Davidson et al. 2002

Davidson EA, Savage K, Verchot LV, Navarro R. 2002. Minimizing artifacts and biases in chamber-based measurements of soil respiration. Agricultural and Forest Meteorology 113: 21-37.

These authors review published examples of measurements of CO2 flux from soils, especially forest soils, with an emphasis on identifying major sources of error in these measurements and possible strategies to minimize these errors. As suggested by the title, they divide these sources of error into artifacts (systematic errors caused by details of the construction or operation of measurement devices), biases (consistent patterns of over- or under-estimation of fluxes), and sampling uncertainties (including spatial and temporal heterogeneities in both actual and measured gas fluxes).

Photosynthesis is the primary driver of CO2 flux, with soil respiration a close second in terrestrial ecosystems. Soil respiration may account for 60-90% of the total respiration of a temperate forest, but can be difficult to measure. The two primary methods of measurement are eddy covariance, often using tower-based systems, and chambers that are typically placed on or embedded into the soil. Direct comparisons between these methods have found large variances, suggesting that one or both methods includes inherent artifacts and biases that need to be addressed.

Chambers can be operated in one of two modes: steady-state and non-steady-state. In the steady-state mode, the flux of gas from the soil is calculated based on the difference in CO2 concentration between the inflow and outflow openings into the chamber, after the conditions inside the chamber have reached dynamic equilibrium. In the non-steady-state mode, the rate of change of gas concentration inside the chamber is measured over some time interval. Both modes have inherent problems.

Achieving a steady-state is time consumming and uncertain. In addition, the details of the rate of flow of carrier gas through the chamber, as well as the soil type, moisture level, vegetation and litter, and other variables can have a large influence on the measurements of gas concentration. There are also significant difficulties associated with determining when a true equilibrium has been attained, and the relevance of that equilibrium to “natural” conditions immediately above undisturbed soil.

The non-steady-state is easier and quicker to achieve, but again there are difficulties and uncertainties surrounding a large number of other variables. Of principle importance is the concentration gradient of the gas of interest through the soil. Higher or lower pressures inside the chamber, either of total atmosphere or the partial pressure of the gas of interest, will alter the concentration gradient and thereby change the rate of gas diffusion (i.e. the flux).

These authors seem to prefer the non-steady-state mode, at least partly because it is very rapid to get measurements compared to the wait of up to days to ensure equilibrium under the steady-state mode. They include some helpful calculations, based mainly on linear regression, to smooth the process of measuring gas flux under non-steady-state conditions. Essentially, near-continuous monitoring of gas concentrations in the non-steady-state mode allows a fit to a regression, and the disturbance effects induced by placing and initiating the chamber can be ignored.

Despite these mathematical techniques, most chambers tend to underestimate true flux rates, usually by between a trivial amount and 15%. Taller chambers and deeper penetration of collars into soil tend to reduce these sources of error, though again many variables including wind conditions and soil parameters have strong influences.

Minimizing pressure effects is another important factor in reducing errors. Fans for mixing air inside the chamber are probably unnecessary except for the largest chamber designs, though vents are probably critical in nearly all cases. External air entering a chamber through the vents will cause some error, but this is likely to be trivial compared to the avoidance of a pressure differential between the inside and outside of the chamber, especially under windy conditions.

Other minor pieces of advice include always restricting the flow rate through the chamber with a valve, to even out the effects of battery charge on the pump, and careful recalibration of chambers after any change in configuration, especially of vents. The disturbance effects associated with the placement of collars or chambers are mostly attenuated within a day, such that it may be very beneficial to emplace collars well in advance of measuring gas flux with chambers.

Spatial and temporal variation can occur at scales both much larger and much smaller than the sampling footprint of a typical chamber. These authors provide statistical advice to determine the sampling design needed to achieve a given level of confidence that one is measuring something close to the actual mean flux; for a temperate forest system in New England, as few as 6 chambers monitored for a little as 5 minutes may be sufficient to achieve 90% confidence in being within 20% of the true mean.

Diel variation is an important part of the patterns of gas flux in soils. Night-time respiration by plants in the absence of photosynthesis can alter concentration gradients to a large degree, such that true long-term fluxes may only be estimated from 24-hour monitoring. How this may apply to Arctic systems experiencing 24-hour sunlight is beyond the scope of this review.

The extensive list of relevant literature in this paper may be very useful for more information and advice for the calibration, use, and analysis of our FTIR system.