| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Biogeochemistry Department, Max Planck Institute for Chemistry, P.O. Box 3060, D-55020 Mainz, Germany, Formerly with the Dep. of Land Resource Science, University of Guelph
b Dep. of Land Resource Science, University of Guelph, Guelph, ON, Canada N1G 2W1
c School of Engineering, University of Guelph, Guelph, ON, Canada N1G 2W1
* Corresponding author (sausma{at}alumni.uwaterloo.ca)
Received for publication November 19, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: DOY, day of year • DSL, Delhi sandy loam • MAC, MacDowell Lake landfarm • RL, refinery landfarm • THC, total hydrocarbon
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The objective of landfarming is to convert or treat substantial quantities of wastes containing degradable constituents into materials that are not a hazard to human health or the environment. In the landfarming process, wastes are spread onto the soil surface and cultivated into the upper soil layer. Biodegradation by indigenous soil microorganisms is considered to be the primary route of waste reduction; however, volatilization, leaching, and adsorption also reduce contaminant concentrations. Soil conditions are controlled through the addition of fertilizer to promote microbial activity and lime to control pH. Amendments in the form of organic matter (e.g., wood chips, sawdust, manure) may be added to improve soil conditions.
There are guidelines in some locations recommending control of emissions from landfarms used for treating wastes or contaminated soils. For example, government agencies in both Canada and the USA have implemented guidelines that recommend or require design changes to limit emissions (Alberta Environment, 1988; Coover and Walker, 1990; Environment Canada, 1993).
Field-scale measurements of trace gas emissions from soil surfaces can be complex and costly, as can experimental trials investigating driving forces or evaluating mitigation methods. Lab- and pilot-scale experiments offer opportunities to perform replicates, evaluate treatments that are otherwise prohibitive, and implement controls that would not be feasible at a field scale. However, results from lab-scale studies may not necessarily be applicable to the field if the design is overly idealized or simplified.
A variety of lab-scale reactor designs have been used to evaluate hydrocarbon emissions from soils contaminated with petroleum products. In each study the design was a function of the experimental goals: designs ranged from those with small surface areas of less than 100 cm2 (Dupont and Reineman, 1986; Lindhardt et al., 1996; Shonnard and Bell, 1993) to those that were a factor of 10 larger (Streebin et al., 1984; Wetherold and Balfour, 1986). Most studies used closed reactor systems in which air was passed through a headspace (Dupont and Reineman, 1986; Lindhardt et al., 1996; Shonnard and Bell, 1993). The study by Streebin et al. (1984) used open wooden boxes filled with soil over which an enclosure was periodically placed to evaluate hydrocarbon fluxes. Wetherold and Balfour (1986) describe a wind tunnel–like device in which a box of soil was placed. In all the studies, the air source was either purified compressed air from cylinders or a centralized compressor. Hydrocarbon evaluation methods ranged from the use of sorbent tubes (Dupont and Reineman, 1986; Lindhardt et al., 1996) to electronic hydrocarbon analyzers (Shonnard and Bell, 1993; Streebin et al., 1984; Wetherold and Balfour, 1986).
The only other trace gas flux that has been monitored from petroleum-contaminated soils undergoing bioremediation is CO2. Lab-scale monitoring of CO2 evolution from contaminated soils is used to evaluate biodegradation through microbial respiration (Bossert et al., 1984; Brown and Donnelly, 1983; Brown et al., 1983; Dibble and Bartha, 1979; Mott et al., 1990). Respirometers less than 1 L in size are typically used to study small volumes of soil.
This paper summarizes the design and construction of a simulator facility that enabled controlled studies that complemented field work previously performed by our research group (Ausma et al., 2001, 2002; Edwards et al., 2001). The results of several experiments assessing the influence of cultivation and rainfall on trace gas emissions are presented. Also considered is the influence of a simultaneous water vapor flux, which is often overlooked in lab-scale studies. During micrometeorological measurements of trace gas fluxes from surfaces corrections are routinely applied for simultaneous water vapor fluxes using what are commonly called the Webb, Pearman, Leuning (WPL) equations (Lee, 2000; Webb et al., 1980). An evaluation of the need to perform these corrections in lab-scale systems is presented here.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Dynamic Simulator Design
To avoid the complications and expense introduced by either scrubbing the source air to strip contaminants or using numerous cylinders of purified air, trace gas fluxes from the simulator soil surface were estimated with a dynamic approach. However, the design of the facility is flexible; incorporation of source air scrubbers and humidification columns can be accommodated.
Compressed air from a centralized compressor was continuously passed through each simulator and the flux of a trace gas was determined from the concentration difference between the source and exhaust air:
 | [1] |
A schematic of the simulator design is presented in the inset of Fig. 1 . The individual simulators were constructed from 150-L plastic barrels cut in half. A new base was constructed of steel held in place with 25.4-mm (1-in) circular angle iron supporting wire mesh mounted approximately 2.5 cm above the base. The wire mesh supported a bed of gravel beneath the soil layer to allow drainage of excess water via a valve tapped into the angle iron. The internal diameter at the base of the simulator was 58 cm. The simulators were approximately 40 cm tall with an upper diameter of 41 cm.
|
The original clamp-on lids supplied with the plastic barrels had 30-cm centers cut out and replaced with Plexiglas to allow visual inspection of the simulator contents. Two holes were then cut through the Plexiglas: a 9-cm hole (sealed with a rubber stopper) to allow physical sampling of the surface soil and a 3-cm hole for the insertion of a rubber stopper equipped with a thermocouple. A small fan installed on the lower surface of the lid ensured circulation and thorough mixing of the air within the simulator headspace.
Simulator Integration
Figure 1 illustrates the integrated simulators along with instrumentation. Five simulators were constructed and integrated with instrumentation through a data acquisition and control system. The footprint occupied by all five simulators was 2.8 m2. Source air flow was controlled upstream of the simulators and sampling system with dual pressure regulators with 5-µm filter elements (ARO Model 129121-500; Ingersoll-Rand Company, Bryan, OH) positioned in series, one of which was equipped with a pressure gauge. Air to the individual simulators was delivered by 12.7-mm (0.5-in) i.d. steel piping. At the simulators flow was limited to 20 L min-1 through 6.4-mm (0.25-in) brass metering valves (Nupro Model B-4MG; Swagelok Canada Ltd., Niagara Falls, ON) mounted at the inlet of each simulator. Flow gauges mounted downstream of the metering valves allowed visual confirmation of flow rates.
Each exhaust port was connected to a solenoid valve (ASCO Model 8262-13; Ascoelectric Limited, Brantford, ON) through 1.8 m of 6.4-mm (0.25-in) o.d. FEP tubing (Fluoroware, Chaska, MN). A manifold constructed of 6.4-mm (0.25-in) stainless steel tubing and fittings supported six solenoid valves that controlled sampling from the simulators and the compressed source air. Rope heaters (FGR-030; Omega Engineering, Stamford, CT) were wound around the manifold to minimize sorption of hydrocarbons to the tubing walls.
Sampling of the source air was performed by teeing into the supply piping at a point upstream of the simulators. A valve controlled flow of compressed air through 1.8 m of black polyethylene tubing. Downstream of the valve a flow gauge, open to the atmosphere, was placed on the branch of a tee, ensuring a continuous flow of air that could be sampled at atmospheric pressure.
A 2-µm stainless steel filter (Nupro Model SS-4TF-2F; Swagelok Canada Ltd., Niagara Falls, ON) positioned downstream of the valve manifold prevented particulate matter from damaging analytical instrumentation.
Simulator and source air sampling was performed with two pumps, placed in parallel, downstream of the valve manifold. A vacuum pump–compressor (Model UN05; KNF Neuberger, Trenton, NJ) drew air at a flow rate of approximately 5 L min-1 for delivery to a total hydrocarbon detector (THD). The second pump (Model L-79200-30; Cole Parmer, Vernon Hills, IL) was limited to a flow of 2 L min-1 through a metering valve (Nupro Model B-4MG; Swagelok Canada Ltd.) for sample delivery to a CO2 analyzer.
Instrumentation
Gases monitored continuously during the study included total hydrocarbons (THCs), CO2 and water vapor. Continuous real-time monitoring was accomplished with a combination of in house–built and commercially available analyzers summarized in Table 1.
|
Data Acquisition and Control System
In house–developed software integrated the total hydrocarbon detector, the CO2 analyzer and the relative humidity probe with the sampling system with a Pentium computer and data acquisition board (Model DT2836; Data Translations, Marlboro, MA). The software acquired data at 1 Hz and supplied continuous visual real-time graphical and numerical concentration values along with 0.5 h real-time estimates of the gaseous fluxes.
The data acquisition and control system was complemented with a datalogger (Model 21x; Campbell Scientific, Logan, UT) and multiplexer (Model AM32; Campbell Scientific), which were used to sample signals from the air and soil thermocouples at 0.1 Hz.
Trace Gas Flux Estimation
Through activation of the solenoid valves, each simulator was sequentially sampled alternately with the compressed source air for 90-s intervals (Fig. 2)
. During each 90 s sampling interval, 90 concentration values were collected. The concentration of each trace gas was determined by averaging the values once the analyzer signal reached steady state. When sampling was alternated between a simulator containing some level of trace gases and source air containing very low levels of the monitored trace gases, an ideal step change (upper waveform in Fig. 2) in concentration did not occur, due to air sample transit time (tlag), analyzer response time (tr), and dispersion of the sample within the tubing (tdisp). The lower waveform in Fig. 2 depicts the actual analyzer response. The lag time (tlag) and transient time (ttrans = tdisp + tr) were independently determined for each analyzer after construction of the integrated simulator facility and prior to experimentation and then incorporated into the software.
|
(µg m-3) of the air sampled from either the compressed inlet air (S) or a simulator (R):
 | [2] |
and contributed to lower noise levels since noise is proportional to (number of samples)-1/2.
The concentration difference (
CR) for each simulator (R) after each 90-s sampling interval was calculated with the average concentrations:
 | [3] |
By measuring the difference between the average of a simulator headspace concentration and the average of the preceding and following compressed source air concentrations, long-term signal drift was filtered from the gradient measurement, since the time scale for intake switching (90 s) was much shorter than the time scale of drift. The acquisition system averaged the measured difference (C) every 0.5 h (tp) for each simulator (R).
The flux of each trace gas from the soil surface was then calculated:
 | [4] |
Correction of Trace Gas Fluxes for Simultaneous Water Vapor Flux
With a simplified form of the Webb, Pearman, Leuning (WPL) equations, which removes the unneeded correction for temperature effects, trace gas fluxes can be adjusted for the presence of a simultaneous water vapor flux:
 | [5] |
is the average density (µg m-3) of the trace gas, 
is the total density of the air (µg m-3), 
is the water vapor density (µg m-3), Ma is the molecular weight of air, Mw is the molecular weight of water, and Eraw is the uncorrected water vapor flux (µg m-2 s-1).
Soil Sources
Two contaminated soils collected from active landfarms at which we had performed field-scale measurements were selected for the lab-scale experiments. An uncontaminated soil was used as a control. To replicate the heterogeneous nature of the field sites, the contaminated soils were not screened or homogenized. Contract laboratories performed the physical and chemical soil analyses, which included texture, carbon, nitrogen, and pH analyses, and hydrocarbon concentration determination. Soil hydrocarbon concentrations were determined following Canadian Council of Ministers of the Environment (1999) protocols. Table 2 contains nutrient and pH values for the three soils used.
|
Diesel Fuel–Contaminated Soil
Soil was collected in August 1999 from one of two landfarms located at MacDowell Lake, Ontario. Soil was stored in sealed 20-L buckets with zero headspace in a walk-in cooler at 4°C until used. The landfarms were used to treat excavated soil contaminated with diesel fuel. Several micrometeorological and chamber studies were performed at this site and are described in detail in Ausma et al. (2002) and Edwards et al. (2001).
The soil within the landfarm was classified as a silty clay loam and was contaminated with unweathered to moderately weathered diesel fuel. The landfarm was relatively heterogeneous with diesel fuel levels within the landfarm facility ranging from approximately 1000 to 6500 mg kg-1, depending on sampling location. Soil contaminants were characterized primarily by hydrocarbons 5 to 13 carbons in size. The average hydrocarbon concentration in the MacDowell Lake landfarm (MAC) simulators was 5100 mg kg-1. The landfarm was fertilized on 18 Aug. 1999, prior to soil collection, with 18–8–27 fertilizer (8% NO3–N, 4% NH3–N, 6% urea N).
The MAC soil aggregates were less than 19 mm in diameter; 59% (w/w) of the aggregates were <2 mm in diameter. There were no stones larger than 2 mm in the soil.
Refinery Landfarm Soil
Soil was collected from an active refinery landfarm in southern Ontario in August 1999 and stored in 20-L sealed buckets with a zero headspace in a walk-in cooler at 4°C until used. The soil was classified as a silt loam. For approximately 10 yr the site experienced intensive waste application and cultivation, resulting in the development of a relatively homogeneous surface soil layer. No waste was applied to the facility between the beginning of August 1999 and the soil collection date. Ausma et al. (2001) describes the facility and presents results of a micrometeorological study.
The hydrocarbon concentration in the soil was approximately 45 000 mg kg-1 and was characterized primarily by gravimetric heavy hydrocarbons larger that C15. Approximately 9% (w/w) of the refinery landfarm (RL) soil was composed of stones 2 mm and larger; 28% (w/w) of the soil aggregates were larger than 19 mm and 23% (w/w) were smaller than 2 mm.
Uncontaminated Soil
Delhi sandy loam (DSL) was collected in October 1998 from the Agriculture and Agri-Food Canada research farm in Delhi, Ontario. The soil was classified as a sand and was excavated from a site undergoing construction. The soil was dry when excavated (<1% w/w) and was stored at ambient temperature. A 4-mm sieve was used to screen out clumps of plant material. Hydrocarbon levels in the soil were below detection limits.
The soils had different textural and nutrient properties. The MAC soil was relatively heterogeneous as it was from a new facility used for only a short time. The RL soil, on the other hand, was more homogenous: for many years the large refinery landfarm underwent daily waste application. Refinery landfarm soil organic carbon concentrations (Table 2) were a factor of 9 larger than in the MAC soil. The RL soil was composed of large, stable aggregates while the aggregates in the MAC soil were primarily <2 mm in size. Organic matter is the major stimulus for the formation and stabilization of aggregates (Brady and Weil, 1999, p. 153); the RL soil had a much larger organic fraction. The DSL control soil did not form stable aggregates. These differences suggest that different mechanisms may ultimately influence trace gas movement through the soils.
The soils underwent relatively long periods of storage before use: MAC and RL soils were stored for 4.5 mo and the DSL soil for 15 mo. There would probably have been some reduction in microbial activity within the MAC and RL soils during storage at 4°C (Stenberg et al., 1998; Zelles et al., 1991) but some microbial activity would continue. The landfarm sources are not used over the winter months (November through April) for waste application, cultivation, or fertilizer application, so stored soil conditions represent a phase that the field soil would naturally have passed through.
The DSL soil was dry when excavated. No significant microbial activity would have occurred during storage.
Experiments
Experiments were designed to evaluate the influence of simulated cultivation and rainfall on the release of THC, CO2, and water vapor from the contaminated soils and to compare simulator results to field results. All experiments were performed at ambient lab temperature (approximately 22°C).
Two simulators were each filled with MAC and RL soil. These were run simultaneously with a control containing DSL. Each simulator was filled with approximately 45 kg of soil. In Experiment 1, the soil in each of the simulators was cultivated to a depth of 10 to 15 cm with a small gardening hand cultivator on three dates (Table 3) over an 11-d period in February 2000. No water was added to the soil during Experiment 1. Soil samples were periodically collected for moisture and hydrocarbon analyses. Trace gas fluxes were monitored continuously before and after cultivation.
|
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
c) was small (103–104 µg m-3) relative to the dry air density term (T - v) (109 µg m-3) resulting in a small correction term.
|
|
|
Experiment 1—Simulated Cultivation
Figure 4
presents flux time series along with soil temperature. Fluxes in the following discussion are presented with their associated standard deviations (i.e., flux ± standard deviation) as determined from repeated cultivations. Three simulated cultivations were performed over a 10-d period (Table 3) and are identified on Fig. 4.
Cultivation of the DSL soil did not result in elevated fluxes. The average fluxes for the three monitored trace gases were -0.06 µg C m-2 s-1, 0.23 µg CO2 m-2 s-1, and 0.19 mg H2O m-2 s-1, which were at or below the detection limits for THCs and CO2. The moisture content was 0.6% for the 10 d study period, explaining the low CO2 and water vapor fluxes.
All simulators displayed a similar diurnal trend in soil surface temperatures driven by diurnal temperature variations in room temperatures.
Total Hydrocarbon Emissions
Each cultivation of the MAC soils resulted in an increase in THC fluxes from the soil followed by a rapid decline. These emission bursts were a factor of 10 larger than undisturbed soil fluxes estimated between cultivations. Average peak fluxes were 105 (±16) and 84 (±19) µg C m-2 s-1 for MAC 1 and 2, respectively. Emissions leveled off to a lower value after each disturbance. Cultivation of the RL soils did not lead to elevated THC fluxes. The average THC flux within the RL simulators for the 10-d experiment was 0.8 µg C m-2 s-1.
The emission of hydrocarbons from soil to the overlying air depends on hydrocarbon availability, the vapor pressures of individual hydrocarbons, the tortuosity of the soil, and temperature. The volatilization potential is highly dependent on the carbon chain length and compound structure. Volatilization is greatest for lower molecular weight compounds such as one- to two-ring aromatic hydrocarbons and carbon chains less than C15 (Douglas et al., 1992) and decreases with increasing length of the hydrocarbon chain. With this in mind, and recalling that the hydrocarbons in the MAC soil were primarily C5–C13 hydrocarbons while C15 and larger hydrocarbons predominated in the RL soil, it is evident why higher THC fluxes were observed from the MAC soil than the RL soil. Fluxes from the RL soil were detectable, unlike the fluxes from the DSL soil, showing that limited volatilization was occurring in the RL soils.
Total hydrocarbon emissions from the RL simulators were low (<1 µg C m-2 s-1). Waste was not applied to the refinery landfarm facility for 1 mo prior to soil excavation. Total hydrocarbon fluxes peaking at 170 µg C m-2 s-1 were measured during spreading and cultivation at the same refinery landfarm (Ausma et al., 2001). The lack of measurable fluxes within the simulators suggests that the primary driving force for THC fluxes at refinery landfarms, such as the landfarm soil source here, is volatiles within the applied waste and that these volatiles quickly emit upon application of the waste to the landfarm.
The peak in THC emissions followed by a sharp decline after cultivation, as observed in the MAC simulators, has been noted in published laboratory (Streebin et al., 1984; Wetherold et al., 1981) and field-scale chamber studies (American Petroleum Institute, 1989b; Dupont and Reineman, 1986; Edwards et al., 2001; Evans, 1988). All but one of these (Edwards et al., 2001) were refinery landfarm studies in which waste had recently been applied to soil, a different scenario than the MAC site, where waste was not applied. Cultivation resulted in peak emissions that were approximately 10 times precultivation fluxes in the MAC simulators. Streebin et al. (1984) noted that volatilization was at least twice that prior to cultivation. From the time series published in an American Petroleum Institute (1989a) study, cultivation resulted in an increase in hydrocarbon fluxes of four to six times. The increase observed in the simulators is not unreasonable considering that these latter two studies were both conducted with refinery wastes.
The magnitude of the observed MAC THC fluxes can be compared with the field studies performed at the MacDowell Lake site (Ausma et al., 2002; Edwards et al., 2001). In these studies, THC fluxes were monitored from the two landfarms with a flux gradient micrometeorological (Ausma et al., 2002) and a static chamber method (Edwards et al., 2001). The THC emissions peaked at 131 µg C m-2 s-1 shortly after construction and tilling of the first MacDowell landfarm (Ausma et al., 2002), 38 to 55% larger than the average peak emissions measured in the MAC simulators. The site was constructed a few days prior to initiation of measurements, diesel fuel concentrations were an average of 12 320 mg kg-1 dry soil, and soil temperatures reached 32°C. High temperatures and availability of hydrocarbons in the freshly excavated soil contributed to the high measured fluxes relative to the simulator results.
Edwards et al. (2001) describes static chamber experiments performed to evaluate the influence of disturbing the soil surface on emissions from the second MacDowell landfarm. Disturbing the soil surface resulted in an average 16-fold increase in THC fluxes. Undisturbed soil fluxes ranged from 0.7 to 3 µg C m-2 s-1 while disturbed fluxes ranged from 5 to 47 µg C m-2 s-1. These values are lower than the simulator fluxes, as the soil at the landfarm was saturated leading to depressed fluxes as a result of inaccessibility of volatile hydrocarbons to air-filled pore spaces.
Carbon Dioxide Emissions
Within the MAC simulators, the response of CO2 fluxes to cultivation was similar to the THC fluxes, although not as pronounced (Fig. 4). Average peak emissions were approximately twice the magnitude of undisturbed soil emissions and reached 110 (±16) and 107 (±1.8) µg CO2 m-2 s-1 for MAC 1 and 2, respectively. The magnitude of the peak declined each time the soil was disturbed within MAC 1. Emissions leveled off to a smaller value after each disturbance. Within the RL simulators, cultivation resulted in increased CO2 fluxes. Average peak CO2 fluxes were 150 (±34) and 143 (±56) µg CO2 m-2 s-1 for RL 1 and 2, respectively. After each subsequent disturbance peak CO2 emissions within RL 2 declined. Fluxes dropped to a lower level after each cultivation.
Carbon dioxide emissions from soil are dependent on soil moisture, aeration, temperature, and availability of organic matter. The small amount of soil water available (Table 3) to microorganisms for the duration of Experiment 1 explains the low respiration rates. Relevant landfarm field studies in which CO2 fluxes were measured include the MacDowell Lake chamber study (Edwards et al., 2001) and a study performed at a second remote diesel fuel–contaminated soil landfarm in 1997 (Wong, 1999) also described in Edwards et al. (2001). In the first study, two pairs of static chambers were run to estimate CO2 fluxes from the MacDowell landfarm. Undisturbed soil fluxes were 93 and 531 µg CO2 m-2 s-1, while disturbed soil fluxes were 463 and 2240 µg CO2 m-2 s-1. In the second study, average daytime CO2 fluxes of 207, 244, and 215 µg CO2 m-2 s-1 were measured with two different micrometeorological techniques and a static chamber technique, respectively. Compared with the results for the MAC simulators, these values are approximately twice the measured peak flux values after cultivation and four times the fluxes several days after cultivation.
Since landfarms are essentially large, cultivated areas of bare soil, other data sources that can be used for comparison are respiration studies performed at bare soil agricultural sites. Rochette and Angers (1999) measured seasonal CO2 emissions of approximately 90 to 180 µg CO2 m-2 s-1, using a dynamic chamber method, from uncultivated sandy loam. Rochette et al. (1997) cite typical soil respiration rates of 100 to 200 µg CO2 m-2 s-1 on mineral soils in Ottawa. Dugas (1993) reported average chamber and micrometeorological CO2 fluxes of 39 and 42 µg CO2 m-2 s-1 in winter Texan weather. La Scala et al. (2000) used a flux chamber method to measure average CO2 emissions of 64 to 123 µg CO2 m-2 s-1 from a tropical soil. The values from these studies suggest that the observed simulator fluxes several days after cultivation (20 to 70 µg CO2 m-2 s-1) were low, probably due to the low soil moisture.
The goal of cultivating landfarm soil is to stimulate microbial activity by aerating the subsurface and to expose previously protected hydrocarbons to attack by microorganisms. Carbon dioxide emitted from the soil immediately after cultivation is either a result of enhanced microbial activity or a burst of emissions resulting from changes to the physical characteristics of the soil. Lower bulk densities, higher air-filled porosities, and the creation of networked macropores are associated with recent plowing (Carter, 1988). Resistance to gaseous transport within the soil is thus lowered and losses during this time probably represent a loss of stored CO2 rather than enhanced production. Enhanced respiration would be evidenced by maintained fluxes that are higher than those measured prior to cultivating (Rochette and Angers, 1999). This did not occur in any of the simulators, here fluxes equilibrated to a lower level after each subsequent soil disturbance indicating that aeration was not limiting respiration.
Water Vapor Fluxes
The water vapor flux from the soil surface in the MAC simulators peaked at the same value after each disturbance: 16 (±1.9) and 15 (±2.7) mg H2O m-2 s-1 for MAC 1 and 2, respectively. Fluxes dropped without reaching a constant value prior to each successive cultivation. Soil moisture content was relatively constant within each simulator for the 10-d duration of the experiment. Moisture contents were low (Table 3). Peak water vapor fluxes in RL 2, with an average of 18 (±1.9) mg H2O m-2 s-1, were higher than RL 1, with an average of 14 (±1.5) mg H2O m-2 s-1, correlating to a higher soil moisture content in RL 2 (11%) relative to RL 1 (4.5%). Following peak fluxes, the values decreased and did not level off prior to the following cultivation.
The evaporation rate of water from soil is controlled by the availability of water and atmospheric conditions. As long as there is adequate moisture and the soil is conductive enough to supply adequate water to the evaporation site, the rate of soil drying is constant. The evaporation rate during this period is controlled by meteorological conditions such as radiation, wind, air temperature, and humidity (Hillel, 1998). As the soil dries out, less water is available for evaporation and a falling rate period begins. In this stage, evaporation is limited by conditions and properties of the soil that control the rate at which the soil can provide moisture to the layer where evaporation occurs (Hillel, 1998). A third stage is ultimately established when liquid conduction through the surface is effectively stopped as a result of the surface drying out. Vapor diffusion is the main route of water vapor transport to the soil surface.
After each cultivation, the water vapor flux in MAC 1, MAC 2, RL 1, and RL 2 peaked and then quickly declined without reaching a constant level. Moisture within all the simulators was too low to maintain a constant level of evaporation, as evidenced by the decline in water vapor fluxes after cultivation and the failure to reach a steady state emission rate.
Experiment 2—Simulated Rainfall
Trace gas fluxes measured during four simulated rainfalls are summarized in Fig. 5
. Table 3 summarizes the experiment along with moisture contents. The mass of water added was increased with each addition. The soil in each simulator was cultivated several times.
Total Hydrocarbon Emissions
Addition of water to the simulators resulted in an immediate decrease in THC fluxes by approximately 50% while cultivation resulted in an increase (Fig. 5). Fluxes continued to drop for a period of about 24 h. After each water addition, fluxes dropped to a lower value.
Cultivation of the wet MAC soil resulted in peak emissions 3 to 10 times larger than precultivation emissions. While wetting a soil creates barriers to gaseous transport, disturbing a wet soil reestablishes channels for gas transport to the surface. Peak cultivation emissions were approximately 50% of those observed in Experiment 1 while postcultivation emissions were of similar magnitude. Cultivation after the second water addition resulted in sustained elevated THC emissions that increased for a period of 6 d after cultivation and then began decreasing. After the third water addition, the soil was not cultivated. Fluxes remained low for 16 d, then THC fluxes increased. When the soil was saturated THC fluxes ranged from 0.5 to 4 µg C m-2 s-1.
Refinery landfarm THC fluxes remained low throughout Experiment 2. Fluxes peaked at approximately 2 µg C m-2 s-1 on Day of Year (DOY) 70 after addition of water. Cultivation of wet soil had little or no effect on THC emissions: between DOY 70 and DOY 107 THC fluxes were consistently less than 1 µg C m-2 s-1. After the third and largest water addition on DOY 107, standing water was observed. Excess water was drained after 24 h when it was expected that aggregates would be wetted throughout. In this experiment, fluxes were completely depressed until the soil began to dry and the soil temperature rose between DOY 126 and 131.
Addition of water had no effect on THC fluxes from the DSL soil. The average THC flux for the DSL soil between DOY 70 and DOY 134 was -0.08 µg C m-2 s-1, less than the minimum resolvable flux.
Increased water content in the soil significantly decreases volatilization (Jury et al., 1990). Diffusion of volatile hydrocarbons is significantly lower in water than in air, thus hydrocarbon transport is effectively impeded when soil pores become filled with water. This was observed in the MAC soil. Figure 6a summarizes the effect of soil moisture content, as water-filled porosity (WFP), on THC flux for Experiment 2. Here, a negative correlation between THC fluxes and soil WFP is evident within the four simulators containing contaminated soil.
|
Carbon Dioxide Emissions
The addition of water to the MAC soils resulted in an initial depression of CO2 fluxes followed by an exponential increase to a constant level of emissions (Fig. 5). Carbon dioxide emissions were consistently two to three times higher in Experiment 2 (80 to 120 µg CO2 m-2 s-1) than Experiment 1. Cultivation resulted in elevated CO2 emissions. After the first water addition the soil was cultivated twice. Cultivation did not result in sustained elevated fluxes.
Peak CO2 emissions were largest after cultivating the MAC soil after the second water addition on DOY 83, and CO2 emissions peaked at 378 and 310 µg CO2 m-2 s-1 within MAC 1 and 2, respectively. Elevated CO2 emissions were sustained after cultivation, and they gradually decreased as soil moisture fell (Table 3).
After the third and largest water addition, CO2 fluxes rose to a relatively constant level 5 d after water addition. Emissions were sustained for 12 d at approximately 100 and 50 µg CO2 m-2 s-1 in MAC 1 and 2, respectively. Between DOY 126 and 131 fluxes increased corresponding to an increase in ambient temperature.
In the RL simulators, CO2 fluxes peaked between 190 and 300 µg CO2 m-2 s-1 and then decreased after each of the first two water additions. After the third water addition, fluxes increased slowly, peaked after excess water was drained on DOY 108, and then remained constant. Cultivating wet soil had little effect on CO2 flux. Emissions were 100 to 150 µg CO2 m-2 s-1 between disturbances (i.e., cultivation and water addition), three to five times larger than in Experiment 1.
Delhi sandy loam CO2 fluxes behaved similarly to the MAC soil, although fluxes peaked to lower values after wetting and cultivation. After the second addition of water, fluxes quickly returned to near-zero values.
Carbon dioxide emissions from all simulators were higher in Experiment 2 than in Experiment 1, as the higher moisture content resulted in increased microbial respiration. The magnitude of the observed fluxes was comparable with the published bare soil studies discussed earlier. Cultivation of the wet soil resulted in sustained CO2 fluxes for up to several weeks.
Water is required for microbial processes, although excess water blocks pores and interferes with the availability of oxygen. The optimal soil moisture for aerobic microbial conditions lies between 50 and 80% of the water-holding capacity (Bossert and Bartha, 1984). At 10% of the water holding capacity, metabolic activity becomes marginal. Figure 6b summarizes the effect of soil moisture content on CO2 fluxes. Studies examining the relationship between respiration and moisture content show that CO2 evolution from soil populations increases with increasing moisture content to near saturation (Howard and Howard, 1993). The simulator soils exhibited different optimal soil moisture levels: maximum CO2 production occurred between 40 and 60, 30 and 50, and 60 and 80% of WFP in the MAC, DSL, and RL soils, respectively.
The larger water additions on DOY 80 and 107 resulted in sustained elevated CO2 fluxes, indicating that water availability was previously limited. The pattern of CO2 emissions behaved differently for the two contaminated soils after wetting. Wetting initially depressed fluxes in the MAC soils and elevated fluxes in the RL soils, suggesting that different mechanisms were driving the observed trends. The MAC soil had a larger fraction of small aggregates and therefore, smaller pore spaces between aggregates. The addition of water would effectively block pores initially, restricting aeration and the exchange of gases between the air and soil subsurface. Wong (1999) also noted marked decreases in CO2 fluxes after rainfalls during a study at a different diesel fuel–contaminated soil landfarm. Soil cultivation created channels allowing rapid transfer of CO2 to the surface and a peak in emissions. Cultivation also stimulated respiration in the MAC soils as evidenced by sustained higher fluxes after cultivation.
The RL soil, with its large stable aggregates, had large channels allowing water to pass through freely without completely impeding gas transport. These large pores would always remain air-filled. Microorganisms remained biologically active on the surface of the aggregates until available water became limited. On DOY 107 and 108 when excess water was added resulting in standing water and complete blockage of the large pores, the CO2 flux behaved similarly to the fluxes from the MAC soil, that is, low fluxes followed by an increase. When the excess water was drained, a peak in CO2 emissions followed in the RL soils, a result of submerged soils exposed to air, permitting microbial activity on the surface of the aggregates.
Water Vapor Emissions
Addition of water to the MAC soil resulted in elevated water vapor fluxes of 24 to 29 mg H2O m-2 s-1 (Fig. 5). After the first addition of water (DOY 70), water vapor fluxes remained elevated. Cultivation on DOY 74 resulted in a decrease, probably because water transport channels were disrupted, providing an additional resistance to evaporation. Cultivation on DOY 78 resulted in an increase followed by a rapid decrease, as some moist soil was brought to the surface but soil moisture content was not high enough to sustain elevated evaporation. After the second water addition (DOY 80), fluxes remained constant for 6 d before decreasing. Cultivation of the wet soil on DOY 82 had no detectable effect on water vapor fluxes. After the largest water addition (DOY 107), fluxes remained elevated at approximately 24 mg H2O m-2 s-1 for the duration of the experiment (26 d). Fluxes increased with increased ambient temperatures between DOY 126 and 131.
Refinery landfarm water vapor fluxes peaked between 26 and 31 mg H2O m-2 s-1 after each addition of water. Fluxes decreased to a stable level within 2.5 d after the first water addition. After the second, 8 d passed before water vapor fluxes leveled off. Twenty-five days after the last addition of water, water vapor fluxes began to decrease. Cultivation on DOY 74 and 78 resulted in the same evaporation pattern as the MAC soil.
Delhi sandy loam water vapor fluxes peaked between 26 and 31 mg H2O m-2 s-1 after each addition of water. Sustained high water vapor fluxes were observed after the second and third additions of water.
Evaporation trends behaved differently in the two contaminated soils. After the first two water additions, RL water vapor fluxes dropped quickly after peaking upon water addition and cultivation. The MAC soil maintained a high flux level before decreasing after the second water addition. The rapid decrease in fluxes seen in the RL soils was probably a function of rapid drying of the surfaces of the large aggregates, which were not wetted to their centers. Not until the largest water addition, when the aggregates were wetted throughout, did the flux profiles behave similarly in both the MAC and RL soils.
Mass Balance
Although performing mass balances and linking the contribution of volatile emissions to changes in soil concentrations was not a primary goal of the experiments, a cursory mass balance evaluation was undertaken to highlight difficulties and determine weaknesses in using mass balances. As a first step in performing a mass balance, the cumulative mass loss of THCs, CO2, and water to the air in each simulator during Experiments 1 and 2 was calculated by integrating the area under the flux time series for each simulator.
There were sufficiently accurate water balance data to facilitate evaluation of the flux estimation method. For the water balance, evaporative loss was estimated for each simulator using two methods (Table 4): first, by integrating the water vapor flux time series in Fig. 4 and 5 (Column A in Table 4) and second, through the gravimetric soil moisture content estimates and Eq. [6] (Column B in Table 4):
 | [6] |
|
The cumulative volatile losses of THCs and CO2 are presented in Table 5. There is confidence in the estimates presented in Table 5 since sampling was performed on well-mixed simulator headspaces and the analytical instrumentation was calibrated regularly. On the other hand, analyses for soil carbon (C) and hydrocarbon concentrations required physical removal of a sample with no guarantee that it was representative of the entire soil bed. The change in soil C and hydrocarbon concentration in the simulator soils over the duration of the experiments was small and, for the most part, within the accuracy limits of the concentration measurements. Thus, a statistically significant reduction in soil C and hydrocarbon concentration could not be demonstrated.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The strengths of the simulator facility include capabilities to monitor multiple gaseous fluxes continuously in real-time and to monitor hydrocarbon emissions as THCs. Monitoring THCs continuously in real-time is feasible while it is difficult to do so for speciated hydrocarbons. Total hydrocarbon monitoring provides a better estimate of total volatile hydrocarbon losses.
Estimating gaseous emissions and deposition by monitoring the air above soil provides an average emission for a volume of soil. In this way a heterogeneous volume of soil, more representative of field conditions, was studied. A limitation of trying to maintain a heterogeneous soil environment was that discrete soil sampling for moisture, nutrient, and contaminant concentrations became complicated when attempting to obtain representative samples. This may be remedied by more extensive sampling of the soil volume, but this is costly, time-consuming, and disruptive to the soil environment.
Soils from two active landfarm facilities were studied. Total hydrocarbons, CO2, and water vapor fluxes were monitored from all simulators for a period of 80 d. Few field or lab-scale landfarm studies have simultaneously monitored THC and CO2 emissions. Volatilization of hydrocarbons is often assumed to be negligible or inferred through mass balances. Through continuous monitoring of both THC and CO2, a complete and well-resolved estimate of volatile carbon losses was made.
Continuous monitoring of trace gas emissions provided a detailed time series of fluxes that were used to evaluate the effect of simulated rainfall and cultivation on the flux of THC, CO2, and water vapor from the soils. Knowledge gained from these types of studies can be used to guide management practices at active facilities and the data generated can aid in modeling studies.
From the experiments presented here, a substantial data set was created highlighting the importance of correcting trace gas fluxes from soils for a simultaneous water vapor flux. Corrections can be significant and by overlooking them, trace gas fluxes may be underestimated or an artificial net deposition or downward flux may be measured.
Experiments showed that cultivating dry or moderately wet soil resulted in brief elevations, enhancement by a factor of up to 10, of THC fluxes followed by a sharp decline. This pattern has been noted in the literature. Total hydrocarbon emissions were also briefly enhanced in wet soils, but to a lesser magnitude than in dry soil. At active facilities, cultivating very wet soils may be difficult, but the incorporation of an effective drainage system combined with cultivation would ensure a rapid return to aerobic conditions.
Cultivating dry soil did not enhance respiration. This implies that at active facilities, cultivating dry soil is of no benefit; volatile hydrocarbon and water losses to the atmosphere are enhanced, as are airborne particulates, while biodegradation is not. On the other hand, cultivating wet soil did result in sustained elevated levels of respiration. The subsurface was aerated and channels allowing gaseous exchange were created.
Hydrocarbon fluxes from the refinery landfarm (RL) soil were very low for the duration of the experiments. This, along with field measurements performed by the authors at refinery landfarms, leads to the conclusion that elevated THC fluxes would only be expected during waste application and would rapidly decline to a low emission rate at active facilities similar to those in southern Ontario.
The immediate effect of a simulated rainfall on diesel fuel–contaminated soil (MAC) was to reduce THC emission rates by an order of magnitude. When the wet soil was cultivated, emission rates peaked. If a goal at a landfarm is to reduce THC emissions, wetting a dry soil surface could be an effective, temporary measure. The desire to reduce emissions may stem from odor issues or the desire to minimize the volatile loss of hydrocarbons.
The addition of water had different effects on gaseous fluxes from the different soils. Soil structure and texture should be taken into consideration when developing management strategies at landfarms.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
