Published online November 16, 2005

Continuous Soil Carbon Dioxide and Oxygen Measurements

and Estimation of Gradient-Based Gaseous Flux
Vasile E. Turcu, Scott B. Jones, and Dani Or*

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

ABSTRACT

of surface CO2 flux are typically based on the closedchamber method whereby surface flux is determined
from changes in gas concentration within an enclosure
on the soil surface (de Jong et al., 1979; Cropper et al.,
1985; Drewitt et al., 2002). Commonly used chambers
are portable devices such as the LiCor Li-6200 or Li6400 systems (LiCor Inc., Lincoln, NE), which are capable of measuring soil CO2 fluxes using high accuracy
research-grade instrumentation (Dugas, 1993). The closed
chambers are static or dynamic, where gas is pumped
from the chamber to an external Infra-Red Gas Analyzer (IRGA). Open dynamic systems use a continuous
flow of gas through the chamber and determine soil
CO2 flux by the difference of CO2 concentration at the
inlet to the chamber and the outlet. A comparison of different chamber types (Norman et al., 1997) shows that
systematic errors are associated with all of these different methods and correction factors are needed. Among
the primary limitations of soil chamber measurements
are the lack of continuous observations, manual setup,
and impact on soil surface boundary conditions that
could alter the nature of the diffusive flux (Davidson
et al., 2002). Attempts to improve temporal coverage (de
Jong et al., 1979; Cropper et al., 1985; Freijer and Bouten,
1991) by continuous air pumping from the enclosure to
a gas analyzer resulted in significant alteration of the
soilatmosphere boundary conditions due to variations
in air pressures within the chamber (Lund et al., 1999)
and perturbation of natural conditions on the soil surface (e.g., gas concentration gradients, precipitation,
radiation). In recent years, researchers developed automated surface chamber measurements capable of capturing short-term changes in soil respiration. Such systems were developed for customized experiments (Ambus
and Robertson, 1998; Liang et al., 2003) or by specialized
companies (i.e., LiCor 8500 system). However, these
quasi-continuous systems are still limited to surface CO2
fluxes lacking details regarding subsurface CO2 dynamics. The urgent need for continuous determination of
soil CO2 flux and associated concentration profiles for
extended periods is widely recognized as a key to reliable integration of total CO2 exchange between soil and
the atmosphere (Parkin and Kaspar, 2004).
The gradient-based approach demonstrating the feasibility of gaseous flux estimations from in situ CO2 and
O2 concentrations and water content measurements was
described previously (Mitchell et al., 1999; Jones et al.,
2000). The availability of new fast-response sensors provides continuous measurements of soil CO2 concentration and enables dynamic observation of soil CO2 evolution with respect to ambient variables such as soil
temperature (Tang et al., 2003).

The magnitude and dynamics of soil gaseous fluxes play critical

roles in the global gas balance; yet, these processes are not captured at
sufficient temporal resolution by standard methods based on periodic
sampling and surface chamber measurements. A novel method for
continuous measurement of soil surface gas fluxes based on subsurface
CO2 and O2 concentration gradient measurements was developed. We
tested the gradient-based method under steady- and transient-state
soil water content and temperature conditions and compared results
with a state-of-the-art surface chamber CO2 flux system. The new aspects of the method include fast-response sensors installed in the soil
profile providing continuous record of concentration gradients coupled
with concurrent estimates of water content-dependent gaseous diffusion
coefficient enabling calculation of surface gaseous fluxes. Low-cost infrared sensors were used for CO2 concentration measurements, and galvanic cells for O2 measurements. An imposed CO2 concentration gradient in a dry soil column resulted in a quasilinear CO2 concentration
profile and surface CO2 flux in agreement with chamber-measured
fluxes. A series of continuous concentration measurements under variable water content conditions and wetting events showed agreement
with surface chamber measurements. Within several days of surface
wetting, soil CO2 concentrations attained 10 mL L1, one order of
magnitude higher than ambient concentrations, whereas O2 concentrations decreased. The gradient-based approach minimizes soil surface
perturbations and provides insights into subsurface soil CO2 and O2
dynamics and the distribution and magnitude of soil respiration processes as related to soil environmental factors. The subsurface gradientbased measurement system represents an order-of-magnitude reduction in cost compared with research-grade surface chamber devices.

t has been estimated that globally, soil contains approximately 2.6 1029 prokaryotic cells (compared
with 1.2 1029 in the open ocean water and sediments),
concentrated in a relatively small volume on the earth
skin (soil volume 1.2 1014 m3 vs. 1020 m3 for open ocean),
making the unsaturated zone the richest compartment
of prokaryotic life on Earth (Whitman et al., 1998). This
abundance of microbial life combined with all higher
plants, makes soil the major component in the ecosystem carbon balance, with a carbon stock of 2 Terra-tons
(Grace, 2001).
The primary methods for gaseous measurements within
the soil include soil air sampling at different depths (Buyanowski and Wagner, 1983) and laboratory analysis of
soil core samples (Cortassa et al., 2001). Measurements
V.E. Turcu, S.B. Jones, Department of Plants, Soils, and Biometeorology, Utah State University, Logan, UT 84322; D. Or, Department
of Civil and Environmental Engineering, University of Connecticut,
Storrs, CT 06269. Received 18 Nov. 2004. *Corresponding author
(dani@engr.uconn.edu).
Published in Vadose Zone Journal 4:11611169 (2005).
Original Research
doi:10.2136/vzj2004.0164
Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

Abbreviations: IRGA, Infra-Red Gas Analyzer.

1161

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

1162

VADOSE ZONE J., VOL. 4, NOVEMBER 2005

The objectives of this study were (i) to introduce and

test a new approach of the gradient-based measurement
method for continuous monitoring of temporal and vertical variations of soil CO2 and O2 concentrations driven by
diurnal soil temperature variations and changes in water
content and (ii) to compare the gradient-based soil CO2
flux determination with standard soil-chamber measurements. In the proposed method, CO2 and O2 sensors
are installed across the soil profile, providing continuous
measurements. Additionally, the method is capable of
capturing and estimating near-surface CO2 and O2 fluxes
without altering soil surface conditions, when measured
concentration gradients are coupled with information
regarding soil diffusivity. The work presented here highlights the advantage of this continuous method for surface CO2 and O2 flux estimation and discusses the drawbacks and uncertainties in the methodology.
Imposing a constant CO2 concentration gradient at
the two ends of an air-dry soil column, where microbial
activity can be assumed negligible and soil gas diffusivity constant, should lead to a linear distribution of CO2
concentration within the profile. We used this simple test
to observe the comparison between the gradient-based
CO2 flux and the soil-chamber flux under steady-state
conditions. Following this, we tested the gradient method
in moist soil, since soil water content plays an important
role in enhancing CO2 production in soils (Wildung et al.,
1975; Zak et al., 1999). In addition to increasing microbial activity, wetting processes can result in a decrease
of soil air-filled porosity and consequently a reduction
in the soil gaseous diffusion coefficient. Hence, increased
CO2 production in the subsurface may not be immediately observable at the soil surface because of reduced
gaseous diffusion. Accurate assessment of soil gas diffusivity is the main drawback of using concentration gradients for calculation of surface flux, especially during
transient water content conditions. The highly dynamic
nature of soil water content during wetting may generate
a nonuniform profile of the air-filled porosity with depth.
Therefore, estimation of soil gas diffusivity from an average air-filled porosity is not adequate. We suggest it
is critical to determine air-filled porosity for each discrete soil layer, where the wettest horizontal layer, having the lowest air-filled porosity, dictates the limiting
gas diffusion coefficient for the whole soil profile. In the
section on theory below, we will introduce the harmonically averaged air-filled porosity for the soil profile. This
approach provides the appropriate estimation of soil
diffusivity, yielding the best estimate of surface CO2 flux.
We compared these fluxes with soil chambermeasured
CO2 fluxes at the soil surface to demonstrate the reliability of the gradient-based approach.
Next, we present basic theoretical considerations and
key assumptions, followed by a description of the experimental setup and methodology and a discussion of measurements obtained using the gradient-based and standard soil-chamber methods.

THEORETICAL CONSIDERATIONS
Diffusion along concentration gradients is the primary mechanism for gaseous transport in soils. In most

soils, pressure gradients are negligible means of gaseous

transport (Glinski and Stepniewski, 1985; Hillel, 1998).
Thus, measurement of CO2 and O2 concentration profiles in soil may be used to estimate gaseous fluxes (de
Jong and Schappert, 1972) according to Ficks Law:
J Ds

dC
C
Ds
dz
z

[1]

where J is the flux of gas species, Ds Ds(,) is the

soil gas diffusion coefficient that varies with soil porosity
and volumetric water content , C is the gas concentration, and z is depth. For flux determination, the gradient
is approximated by discrete differences C and z.

Soil Gaseous Diffusion Coefficient

For a dry porous medium, the gaseous diffusion coefficient is often expressed as a function of the porosity
of the medium, such as in the expression proposed by
Millington (1959):
Ds Da4/3

[2]

where Da is the diffusion coefficient of gas species in

free air, where Da 1.64 105 m2 s1 for CO2 and Da
1.98 105 m2 s1 for O2 (Hillel, 1998) at standard conditions of temperature (25C) and pressure (1025.13 kPa).
Variations of Da with temperature and molar fraction
(Jaynes and Rogowski, 1983) are neglected in this study.
For wet soils, the fraction of air-filled porosity determines the soil gaseous diffusion coefficient. Many expressions have been proposed to relate soil water content
or air content to gaseous diffusion, and a recent review
of field and laboratory determined diffusivities (Werner
et al., 2004) pointed to the following simple relationship
proposed by Moldrup et al. (2000) as the best predictor
of gaseous diffusion coefficient as a function of air-filled
porosity not only for sieved and repacked soils but also
for in situ measurements:
Ds Da

( )2.5

[3]

An important aspect associated with measurements

and calculations of gaseous fluxes in field soils is the role
of spatial variations in textural properties in a soil profile
and nonuniform vertical distribution of soil water content. Variations in these attributes, especially soil water
content, which is highly variable but assumed uniform
horizontally, affect the calculations of an effective gaseous diffusion coefficient for flux estimation in a given
soil layer. The effective diffusion coefficient for a layered
nonuniform soil profile follows a similar calculation as
for the effective hydraulic conductivity for flow perpendicular to layering based on harmonic averaging of individual conductivities (or diffusivity) of each layer:
n

Ds

zk

k1

zk
Dsk(k)
k1
n

[4]

where Ds is the equivalent (effective) soil diffusion coefficient for the entire spatial domain, Dsk(k) represents

1163

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

www.vadosezonejournal.org

Fig. 1. Detailed view of sensor arrangement showing the combination of CO2O2 sensors and the thermocouple inserted into the
soil profile.

soil gaseous diffusivity for a discrete layer k of thickness

z, and water content k; n is the total number of layers
within the domain (for soil porosity that varies with
depth, Dsk(k,k) is required in Eq. [4]). Note that this
necessitates more detailed information regarding water
content vertical distribution than is provided by simple
mass balance models or by vertically averaged water
content measurements.
MATERIALS AND METHODS
A new generation of commercially available sensors enables
simple and reliable in situ measurements of CO2 and O2 concentrations. For CO2 measurements, we used Vaisala GMD20
analyzers (Vaisala Inc., Woburn, MA) based on a Single-Beam
Dual Wavelength infrared absorption method and silicon
CARBOCAP sensor technology (Fig. 1). Each signal pulse is
compared with a reference signal obtained under the same
conditions of temperature and humidity, ensuring a response
independent of ambient conditions. The sensors provide a
linear response within 1 min of variation in CO2 concentration
over a preset range (for dry soils we used a calibration span
of 05000 L L1, whereas for moist soils we used 020 000 L
L1) with an accuracy of 30 L L1 2% of reading. The
choice of sensor calibration was guided by the depth of sensor
installation and expected moisture content. A stable 24-V
power supply is required by the sensors that output an analog
voltage or current. The sensors relatively small dimensions
(length 140 mm; diameter 15 mm) facilitates simple deployment within a soil profile. Oxygen concentrations were
measured using galvanic cell sensors KE-25 (Figaro Inc., Glenview, IL). These low cost, long-life (5 yr) sensors require no
power input and provide a rapid (12 s) and linear analog
response to variations in oxygen concentration. Independent
measurements of soil surface CO2 fluxes were obtained using
a standard Li6400 portable soil chamber (Li-Cor Inc.). A PVC
collar installed directly above the subsurface sensors in the
soil surface provided physical support of the soil chamber
during measurements. Soil water content was determined with
a CS-615 reflectometer (Campbell Scientific, Logan, UT),
which averaged water content over the length of the 30-cm
rods. The sensor was calibrated using gravimetric water content measurements in the Millville silt loam soil (coarse-silty,
carbonatic, mesic Typic Haploxeroll).
We compared two methods for estimating soil gas diffusivity
for the 0- to 8-cm soil layer. First, using Eq. [3], we obtained a
diffusivity value based on an average water content for the soil
layer as measured with the reflectometer. This value was inserted in Eq. [1] to yield an average surface CO2 flux denoted
as Jav. In the second approach, we used values of water content
for each 2-cm soil layer to determine harmonically averaged

Fig. 2. Laboratory soil column showing experimental setup for steadystate CO2 flux determinations. Measured surface soil chamber fluxes
were compared with calculated fluxes obtained from the gradient
method (Table 1).

diffusivity using Eq. [4] with the resulting surface CO2 flux denoted as Jh. For this situation we assumed a binary gas system.

Steady-State CO2 Flux Measurements in Soil Columns

For preliminary tests of the gradient method we used a dry
soil column uniformly packed with Kidman soil (coarse-silty,
carbonatic, mesic Typic Haploxeroll). The bottom boundary
of the column was maintained using a constant, low pressure
(10 Pa) flow rate of calibration-grade CO2 exiting through
a large diameter (2.5 cm) outlet tube. The exhaust tube exited
the room to avoid buildup of CO2 in the laboratory. The fixed
CO2 concentration gradient resulted in steady-state flux through
the soil column. The experimental setup is depicted in Fig. 2,
where the horizontal placement of two CO2 sensors inside the
sand column is shown at 9 and 24.5 cm below the surface,
respectively. Measurements were made for two different CO2
concentrations at the bottom of the soil column, each leading
to different concentration profiles in the column and different
CO2 flux at the surface (Table 1). For each case, the surface
CO2 flux was estimated based on the concentration gradient
and diffusion coefficient in dry sand and was periodically and
independently measured using the soil chamber. To minimize
perturbation at the surface of the soil column, we installed
the soil chamber for only short measurement periods lasting
5 min. Gas diffusion coefficients were determined based on
soil porosity using Eq. [2] and [3] and assuming that water
content was negligible in the air-dried soil.

Dynamic CO2 Flux Measurements

in Greenhouse Soil Columns
A large container 1.20 m tall with a horizontal cross-sectional area of 0.6 m2 was filled with air-dry Millville silt loam

1164

VADOSE ZONE J., VOL. 4, NOVEMBER 2005

Table 1. Carbon dioxide measurements in a column of dry sandy

soil under steady-state conditions.

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

Depth

CO2
concentration

cm

L L1

0
9
24.5
30

400
1544
3200
3905

0
9
24.5
30

400
1380
2570
3190

Diffusivity

Gradient
method flux

Soil
chamber flux

m 2 s1
Case 1
6.51E-06

mol m2 s1
2.61

2.79

Case 2
6.51E-06

2.04

2.49

Calculated with Eq. [2].

soil. A 25-cm-diameter PVC pipe was placed on the soil surface, and the soil was carefully excavated from the circular
area delimited by the inner edge of the pipe, leaving the surrounding soil with minimal disturbance. The pipe was driven
gradually into the soil during excavation until it reached the
desired depth of 30 cm. A plastic disc was then placed at
the bottom of the cavity and sealed with silicone. A 2.5-cmdiameter auger was used to bore a horizontal sensor cavity
into the soil through the predrilled holes in the large PVC
pipe wall where sensors were installed (Fig. 3). In 2001, for
experiments with two sets of sensors (11 d), these were installed at depths of 8 and 22 cm. In 2002 (7 d), we used three
sets of sensors installed at 4, 8, and 22 cm. Each measurement
depth was instrumented with a pair of CO2 and O2 sensors.
The sensors were mounted end to end inside of a PVC tube
(2.5 cm in diameter and 20 cm long) with a gap between gas
entry ports and a series of slits in the bottom of the PVC tube
to allow unhindered gas exchange as seen in Fig. 1. A copperconstantan thermocouple was enclosed within the same head
space for soil air temperature measurement. During the experiments, the nonvegetated soil was wetted periodically by applying known amounts of water at the soil surface (12 mm
per application). Sensor outputs were logged by a CSI-21x
datalogger (Campbell Scientific) and stored as 10-min averages. A collar for closed-chamber periodic measurements was
installed in the soil surface directly above the gradient-based
sensor bank. Surface CO2 efflux was measured daily with the

closed-chamber device for comparison to fluxes determined

from the gradient-based approach.
The gradient-based CO2 surface flux was determined using
Ficks first law of gas diffusion presented above. The nearsurface atmospheric CO2 concentration can be affected by
advective, diffusive, and thermally induced processes. Several
measurements were made to determine the aboveground CO2
concentration in atmospheric air, using the Li-Cor 6400 IRGA.
The soil chamber was laid on its side on the soil surface, and
concentration of CO2 entering the chamber was monitored
(LI-6400 instruction manual). The concentrations averaged
380 20 L L1 for low soil water contents and 380 50 L
L1 for wetter soil. These variations in surface CO2 concentration equate to a 1% error in CO2 flux calculation. We assumed that the CO2 concentration at depth z 0 was equal
with the concentration in the soil chamber during these initial
measurements. The CO2 concentration at depth z, Cz, was
obtained via direct measurement with the sensors described
previously. For this study, all the surface flux data presented
were obtained by calculating the concentration gradient between the surface and a depth of 8 cm. The information from
sensors at deeper depths was used to observe subsurface CO2
concentration dynamics. When concentrations lower in the
profile fall below near-surface peak measurement, downward
flux of CO2 is likely to occur.
We used the numerical simulation model Hydrus-2D (Simunek et al., 1999) to determine vertical and temporal water
content distributions, subject to known water flux input at the
surface and predefined soil hydraulic properties given in Mmolawa and Or (2003). The simulation domain was considered
a one-dimensional vertical soil column, of 70-cm depth, with
uniform soil characteristics, and with the observation nodes
for simulation outputs at 1, 3, 5, , 19 cm below the soil
surface. The lower boundary condition was set as a free drainage condition while the upper boundary condition for water
flow was specified as a variable flux condition (e.g., irrigation,
evaporation), where the flux of water was obtained from the
wetting record during the experiments. The entire soil surface
was uniformly sprinkled with water using a 1-L volume. Each
wetting event recorded included the amount of water per unit
area per time of wetting providing the input surface water
flux information for both 2001 and 2002 experiments.

RESULTS AND DISCUSSION

Steady-State CO2 Flux Measurements
in Soil Columns
Imposing a constant CO2 concentration gradient at
the two ends of an air-dry soil column (where microbial
activity was assumed negligible) should produce a linear
gradient CO2 concentration within the profile as confirmed by the results in Fig. 4. The gradient-based flux
estimates were in good agreement with soil chamber
measurements obtained at the soil surface (Table 1).
These simple tests in a dry soil (uniform diffusion coefficient) with negligible CO2 production demonstrate the
gradient-based methods ability to provide estimates of
gaseous flux based on soil CO2 concentration measurements and known concentrations within the soil profile.

Fig. 3. Cut-away of the experimental apparatus showing the sealed

25-cm-diameter PVC pipe used to house the sensors and datalogger. The soil chamber collar is placed directly above the gradientbased sensors, and a soil moisture sensor installed obliquely averages water content over the 0- to 15-cm depth.

Dynamic CO2 Flux Measurements

in Greenhouse Soil Columns
At the beginning of the 2001 and 2002 greenhouse experiments, CO2 and O2 concentrations in air-dried and

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

www.vadosezonejournal.org

1165

CO2 concentrations in the wet soil profile were similar

at the 4- and 8-cm depths, resulting in only a small CO2
concentration gradient between these two depths. This
suggests the possibility of a broad layer of maximum
CO2 production located within and perhaps surrounding the 4- to 8-cm depth. More detailed insight into this
active zone would require a concentrated sensor array
to better characterize the shape of the concentration profile and the resulting gradients driving diffusion both
upward and downward.

Effects of Variations in Water Content

on Soil Gaseous Diffusion Coefficient

Fig. 4. Soil CO2 concentration profile measured in the laboratory

under steady-state conditions in dry Kidman soil.

repacked Millville silt loam soil remained near atmospheric values 1 wk before the first wetting event. Consistent with the assumption of negligible microbial CO2
production in dry soil, attempts to measure with the soil
chamber confirmed that CO2 efflux was below detection
limits during the same period. Upon wetting of the soil
surface, the concentration of soil CO2 increased rapidly
(Fig. 5a and 5b) whereas O2 concentration dropped, as
depicted in Fig. 5c and 5d. These changes in CO2 and O2
concentration profiles are indicative of increased microbial respiration in the wet soil. The CO2 and O2 concentration profile became nonlinear on wetting, with the
measured maximum CO2 concentration associated with
the lowest O2 concentration. The increase of CO2 and
decrease of O2 concentrations during the two experiments (Fig. 5a5d) show a time delay at the 22-cm depth.
This resulted from the retarded arrival of the wetting
front that creates a large difference in soil water content
from top to bottom. This creates a discrepancy in CO2
measured concentration at different depths. In the 2001
experiment, CO2 concentration at the 8-cm depth exceeded 9000 L L1, compared with normal atmospheric
concentrations of 380 L L1. The oxygen in the same
soil layer was depleted to 180 mL L1 from its initial
(atmospheric) value of 209 mL L1. These values indicate the presence of a strong concentration gradient
from the 8-cm layer toward the surface with upward
flux of CO2 and downward O2 flux. Interestingly, sensors
at 22 cm below the soil surface show lower CO2 concentrations and higher O2 concentrations than the 8-cm
measurements, indicating the majority of microbial activity (CO2 source and O2 sink) occurred between the
surface and the 22-cm depth. The resulting CO2 fluxes
were driven both downward and upward from this optimal layer (e.g., wetter and warmer), with O2 being consumed from both directions. With the addition of a
sensor at 4 cm in the 2002 experiment, we observed that

Water was applied on the soil surface in pulses as shown

in Fig. 6a (for 2001) and 6b (for 2002). The amounts and
application times were used as inputs for detailed simulations of water content distribution in the profile using
the Hydrus-2D model. Simulated water content is presented in Fig. 6a and 6b for layers at depths of 0 to 2,
6 to 8, and 18 to 20 cm. Measured (average) soil water
content is also shown for the 0- to 15-cm profile for comparison. Each water application is marked by a sharp
increase in (simulated) surface layer water content followed by internal drainage into deeper layers. Vertically
averaged water content measurements tend to smear
the sharp increase in near-surface water content that
acts as a throttle in controlling gaseous diffusion into
and out of the soil. The importance of surface wetness
in controlling diffusion processes is illustrated next.
No water was applied between Days 4 and 7 in the
2001 experiment, allowing internal water redistribution
and a decrease in surface water content. During this time,
a decrease in the 8-cm CO2 concentration occurred, in
parallel with an increase in O2 concentration at the same
depth. These changes did not correlate with the steady
increase in temperature (the CO2 decrease and O2 increase lasted for 2.5 d, while temperature oscillations are
based on a daily cycle). These changes are better explained by higher surface layer gas diffusivity as the
water evaporated and drained to deeper layers. Increased
surface diffusivity allows a larger gas flux across the soil
atmosphere interface, reducing the stored CO2 from
microbial production and increasing the supply of oxygen
in the soil air (aeration). When wetting applications were
resumed, a sharp decrease in surface gas diffusivity was
followed by a buildup of soil CO2 concentration and a
decrease of O2. A similar situation can be observed in
the 2002 data, between Days 3 and 4 of the experiment,
when no water was applied for more than 24 h.
Surface CO2 flux, Jav, was obtained using the average
soil profile water content and is depicted in Fig. 6e and 6f
(dashed line). The correlation between Jav and the soil
chamber measured surface fluxes (circles) was poor, where
Jav was about three times higher than the total CO2 fluxes
measured with the soil chamber. This flux discrepancy is
much larger than errors associated with the soil-chamber
measurements. We reevaluated the possibility that soil
gaseous diffusivity was overestimated using the average
bulk air-filled porosity in calculations, disregarding spatial distribution of water in the soil profile. Gas diffusiv-

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

1166

VADOSE ZONE J., VOL. 4, NOVEMBER 2005

Fig. 5. Greenhouse measurements of (a, b) CO2 and (c, d) O2 gas concentrations and (e, f) soil temperatures measured at two or three depths
in the soil profile, during the 2001 and 2002 greenhouse experiments. In 2001, between Days 4 and 7, no water was applied.

ity is controlled by the wettest soil layer, suggesting that

harmonic averaging for diffusivity values would be required (Eq. [4]). Harmonically averaged diffusion coefficients were used to recalculate the surface CO2 flux Jh
shown in Fig. 6c and 6d (thin dotted line). The harmonically averaged flux obtained using the same diffusion
model but accounting for a layered water distribution
is in better agreement with soil chamber surface CO2
fluxes during both years of measurements where peak
fluxes occur just before irrigation. The frequent application of water resulted in a highly dynamic soil CO2
surface flux (Jh) that is not captured by the soil-chamber
measurements, whereas the gradient method provides
the missing flux information between consecutive soilchamber measurements. For example, Fig. 6 shows the
linkage between the dynamics of soil gaseous diffusivity
and surface wetting and their impact on surface CO2
and O2 flux. We used flux values from Fig. 6f to calculate
the cumulative C evolved from the soil surface during
about 5 d via the gradient-based and soil-chamber meth-

ods. Integration of Jh, which captured the sudden reductions in flux following each surface wetting event,
yielded 283 mmols CO2 m2. Considering the soil-chamber measurements, nearly 50% more C (411 mmols CO2
m2) was evolved. This suggests that for cases when
soil-chamber measurements are made during peak flux
periods (e.g., during high soil temperature or before
irrigation) the total soil C balance may be overestimated
if additional nonpeak periodic measurements are not included.
Near-surface oxygen flux was determined using the
gradient method, based on similar principles, with data
collected continuously at the 8-cm depth. Harmonically
averaged air-filled porosities were used to determine
the diffusion coefficient as for CO2, multiplied with a factor of 1.21 to account for the difference in Da values
between CO2 and O2. The flux of oxygen into the soil
(Fig. 6g and 6h) was much larger than CO2 flux out
because of a larger gradient of oxygen recorded in the
same soil layer (08 cm). This discrepancy may be par-

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

www.vadosezonejournal.org

1167

Fig. 6. Measured (015 cm, dashed line) and Hydrus2D simulated (i.e., for 02, 68, and 1820 cm depth) (a, b) soil water content and (c, d)
estimated soil gas diffusivity for the 2001 and 2002 greenhouse experiments. (e, f) Surface CO2 flux determined using the gradient-based
method (lines) and periodic soil chamber measurements (circles) are shown for both years. (g, h) Calculated near-surface O2 flux.

tially explained by a larger solubility of CO2 than O2 in

soil water (28 times higher for CO2; Paul and Clark,
1996) resulting in a smaller CO2 concentration gradient
(Cortassa et al., 2001). Unfortunately, there was not
an independent oxygen flux measurement available for
comparison with the gradient method. Therefore, it is
difficult to verify the gradient-based flux determination. Nevertheless, the estimation of O2 flux is valuable
for soil-respiration studies (e.g., establishing respiratory
quotient).
A summary of comparisons between surface CO2
fluxes determined using the gradient-based method and
closed-chamber measurements (Fig. 7) shows reasonable
correlation between the two methods. The three outlier
values of CO2 flux measured with the soil chamber at

the end of the 2001 experiment were omitted in this graph

because of suspected equipment failure during those measurements. The high surface CO2 flux measured at the
beginning of the 2002 experiment (Fig. 6f) immediately
after the initial wetting could be attributed to the effect
of CO2 desorption from soil during water intrusion. With
a tendency for underprediction of surface chamber measured fluxes (de Jong et al., 1979), the gradient-based
method is in need of further study. The method offers a
viable and valuable alternative to closed-chamber measurements by providing detailed temporal and spatial
(soil profile) information on CO2 and O2 concentrations
not available with surface chamber measurements. Our
analysis of the gradient-based approach suggests successful application of this technique requires detailed

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

1168

VADOSE ZONE J., VOL. 4, NOVEMBER 2005

soil profile. Consequently, information regarding soil

microbial activity is not limited to CO2 production only,
since consumption of O2 can also be observed yielding
possible insights regarding soil respiration type (i.e., aerobic or anaerobic). The measurement of continuous
subsurface gas concentration using this new method is
a valuable tool for calibration of soil CO2 transport and
production models and for improved understanding of
biogeochemical phenomenology in the soilatmosphere
continuum. Accurate estimates of soil gas diffusivity required for gas flux calculation hinge on accurate spatial
measurement of soil water content within the profile.
We envision that the availability of more accurate and
lower cost water content sensors installed adjacent to
the gas sensor array will improve the accuracy and usefulness of the gradient-based method. Additionally, for
known surface inputs (e.g., precipitation, evaporation),
modeling coupled with measurements should improve
surface CO2 flux estimates.
ACKNOWLEDGMENTS
Fig. 7. A 1:1 comparison of the soil surface CO2 flux measurement
using the closed-chamber method vs. the proposed gradientbased method.

information regarding water content distribution within

the soil profile.

This research was funded by the National Science Foundation project DEB-9807097 and supported by the Utah Agricultural Experiment Station under UAES paper no. 7622. The
authors thank Bill Mace, Brent Bingham, and Alan Mitchell
for their assistance in this work.

REFERENCES

CONCLUSIONS
A novel gradient-based measurement method for soil
CO2 and O2 concentration and flux was developed and
tested. The new aspects of this method are the use of
newly available sensors capable of measuring continuous soil CO2 and O2 concentrations in situ, with minimum soil disturbance and detailed consideration of soil
gas diffusivity for use in surface CO2 flux estimation. As
a supplement and potential alternative to the standard
soil surface chamber technique, the gradient method offers several advantages, including continuous monitoring
and insights into the subsurface CO2 and O2 concentration
dynamics and spatial distribution. Simultaneous measurements of oxygen concentrations and soil temperature at the same depth provide useful information about
the correlation between soil respiration and temperature. A continuous surface CO2 flux determination is
particularly important for capturing short-duration and
often large CO2 fluxes occurring after rapid wetting (e.g.,
rainfall events in arid zones). Considering the minimal
impact on surface conditions using the gradient method,
surface CO2 fluxes are more reliably related to dynamics
of environmental factors such as water content and temperature. However, abrupt changes in the near-surface
soil water content above the top sensor may not be captured using the gradient method. The reduction in nearsurface diffusivity and commensurate buildup of subsurface gaseous concentrations are more likely to be
observed by the gradient method. The gradient method
can be used for in situ continuous monitoring of soil
respiration dynamics because it provides continuous and
simultaneous data on CO2 and O2 concentrations in the

Ambus, P., and G.P. Robertson. 1998. Automated near-continuous

measurement of carbon dioxide and nitrous oxide fluxes with a
photoacoustic infra-red spectrometer and flow-through soil cover
boxes. Soil Sci. Soc. Am. J. 62:394400.
Buyanowski, G.A., and G.H. Wagner. 1983. Annual cycles of carbon
dioxide level in soil air. Soil Sci. Soc. Am. J. 47:11391145.
Cortassa, S., M.A. Aon, and P.F. Villon. 2001. A method for quantifying rates of O2 consumption and CO2 production in soil. Soil
Sci. 166:6877.
Cropper, W.P., K.C. Ewel, and J.W. Raich. 1985. The measurement
of soil CO2 evolution in situ. Pedobiologia 28:3540.
Davidson, E.A., K. Savage, L.V. Verchot, and R. Navarro. 2002.
Minimizing artifacts and biases in chamber-based measurements
of soil respiration. Agric. For. Meteorol. 113:2137.
de Jong, E., R.E. Redmann, and E.A. Ripley. 1979. A comparison
of methods to measure soil respiration. Soil Sci. 127:300306.
de Jong, E., and H.J.V. Schappert. 1972. Calculation of soil respiration
and activity from CO2 profiles in the soil. Soil Sci. 113:328333.
Drewitt, G.B., T.A. Black, Z. Nesic, E.R. Humphreys, E.M. Joek,
R. Swanson, G.J. Ethier, T. Griffis, and K. Morgenstern. 2002.
Measuring forest floor CO2 fluxes in a Douglas-fir forest. Agric.
For. Meteorol. 110:299317.
Dugas, W.A. 1993. Micrometeorological and chamber measurements
of CO2 flux from bare soil. Agric. For. Meteorol. 67:115128.
Freijer, J.I., and W. Bouten. 1991. A comparison of field methods for
measuring soil carbon dioxide evolution: Experiments and simulation. Plant Soil 135:133142.
Glinski, J., and W. Stepniewski. 1985. Soil aeration and its role for
plants. CRC Press, Boca Raton, FL.
Grace, J. 2001. Carbon cycle. p. 609628. In Encyclopedia of biodiversity. Vol. 1. Academic Press, New York.
Hillel, D. 1998. Environmental soil physics. Academic Press, New
York.
Jaynes, D.B., and A.S. Rogowski. 1983. Applicability of Ficks law
to gas diffusion. Soil Sci. Soc. Am. J. 47:425430.
Jones, S.B., B.W. Bingham, and D. Or. 2000. Measurement of soil
CO2 gradient for in-situ estimation of carbon fluxes. In Annual
Meetings Abstracts. ASA, CSSA, and SSSA, Madison, WI.
Liang, N., G. Inoue, and Y. Fujinuma. 2003. A multichannel auto-

Reproduced from Vadose Zone Journal. Published by Soil Science Society of America. All copyrights reserved.

www.vadosezonejournal.org

mated chamber system for continuous measurement of forest soil

CO2 efflux. Tree Physiol. 23:825832.
Lund, C.P., W.J. Riley, L.L. Pierce, and C.B. Field. 1999. The effects
of chamber pressurization on soil-surface CO2 efflux and the implications for NEE measurements under elevated CO2. Global
Change Biol. 5:269281.
Millington, R.J. 1959. Gas diffusion in porous media. Science (Washington, DC) 130:100102.
Mitchell, A.R., D. Or, M. Caldwell, and S.B. Jones. 1999. Gradient
method for in-situ measurement of CO2 Flux. In Annual Meetings
Abstracts. ASA, CSSA, and SSSA, Madison, WI.
Mmolawa, K., and D. Or. 2003. Experimental and numerical evaluation of analytical volume balance model for soil water dynamics
under drip irrigation. Soil Sci. Soc. Am. J. 67:16571671.
Moldrup, P., T. Olesen, J. Gamst, P. Schjonning, T. Yamaguchi, and
D.E. Rolston. 2000. Predicting the gas diffusion coefficient in repacked soil: Water-induced linear reduction model. Soil Sci. Soc.
Am. J. 64:15881594.
Norman, J.M., C.J. Kucharik, S.T. Gower, D.D. Baldocchi, P.M. Crill,
M. Rayment, K. Savage, and R.G. Striegl. 1997. A comparison
of six methods for measuring soil-surface carbon dioxide fluxes.
J. Geophys. Res. 102:28,77128,777.
Paul, E.A., and F.E. Clark. 1996. Soil microbiology and biochemistry.
2nd ed. Academic Press, San Diego, CA.

1169

Parkin, T.B., and T.C. Kaspar. 2004. Temporal variability of soil carbon dioxide flux: Effect of sampling frequency on cumulative carbon loss estimation. Soil Sci. Soc. Am. J. 68:12341241.
Simunek, J., M. Sejna, and M.Th. van Genuchten. 1999. The Hydrus-2D
software package for simulating the two-dimensional movement
of water, heat, and multiple solutes in variably saturated media.
IGWC, Riverside, CA.
Tang, J., D.D. Baldocchi, Y. Qi, and L. Xu. 2003. Assessing soil CO2
efflux using continuous measurements of CO2 profiles in soils with
small solid-state sensors. Agric. For. Meteorol. 118:207220.
Werner, D., P. Grathwohl, and P. Hohener. 2004. Review of field
methods for the determination of the tortuosity and the effective gas-phase diffusivity in the vadose zone. Available at www.
vadosezonejournal.org. Vadose Zone J. 3:12401248.
Whitman, W.B., D.C. Coleman, and W.J. Wiebe. 1998. Prokaryotes:
The unseen majority. Proc. Natl. Acad. Sci. USA 95:65786583.
Wildung, R.E., T.R. Garland, and R.L. Buschbom. 1975. The interdependent effects of soil temperature and water content on soil respiration rate an plant root decomposition in arid grassland soils. Soil
Biol. Biochem. 7:373378.
Zak, D.R., W.E. Holmes, N.W. MacDonald, and K.S. Pregitzer. 1999.
Soil temperature, matric potential, and the kinetics of microbial
respiration and nitrogen mineralization. Soil Sci. Soc. Am. J. 63:
575584.