Evaluating conditional sampling strategies for trace-gas flux measurements in urban environments

In contrast to forests and agricultural surfaces, micrometeorological measurements of turbulent trace-gas exchange between urban surfaces and the atmosphere are rare. There are a small number of experiments limited to the study of turbulent CO2-exchange in urban environments. However, no attempts have been made to directly monitor the turbulent exchange processes of the large variety of other relevant trace-gases or aerosols in urban environments using eddy-covariance.

Most trace-gases are not measurable in-situ with the high frequency needed for eddy covariance. Therefore, a variety of conditional sampling techniques like Relaxed Eddy Accumulation (REA) or Disjunct Eddy Covariance (DEC) have been developed in order to operate slower gas-analyzers. Unfortunately, they introduce new uncertainties by their parametrization or by lowering statistical significance. More sophisticated and direct approaches like Disjunct Eddy Accumulation (DEA) are promising but are still affected by technical limitations.

In urban environments, further methodical considerations might inhibit the experimental determination of trace-gas fluxes by these conditional sampling strategies. The non uniformity of sources, the enormous roughness of urban surfaces, and the resulting turbulent exchange patterns question any parameterized approach. For example, little is known on the factor beta of the REA in urban turbulence. But also direct approaches like the DEC are critical, since over rough urban surfaces, fluxes are typically associated with large coherent structures, which occupy only small time fractions.

With the help of existing data-sets of high frequency CO2 and turbulence measurements from four cities worldwide (Berlin, Germany; Basel, Switzerland; Baltimore, USA and Tokyo, Japan) different approaches of conditional sampling are simulated. Their suitability for trace-gas flux measurements in urban areas is evaluated. Requirements for technical specifications and their statistical significance are defined.