Abstract

Simultaneous deployments of two parallel closed dynamic chamber soil respiration measurement systems were conducted to compare internal chamber pressure and soil carbon dioxide efflux in response to fan-controlled and natural wind movement. When sealed to an impermeable plexiglass plate, the chamber systems demonstrated negative internal chamber pressurization in response to horizontal wind movement. However, horizontal wind-driven chamber pressurization was negligible in the chamber systems when deployed on soil. Under both fan-controlled and natural wind conditions, increased soil-atmosphere carbon dioxide fluxes corresponding to horizontal wind events were observed. The carbon dioxide flux increased approximately 20 percent (above non-fan interval fluxes) for every meter per second increase in horizontal wind velocity during controlled-wind fan trials. During natural wind condition trials, carbon dioxide flux increases, ranging from 42-121 percent, were observed corresponding to horizontal wind gust velocities less than 1 m s-1 in magnitude. The experimental results suggest that the closed dynamic chamber method may systematically overestimate soil carbon dioxide flux in response to the selective condition of horizontal wind movement.

Introduction

1.1 Carbon Dioxide

The Earth’s atmosphere functions in its radiative balance as a thermal boundary layer, absorbing infrared radiation re-emitted from the Earth’s surface resulting from solar radiation influx. Carbon dioxide (CO2) and other greenhouse gases are the primary absorbers of this infrared radiation. Atmospheric CO2 levels have been increasing since the Industrial Revolution largely due to anthropogenic manipulation of the carbon cycle. The global carbon cycle determines the relative partitioning of carbon into its global reserves, including atmospheric CO2, soil carbon, and terrestrial organic biomass. The primary mechanism of anthropogenic carbon release is through fossil fuel combustion, which releases inorganic carbon stored in the lithosphere to the atmosphere in the form of CO2. The increasing level of CO2 in the atmosphere has the potential to affect global climate by shifting the Earth’s radiative balance, with unknown climatic effects.

Approximately half of anthropogenic CO2 has been re-partitioned into non-atmospheric reserves by global carbon cycling, including carbon sequestration in the oceans and terrestrial biosphere. During the 1990s, the terrestrial biosphere removed approximately 1.4 gigatonnes of carbon per year from the atmosphere [Schimel et al. 2001]. An important mechanism in this carbon uptake is the reforestation of abandoned agricultural land in the northern mid-latitudes [Schimel et al. 2001]. Barford et al. [2001] estimated a net uptake of 2.0 (+/- 0.4) megagrams of carbon per hectare per year in a 60-80 year old deciduous northern mid-latitude forest during the 1990s. Accurately quantifying the uptake of atmospheric CO2 by the terrestrial biosphere is important for estimating the global carbon budget and understanding the carbon cycle processes that determine it.

Atmosphere-Forest Carbon Exchange

The processes of photosynthesis and respiration drive the exchange of carbon between the atmosphere and forest ecosystems. During photosynthesis, CO2 is removed from the atmosphere and sequestered as organic carbon in the woody plant material of growing forests. Respiration returns a portion of the photosynthetically fixed carbon to the atmosphere as CO2 from live plant respiration and microbial decomposition of organic material. The net forest-atmosphere exchange of carbon is determined by the balance between photosynthetic uptake (sinks) and release by respiration (sources). If the photosynthetic activity of a forest ecosystem is greater than its respiration, there will be a net uptake of atmospheric CO2. Thus, measuring the rates of photosynthesis and respiration is necessary for measuring carbon exchange.

1.2 Soil Respiration

There are two main components of forest respiration: soil respiration and the above-ground respiration of living plants from stems and leaves. Soil respiration is a major component of forest ecosystem carbon release. Soil respiration in northern mid-latitude forests is the sum of respiration by living plant roots (autotrophic) and the decomposition of plant and root litter (heterotrophic). Bowden et al. [1993] estimated roughly equal contributions to total soil respiration from live root respiration, decomposition of below-ground litter, and decomposition of above-ground litter. Below-ground respiration (the sum of live root respiration and decomposition of below-ground litter) from soils can account for 80 percent or more of total forest respiration [Davidson et al. 1998, Wofsy et al. 1993]. The accurate measurement of soil respiration is therefore essential to accurately quantifying atmosphere-forest carbon exchange.

Physical Processes of Soil Respiration

The biological process of soil respiration releases CO2, but physical processes determine the transport of carbon dioxide from the soil column to the atmosphere. The movement of CO2 across the soil-atmosphere boundary is primarily driven by molecular diffusion. The concentration of CO2 is much greater in soil than the atmosphere, which results in a soil-atmosphere concentration gradient that drives efflux [Nakayama 1990].

However, air pressure gradients within or above the soil column can affect normal soil trace gas diffusion. These gradients can lead to movement of soil gases at rates greater than diffusive flux (mass flow). Turbulent air movement at the soil surface is one mechanism by which air pressure gradients are formed [Kimball and Lemon 1972].

Kimball and Lemon [1971] studied the evaporation of heptane through a series of media during surface air turbulence and pressure fluctuation. Under fan-induced surface turbulence, soil-atmosphere gas exchange increased in materials ranging in coarseness from straw to silt loam. However, they found that under natural conditions, including wind speeds 0.1-0.5 meters per second (ms-1), there was no significant turbulent effect on soil-atmosphere gas exchange in textures ranging from coarse sand to silt loam. Kimball and Lemon [1971] concluded that diffusion is the primary process for soil aeration in natural conditions, but suggested that air turbulence can contribute to gas exchange in coarse-textured mulches and in the very shallow depths of dry soil, especially in regions of high wind speeds or turbulence. The magnitude of this pressure-induced mass flow increases with soil permeability but decreases with soil depth [Kimball and Lemon 1972].

This turbulent effect may be evident in periods of unusually high soil respiration in poorly drained or bog soils corresponding to high wind speeds [Goulden et al. 1996a]. A similar effect on upland forest soils was not noted [Goulden et al. 1996a, Baldocchi et al. 1986]. Thus, molecular diffusion is considered the primary process controlling the efflux of CO2 due to soil respiration in upland forest ecosystems under natural conditions.

1.3 Soil Respiration Measurement Methods

There are two widely used methods for the measurement of soil respiration: micrometeorological techniques and point-based chamber systems [Lund et al. 1999].

Micrometeorological Methods

The eddy correlation method is a common micrometeorological technique. Forest soil respiration can be measured by the eddy correlation method above the canopy at night or by below-canopy towers during the daytime [Rustad et al. 2000]. Theoretically, the eddy correlation method is based upon a conservation equation: the change in concentration of CO2 is equal to the sum of 1) mean horizontal and vertical advection, 2) mean horizontal and vertical divergence or convergence of turbulent flux, 3) molecular diffusion, and 4) any atmospheric sources or sinks [Baldocchi et al. 1988].

This equation is simplified by a set of assumptions. First, a horizontally uniform and level surface at a micrometeorological site should minimize advective transport, which is driven by the mean motion of the atmosphere [Baldocchi et al. 1988]. Second, atmospheric molecular diffusion, which is driven by random thermal motion of molecules, is negligible compared to atmospheric turbulence in most conditions [Baldocchi et al. 1988]. Finally, atmospheric sources and sinks should not be present at a micrometeorological measurement site [Baldocchi et al. 1988]. With these simplifications, the conservation equation can be reduced and the CO2 flux can be determined as the mean covariance between turbulent fluctuations (the difference between mean and instantaneous values) in vertical wind velocity and the mixing ratio of CO2 [Baldocchi et al. 1986].

The advantages of micrometeorological methods include minimal disturbance to the environment under measurement, relatively continuous long-term measurements, and area-based, integrated fluxes [Baldocchi et al. 1988]. However, there are methodological limitations and high costs associated with this method. For example, accurate measurements of the CO2 flux can be difficult during changes in the internal boundary layer, which are common at dusk and dawn, or in cases of extreme surface topography or vegetative roughness [Baldocchi et al. 1988]. Yet, eddy correlation has proven to be a useful method for determining carbon exchange between terrestrial biosphere ecosystems and the atmosphere [Baldocchi et al. 1986, Wofsy et al. 1993, Hollinger et al. 1995].

Chamber Systems

Chamber systems are divided into three basic categories: static, open dynamic, and closed dynamic. Static chambers measure soil respiration by absorption of CO2 with an alkali solution or soda lime in a closed chamber [Lund et al. 1999]. Nay et al. [1994] found that static chamber methods overestimate low fluxes and underestimate high fluxes of CO2 by as much as 50%. Static chamber methods are generally considered less desirable than open and closed dynamic measurements [Lund et al. 1999, Fang and Moncrieff 1996].

Open dynamic chambers continuously pass ambient air through the chamber until steady state is reached. The CO2 flux is then calculated as the difference in CO2 concentration for air entering and leaving the chamber [Lund et al. 1999]. Open chamber methods have been used in numerous studies of soil carbon exchange [Lund et al. 1999, Goulden et al. 1996b] and are preferred for extended or continuous chamber measurements.

Closed dynamic chamber methods are non-steady state. CO2 concentrations increase in the chamber headspace, which continuously affects the concentration gradient driving soil-atmosphere diffusion [Hutchinson et al. 2000]. Chamber headspace air is circulated in a closed loop through an external sensor, usually an infrared gas analyzer [Lund et al. 1999]. When deployed over short measurement periods (on the order of minutes), the closed dynamic chamber method can estimate soil respiration. Closed dynamic methods have been used extensively [Norman et al. 1992, Davidson et al. 1998, Davidson and Trumbore 1995].

Chamber systems are usually low-cost and easy to use[Hutchinson and Livingston 1993]. However, there are numerous potential errors associated with chamber systems because they disturb the processes controlling soil respiration and trace gas efflux in the local environment under study, leading to measurement error [Baldocchi et al. 1988]. Additionally, there are potential biases due to the spatial variability of fluxes [Folorunso and Rolston 1984]. This spatial variability is important when a small number of chamber measurements are used to estimate carbon budgets for large regions [Denmead and Rapauch 1993].

1.4 Systematic Measurement Error

The magnitude of potential error in soil respiration measurements is large and hampers attempts to measure forest carbon budgets and carbon exchange by terrestrial ecosystems. Measurement error is dependent upon the difference between measured and true flux.

Hutchinson and Livingston [1993] note that measured flux () is equal to the sum of true flux (), error due to systematic bias (), and error due to random variability ():

 =  +  + 

Measurement error due to random variability () can be estimated by sampling principles and statistical measures [Hutchinson and Livingston 1993]. However, systematic error () cannot be determined without a known flux. Thus, identification of systematic error in soil respiration measurements is important for accurate measurement of true flux ().

Systematic error can be either uniform or selective in nature. Goulden et al. [1996b] define “uniform” systematic error as systematic error that is constant and independent of measurement conditions and “selective” systematic error as error that results when the accuracy of a measurement varies as a function of the physical environment. Both uniform and selective systematic errors are present in chamber methodologies.

One source of systematic error is the biological effect of chamber presence. For example, Schwartzkopf [1978] concluded that the velocity of air flow through an open chamber affects the respiration rate, possibly through increased oxygen supply to soil biota. Additionally, biological soil respiration is related to soil temperature [Davidson and Savage 2001], which can lead to chamber temperature effects. However, the biological effects of chamber measurement are considered minimal [Hutchinson and Livingston 1993]. In contrast, systematic physical error is potentially important in chamber systems.

Physical-Based Systematic Error

Chamber systems should be designed to model natural conditions that would prevail in the absence of the chamber [Hutchinson et al. 2000]. Yet, systematic error resulting from the alteration of the physical environment under chamber measurement can be significant. Research has suggested that two potential sources of physical systematic error exist. These potential error sources are the suppressed transmission of turbulent pressure fluctuations to the soil surface and chamber pressurizations, or pressure gradients between the internal chamber and external environment, including the soil [Hutchinson and Livingston 1993, Lund et al. 1999]. Generally, a pressure equilibration tube or vent is considered sufficient to transmit turbulent fluctuations to the soil surface under closed chamber measurement [Hutchinson and Livingston 1993].

Pressure differences between internal and external environments of chambers pose serious problems for accurate measurement. Chamber pressurizations have been shown to lead to mass flow of trace gases in the soil that would not arise under natural conditions [Kanemasu et al. 1974]. This occurs because soil gases will follow an air pressure gradient from areas of higher air pressure to areas of lower air pressure. A negative pressurization will create an air pressure deficit in the chamber headspace that will drive mass flow of soil gases, including CO2, from the soil into the chamber headspace. Alternatively, a positive pressurization will create a pressure gradient from the chamber headspace into the soil, thereby suppressing diffusion and potentially flushing soil gases from the chamber.

The force of a pressure gradient on soil gases will overwhelm the force of any concentration gradient driving molecular diffusion. Because true soil-atmosphere CO2 flux is driven by molecular diffusion, a measurement system that creates an air pressure gradient in the soil column micro-environment will lead to non-diffusive flux and, thus, systematic error.

Chamber pressurizations have been noted in open dynamic systems. Kanemasu et al. [1974] found significant flux measurement errors resulting from chamber pressurizations. Measured pressure gradients on the order of a few Pascal between the internal chamber and external atmosphere environment developed when air was blown or drawn through an open chamber. Positive chamber pressurizations occurred when air was blown into the chamber and negative pressurizations resulted from drawing air through the chamber. Negative pressurizations resulted in fluxes an order of magnitude higher than fluxes observed under positive chamber pressurizations, indicating that negative pressurizations can create mass flow of carbon dioxide by suction through the soil medium [Kanemasu et al. 1974]. Increased fluxes during negative chamber pressurizations in open chamber systemswere also noted by both Lund et al. [1999] and Fang and Moncrieff. [1996]. Pressure differences on the order of 0.5 Pascal can cause significant variation in measured efflux; Lund et al. [1999] found that positive pressurizations reduced measured flux by up to 70% in open chamber systems.

Positive chamber pressurizations have been noted in closed dynamic systems. These pressurizations can occur during chamber deployment, which may lead to underestimation of measured flux by suppressing flux or disrupting the soil-atmosphere concentration gradient [Hutchinson et al. 2000]. Additionally, Lund et al. [1999] and Denmead [1979] noted that negative pressures could result in closed dynamic systems due to problems in the circulation of headspace air through a closed loop. However, systematic error due to negative chamber pressurizations in normally operating closed dynamic systems has not been previously addressed. The purpose of this research paper is to address the possibility of systematic error in closed dynamic chamber systems.

1.5 Eddy Correlation and Chamber Measurement Disparities

Rustad et al. [2000] notes that a preferred method of soil respiration measurement has not been established. This methodological debate has centered around significant differences in measured soil-atmosphere CO2 efflux between eddy correlation methods and chamber system methods. Soil respiration measurements conducted using closed dynamic chamber methods have estimated fluxes 50% greater than those measured by eddy correlation [Davidson et al. 1998]. Goulden et al. [1996a] measured 50% greater soil respiration fluxes with both open and closed dynamic chamber systems compared to eddy correlation methods during windy nights. These results suggest that chamber systems may overestimate soil-atmosphere CO2 efflux [Goulden et al. 1996a]. However, Davidson et al. [1998] concluded that “no plausible” explanation for systematic overestimation of CO2 efflux by closed dynamic chamber methods is available, with the implication that eddy correlation techniques may underestimate ecosystem respiration at night.

1.6 Potential Systematic Error in Closed Dynamic Chamber Systems

Both Norman et al. [1992] and Matthias et al. [1980] noted that wind movement around closed dynamic chambers can affect measured fluxes, recommending that chambers not be used in bare fields or other areas unsheltered from the wind. Minor vegetation, such as the forest understory or agricultural row-crops, is considered sufficient to provide windbreak protection. However, the possibility that systematic error in the closed dynamic chamber methodology is present due to wind movement, even in areas of vegetation cover, has not been considered.

As noted previously, closed dynamic chambers are equipped with a pressure equilibration tube or vent to transmit atmospheric turbulent fluctuations to the soil surface under study. Wind movement across the opening of the pressure vent tube could affect internal chamber pressure. We suggest that horizontal air flow across the pressure vent tube of closed dynamic chambers will lead to negative chamber pressurizations.