Uncertainties in the assessment of the isotopic composition of surface fluxes: A direct comparison of techniques using laser-based water vapor isotope analyzers

The isotopic composition of surface fluxes is a key environmental tracer currently estimated with a variety of methods, including: Keeling mixing models, the flux-gradient technique, and eddy covariance. We present a direct inter-comparison of these three methods used to estimate the isotopic ratio of water vapor in surface fluxes (delta(ET)) over half-hour periods, with a focus on the statistical uncertainty of each method (sigma(delta ET)). We develop expressions for sigma(delta ET) as a function of instrument precision, sample size, and atmospheric conditions. Uncertainty estimators are validated with high frequency (1 Hz) data from multiple configurations of commercial off-axis integrated cavity output spectroscopy (ICOS) systems. We find measurement techniques utilizing the high frequency capabilities of ICOS system outperform those methods where a single average of the isotopic composition is obtained at each height, with improvements attributed to large sample counts and increased variation in observed concentrations. Analytically, and with supporting data, we show that over 30 minute periods the Keeling plot and flux-gradient techniques produce nearly identical delta(ET) and sigma(delta ET) values, while eddy covariance calculations always introduce more uncertainty given the same high frequency data. This additional uncertainty is proportional to the reciprocal of the correlation coefficient between vertical wind speed and water vapor mixing ratio. Finally, given the inverse relationship between delta(ET) uncertainties and the range of water vapor observed, we propose that experimental designs should attempt to maximize both sample count and the coefficient of variation in atmospheric water vapor.