is moderately sensitive to the condition of the porous medium immediately next to the cylinder surface, suggesting that challenges exist for the deployment of the instrument in the field.