Abstract |
The signals from sensors used to measure atmospheric scalar fluctuations ordinarily produce characteristic spectra and cospectra with vertical wind velocity that are strongly influenced by vertical transport processes. Unwanted signal noise generated by some of the fast response chemical sensors, however, distort the spectra and cospectra and increase run-to-run variability of eddy-correlation flux estimates. Instead of data collection periods of only 30 min needed to provide a precision of approximately 10-20% in each flux estimate obtained at heights of 5-10 m during ideal daytime conditions, an ensemble derived from several such runs is necessary with certain sensors in order to achieve comparable precision. This has been the practical experiences obtained in numerous experiments above various natural surfaces, and is verified to a limited extent by both empirical and theoretical investigations. |