Experimental validation of the Webb correction for CO2 flux with an open-path gas analyzer

The evaluation of the momentum, sensible heat, latent heat, and trace gas (e.g., CO2) fluxes between various surfaces and the atmosphere is one of the most basic process studies. Turbulent flux by the eddy covariance technique using open-path instruments is the only direct measurement over surfaces. CO2 and water vapor are the only trace gases that can measure turbulent flux using the open-path gas analyzer. The open-path CO2/H2O gas analyzer does not measure nondimensional CO2 and water vapor concentrations as mixing ratios but it measures CO2 and water vapor densities. For this reason, the trace gas flux using this analyzer needs to correct for the mean vertical flow due to air density fluctuation. Webb et al. (1980) suggested that the flux due to the mean vertical flow cannot be neglected for trace gases such as water vapor and CO2.

To evaluate the magnitude of the influence by the mean vertical flow, Webb et al. (1980) assumed that the vertical flux of dry air should be zero. Practically, sensible and latent heat fluxes evaluated by the eddy covariance technique are used to calculate water vapor and CO2 fluxes by the mean vertical flow (Webb or WPL correction).

Ohtaki et al. (1989) suggested that the Webb CO2 flux due to air density fluctuation is larger than the raw CO2 flux and changes the transport direction of the total CO2 flux over the coastal sea surface. Kondo and Tsukamoto (2007) also confirmed these results over the open ocean surface using ship- motion corrected eddy-covariance method.

Recently, Liu (2005) presented an alternative equation to correct effects of density fluctuations due to sensible and latent heat fluxes that does not require the original Webb correction assumption that dry air is conserved and the use of mean vertical velocity, and concluded that the original Webb correction is incomplete. However, Kowalski (2006) and Massman and Tuovinen (2006) argued that Liu's theory is mistaken. Leuning (2007) revised the correction theory presented by Webb et al. (1980) on the assumption that there is no source or sink of dry air at the ground or in the layer below the height of measurement.

The basic Webb correction theory has been discussed well, but this correction is still poorly understood from actual observed data in various conditions. Therefore, it is important to evaluate the original Webb correction by applying it to data measured over various surfaces. This paper experimentally evaluates the accuracy of the original Webb correction due to the air density fluctuation from water vapor and CO2 flux measurements by the eddy covariance technique using an open-path CO2/H2O gas analyzer.

For this purpose, we measured the turbulent flux over an asphalt surface (large parking lot), where CO2 and water vapor fluxes were considered almost zero and sensible heat flux was significant. Time series of air temperature and CO2 density shows clear negative correlations and resulted remarkable positive sensible heat flux and negative CO2 flux. Naturally, this negative CO2 flux is an apparent flux because the asphalt surface does not act as CO2 source or sink. The power spectrum of CO2 density fluctuation had similar frequency structure to those of vertical wind velocity, sound virtual temperature, and water vapor density fluctuations. The cospectrum of raw CO2 flux was negatively correlated with that of sensible heat flux. The cospectrum of latent heat flux was almost zero and poorly correlated with that of raw CO2 or sensible heat fluxes. Raw CO2 flux was apparently downward from the atmosphere to the asphalt surface during this observation period. The Webb CO2 correction term for sensible heat flux was upward and larger than the raw CO2 flux, although the Webb CO2 correction term for latent heat flux was almost zero. Total latent heat and CO2 fluxes were evaluated as significant upward fluxes.

The raw CO2 flux showed a diurnal variation, and was always negative (downward from the atmosphere to the asphalt surface) between –0.095 and –0.368 mg m^–2s^–1 (the average was –0.283 mg m^–2s^–1). The Webb CO2 correction term due to sensible heat flux also showed a diurnal variation, and was positive between 0.167 and 0.462 mg m^–2s^–1 (the sensible heat flux was between 87 and 243 Wm^–2). The Webb correction term due to latent heat flux was between –0.008 and 0.016 mg m^–2s^–1 (raw latent heat flux was between –21 and 41 Wm^–2), and was not influenced by the Webb correction. Total CO2 flux was always positive between 0.001 and 0.097 mg m^–2s^–1 (the average was 0.064 mg m^–2s^–1), and did not show a clear diurnal variation, as did raw CO2 and sensible heat fluxes. Ham and Heilman (2003) performed a similar experiment over an asphalt surface, and reported total CO2 flux by the eddy covariance technique of 0.02 mg m^–2s^–1 (=1.8 g m^–2day^–1) and CO2 flux by the chamber technique of 0.03 mg m^–2s^–1 (=2.8 g m^–2day^–1) ) over the asphalt surface. The average total CO2 flux in the present study was about two to three times larger than in their result. Although total CO2 flux measured by the eddy covariance technique as reported by Ham and Heilman (2003) included negative values, total CO2 flux in the present study retained a significantly positive value.

This result shows that Webb correction may cause an overestimated correction to CO2 flux by the eddy covariance technique using an open-path CO2/H2O gas analyzer over surfaces, especially where small CO2 flux by soil respiration or air-sea exchange process is observed.