In Situ Measurement of Methane Oxidation in Groundwater by Using ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUlY 1991, P. 1997-2004 0099-2240/91/071997-08$02.00/0 Copyright © 1991, American Society for Microbiology

Vol. 57, No. 7

In Situ Measurement of Methane Oxidation in Groundwater by Using Natural-Gradient Tracer Testst RICHARD L.

SMITH,'* BRIAN L. HOWES,2

AND

STEPHEN P. GARABEDIAN3

Water Resources Division, U.S. Geological Survey, Lakewood, Colorado 802251; Biology Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 025432; and Water Resources Division, U.S. Geological Survey, Marlborough, Massachusetts 017523 Received 11 February 1991/Accepted 17 April 1991

Methane oxidation was measured in an unconfined sand and gravel aquifer (Cape Cod, Mass.) by using in situ natural-gradient tracer tests at both a pristine, oxygenated site and an anoxic, sewage-contaminated site. The tracer sites were equipped with multilevel sampling devices to create target grids of sampling points; the injectate was prepared with groundwater from the tracer site to maintain the same geochemical conditions. Methane oxidation was calculated from breakthrough curves of methane relative to halide and inert gas (hexafluoroethane) tracers and was confirmed by the appearance of 13C-enriched carbon dioxide in experiments in which "3C-enriched methane was used as the tracer. A Vmax for methane oxidation could be calculated when the methane concentration was sufficiently high to result in zero-order kinetics throughout the entire transport interval. Methane breakthrough curves could be simulated by modifying a one-dimensional advection-dispersion transport model to include a Michaelis-Menten-based consumption term for methane oxidation. The Km values for methane oxidation that gave the best match for the breakthrough curve peaks were 6.0 and 9.0 ,uM for the uncontaminated and contaminated sites, respectively. Natural-gradient tracer tests are a promising approach for assessing microbial processes and for testing in situ bioremediation potential in groundwater systems.

Estimating the distribution and rates of microbial processes in groundwater environments has proven to be a difficult task. This is because many problems arise when the traditional methods developed for soils or aquatic sediments are applied to subsurface systems (8). These approaches usually include attempting to obtain undisturbed and uncontaminated subsurface samples and maintaining in situ conditions during activity measurements (29). However, many groundwater systems present the opportunity to assess microbial processes with a unique approach, the use of in situ natural-gradient tracer tests. These tests were originally designed and have been used primarily to study the hydro-

conditions. Our results demonstrate the utility of in situ natural-gradient tracer tests to study microbial processes in groundwater systems. MATERIALS AND METHODS

Study site. The study site is a freshwater sand and gravel glacial outwash aquifer located on Cape Cod, Mass., which has been contaminated by land disposal of treated sewage since 1936. The resulting contaminant plume is more than 3.5 km long, 0.9 km wide, and 23 m thick (17, 30). In general, the contaminant plume consists of vertical and horizontal gradients of specific conductance (50 to 400 iuS), dissolved oxygen (0 to 8 mg/liter), pH (5 to 7.1), dissolved organic carbon (1 to 4 mg/liter, much of it refractory in nature), and inorganic compounds, such as chloride (0 to 28 mg/liter), sulfate (4 to 30 mg/liter), nitrate (0 to 16 mg of N per liter), and ammonium (0 to 10 mg of N per liter) (4, 17, 18, 31, 33). These gradients have resulted in corresponding distribution patterns of bacterial abundance and heterotrophic uptake (12, 31). The chemical makeup of the plume, especially with regard to inorganic nitrogen species, is similar to that associated with more than 17 x 106 residential septic systems across the United States (3) and therefore represents increasingly common contamination conditions in regional aquifers. The aquifer receiving the contaminants is composed of stratified sand and gravel with a mean particle size and porosity of 0.5 mm and 0.39, respectively (17), and a horizontal flow rate of about 0.4 m per day (6). The groundwater temperature in the contaminant plume is 10 to 14WC

geology of aquifers (6, 19), but with appropriate modifications they have the potential to assess microbial processes as well (23). The advantage of such tracer tests over other approaches is that they use natural groundwater flow (i.e., natural gradients) to transport tracers through undisturbed sections of an aquifer (6, 7, 11, 19, 23, 27). In this study, we adapted the tracer test technology to study a microbially mediated process, methane oxidation, in situ in an unconfined sand and gravel aquifer. We chose to study methane oxidation because of the potential use of methane injection as an in situ treatment process for removal of halogenated aliphatic compounds from contaminated groundwater (25, 27, 35) and preliminary data indicated that the potential for methane oxidation existed in the aquifer under study. Using in situ tracer techniques, we were able to demonstrate that methane oxidation could occur within both a pristine, well-oxygenated region of the aquifer and a nitrate-enriched, anoxic section, and by coupling the results with a groundwater transport model we were also able to determine the kinetics of methane oxidation under both Corresponding author. t Woods Hole Oceanographic Institute contribution

(17).

Tracer test sites. Methane oxidation was assayed in situ in the center of the contaminant plume 0.25 km downgradient from the contaminant source and in an uncontaminated portion of the aquifer by using natural-gradient tracer tests. Each site consisted of an array of 15-port multilevel sampling

*

no. 7633.

1997

1998

SMITH ET AL.

APPL. ENVIRON. MICROBIOL.

TABLE 1. Injectate composition for each of the Tracer test

natural-gradient tracer tests Injectate constituent

Contaminated site Test 1 Test 2

C1-

C2F6 CH4 Br13CH4

Uncontaminated site

Injectate concn' 41.1 mM 25.1 ,uM 85.0 p.M 14.7 mM 180.2 F.M

Br-

12.8 mM 66.3 ,uM 225.0 ,uM a The values are averages of the values for subsamples taken during the injection process.

CH4 02

and the contents of the bag were thoroughly mixed. The tracer solutions were pumped back into the ground through the same two adjacent ports (0.6 m apart) of the injection well with a peristaltic pump, taking care to avoid injection of any gas bubbles. Samples for gas and ion analysis of the tracer solution were collected at the beginning, middle, and

FIG. 1. Diagram of the well locations at the two small-scale, natural-gradient tracer test sites.

devices (MLSs) situated in the same configuration (Fig. 1). Construction and installation of the MLSs have been described previously (19, 30). At the contaminated site (well F347), the vertical spacing between the MLS ports was 0.6 m, covering the interval from 5.6 to 13.7 m below the land surface. For the uncontaminated site (well F168) the vertical spacing between the ports was 0.4 m for the interval from 5.0 to 10.3 m below the land surface. Water samples were obtained from each port of the MLSs with a peristaltic pump (30). Specific conductance was determined by using an electrode, oxygen was determined by iodometric titration (1), and nitrate was determined with an Autoanalyzer (Lachat Instruments, Menquon, Wis.). Tracer tests. The methods used to prepare the tracer solutions differed at the two tracer sites; each method was chosen with particular regard to maintaining the in situ oxygen concentration at the site. In both cases, the tracer solution was composed of groundwater (100 liters) withdrawn from two adjacent ports of the upgradient injection well MLS with a peristaltic pump. Within the anoxic zone of the contaminated site, the groundwater was pumped into a gas-impermeable bag which had been previously flushed five times with N2 and contained 1 liter of a concentrated N2-sparged solution of a conservative tracer (NaCl or NaBr). During the first tracer test, the bag was vented and tracer gases were added by simultaneously bubbling the solution with C2F6 (14.7 liters/min) and CH4 (1.6 liters/min) for 1 h. For the second tracer test 1 liter of 13CH4 (99% pure; Cambridge Isotope Laboratories, Woburn, Mass.) was added to the bag (which was not vented) and was allowed to equilibrate for 12 h. The excess CH4 gas was then vented,

end of the injection. For the uncontaminated, aerobic groundwater system, groundwater was pumped into an open container, MgBr2 was added, and the solution was bubbled with CH4 (1.6 liters/min) and air (14.7 liters/min) for 45 min. The tracer was then injected into the ground in the same manner as described above for the contaminated site. The injectate composition for each of the tracer tests is shown in Table 1. Water samples were taken daily with a peristaltic pump from each of the 60 ports in each of the sampling grids. The first 120 ml of each sample was discarded, and the next 20 ml was collected in a syringe and injected into a stoppered 50-ml serum bottle, which was then inverted. Most samples were analyzed on the day of collection; a few were stored for 1 to 3 days at 4°C. The dissolved gases in the bottle headspace after equilibration at room temperature were assayed by gas chromatography (Porapak N; flame ionization detector), after which the halides were assayed with an ion-specific electrode (corroborated by using ion chromatography). For the 13C tracer test, 1-liter samples were collected, dissolved inorganic carbon was precipitated as SrCO3 (by adding 60 ml of SrCl2 6H20 in NH4OH [4 g/g] per liter of water sample) and filtered, and the 13C/12C content was determined by using mass spectrometry (31). Model formulation. A modified advection-dispersion equation with a Michaelis-Menten utilization term was used to simulate the movement, spreading, and losses to microbial consumption of methane in the small-scale injection tests. The one-dimensional transport equation, which contains terms for storage, advection, dispersion, and degradation, can be written as: dc a2c ac Vmaxc =aD axv + (1) at c + Km ax aX2 where c is the concentration of CH4 (or solute) in solution, t is time, D is the dispersion coefficient (= aLv), aL iS dispersivity, x is the spatial coordinate, v is the fluid velocity, Vmax is the maximum rate of CH4 degradation, and Km is the Michaelis-Menten constant. The modeling approach used in this study to simulate the transport of methane was to find the physical parameter

METHANE OXIDATION IN GROUNDWATER

VOL. 57, 1991

values (aL, v) by calibrating analytical advection-dispersion model solutions to the observed breakthrough curves of the nonreactive solute (Cl- or Br-) and to find the microbial degradation parameters (Vmax, Kin) by calibrating numerical solutions of equation 1 to observed methane breakthrough curves. In the case of chloride, there was no degradation (Vmax = 0), and the fluid velocity (v) was calculated from the known distance between the injection point and the sampling point (x1) and the time to the peak concentration (tpeak) for the breakthrough curve (v = xl/tpeak). The dispersivity parameter (aL) was obtained from the chloride curve by using the following relationship, which was derived from the one-dimensional solution of the advection-dispersion equation with a pulse input (9):

XL(At/tpeak)2

-v

c(x - lAxt + At)[)2 + 4(Ax)

C(X,t + L+Att t)

1

D

+

+(AX)2

c(x + Ax,t + At)

c(x - AX,t) + c(x9t)[K.LAt

+

1 Vmax + + 2[Km +I~ c(x,t +

At)]_

-D

(A[)2 D

v

=

+ 4(x v

T(A-x+ 4(Ax) +

+

Lm + ~~c(x,t)] 2[Km

D

TABLE 2. Groundwater chemistry at the tracer test sites' Test site

Contaminated Uncontaminated a

Specific conductance

Oxygen concn

Nitrate concn

Methane concn

(,uS)

($LM)

(>LM)

(>M)

281 58

0.01 at the CH4 peak). In general, the relationship at this site indicated a first-order reaction when methane concen-

trations at the well fence were

15

maximum amount of methane consumed was 7

,uM.

,uM.

The

The rate

of methane consumption could be calculated from this value for any sample which had a methane concentration >15

because the concentration would have been >15

,uM

,uM

during

the entire transport interval. This rate (Vmax) was 0.4 FM/ day at the contaminated tracer site. The computer model was able to effectively simulate the breakthrough curves of both chloride and methane at the contaminated (Fig. 6) and uncontaminated tracer sites. The model velocity (v) and the dispersivity (CXL) parameters were calibrated by fitting the chloride breakthrough curves (Fig. 6A) with the Vmax set to zero (Table 3). The velocity and dispersivity were found to be somewhat lower for the uncontaminated site, reflecting the longer transport time at that site (Fig. 2 and 3). Then, by using these values for v and aL, the Michaelis-Menten parameters (Ki,m Vmax) were determined by adjusting their values to match the model predictions to the methane breakthrough curve. The modelsimulated concentration values were found to be inversely related to Vmax and particularly sensitive to Vmax at the peak concentration values (Fig. 6B). Simulated concentrations were found to be directly related to the Km (Fig. 6C), and peak concentrations were again most sensitive to its changes. The model does not have a unique solution for Km and Vmax; similar simulations can be obtained for completely different sets of values. However, because Km and Vmax are mathematically related, a unique Km can be obtained for each given Vmax by one of two methods. The Km value can

METHANE OXIDATION IN GROUNDWATER

VOL. 57, 1991

2001

0

w LLJ 0

CD 0a)

0 o

0

wE Z =L

C,)

z 03

I

w 0 20 0

o

0

0

4

8

12

16

20

24

CH4 (>M)

FIG. 5. Relationship between the amount of methane consumed for any given tracer sample and the measured methane concentration in that sample. The plot is a composite of the data from four depths (8.5, 9.1, 9.7, and 10.2 m below the land surface) from both tracer tests that were conducted at the contaminated site.

0 m

C) Co

potential to assess microbial processes in groundwater habitats. Because the groundwater environment is similar to the /X %/7 _ sediments of surface water environments and to soils in sediet sunsaturated ofenvironments (i.e., in each case the vast majorI ] , ity of the microbial community is attached to particulate surfaces), natural-gradient tracer tests represent a unique 22 24 10 12 14 18 20 16 opportunity to measure the in situ rate of a microbial process DAYS AFTER INJECTION in an undisturbed, sediment-dominated ecosystem. The results of this study demonstrate that small-scale FIG. 4. Time course of methane consumption (A) and 813C natural-gradient tracer tests can be used to measure methane values f or dissolved inorganic carbon (DIC) (B) during a second oxidation in a sand and gravel freshwater aquifer. The loss of tracer te st at the contaminated site with 13CH4 (99% pure). The data methane during the tracer tests was not due to physical are from a sampling port that was 9.7 m beneath the land surface. processes, such as dissolution of dissolved gases into bubbles, because both an inert gas (C2F6) and conservative ions were transported identically, or to retardation of methane by be obtaiined by selecting the value either by using a statistical sorption, because the shapes of the methane breakthrough best fit between the simulation and the methane breakcurves were identical to the shapes of the conservative ion through curve (Fig. 7) or by visually selecting the simulation that be,st matches the peak concentration (Fig. 6B and C). curves. The observed attenuation of methane was most likely the result of methane oxidation by the resident microThe fir st approach gives equal emphasis for the entire bial population within the aquifer. This process was clearly breakth rough curve, while the latter approach places more demonstrated as the operative mechanism with 13CH4; '3C emphas;is on methane consumption in the center of the tracer enrichment of the dissolved inorganic carbon pool could cloud. By using the Vmax of 0.40 ,imol of CH4 consumed per only have resulted from oxidation of the '3CH4 tracer. The liter peir day determined from the relationship shown in Fig. symmetry of the CH4 breakthrough curves (Fig. 2 and 3) 5, the Km for methane consumption at the contaminated suggests that the adaptation period for CH4 as a substrate tracer s;ite was determined to be 5.2 FiM when the least root was relatively short (