Time domain characterization of micrometeorological data based on a two sample variance

Abstract

In ecosystem research laser based gas monitors are increasingly used to measure fluxes of methane, nitrous oxide and even stable carbon isotopes. As these complex measurement devices under field conditions cannot be considered as absolutely stable, drift characterization is an issue to distinguish between atmospheric data and sensor drift. In this paper the concept of the two sample variance is utilized in analogy to previous stability investigations to characterize the stationarity of spectroscopic and micrometeorological data in the time domain. The main results of the study are practical guidelines how to use the method for eddy-covariance determination of ecosystem exchange by laser-optical instruments suffering from signal instability. As an example, the method is applied to assess the high-pass filter time constant for detrending of time series data. The method described here provides information similar to existing characterizations as the ogive analysis, the normalized error variance of the second order moment as well as information about the spectral characteristics of turbulence in the inertial subrange. The method is easy to implement and, therefore, well suited to assist as a useful tool for a routine data quality check for both, new practitioners and experts in the field.