An approach to minimizing artifacts caused by cross-sensitivity in the determination of air-sea C[O.sub.2] flux using the eddy-covariance technique

The air-sea C[O.sub.2] flux was measured from a research vessel in the North Yellow Sea in October 2007 using an open-path eddy-covariance technique. In 11 out of 64 samples, the normalized spectra of scalars (C[O.sub.2], water vapour, and temperature) showed similarities. However, in the remaining samples, the normalized C[O.sub.2] spectra were observed to be greater than those of water vapour and temperature at low frequencies. In this paper, the noise due to cross-sensitivity was identified through a combination of intercomparisons among the normalized spectra of three scalars and additional analyses. Upon examination, the cross-sensitivity noise appeared to be mainly present at frequencies < 0.8Hz. Our analysis also suggested that the high-frequency fluctuations of C[O.sub.2] concentration (frequency > 0.8Hz) was probably less affected by the cross-sensitivity. To circumvent the cross-sensitivity issue, the cospectrum in the high-frequency range 0.8-1.5 Hz, instead of the whole range, was used to estimate the C[O.sub.2] flux by taking the contribution of the high frequency to the C[O.sub.2] flux to be the same as the contribution to the water vapour flux. The estimated air-sea C[O.sub.2] flux in the North Yellow Sea was -0.039 ± 0.048 mg [m.sup.-2] [s.sup.1], a value comparable to the estimates using the inertial dissipation method and Edson's method (Edson et al., J Geophys Res 116:C00F10, 2011). Keywords Air-sea C[O.sub.2] flux * Cospectrum * Cross-sensitivity * Eddy covariance * Yellow Sea