Eddy-covariance measurements are suitable for monitoring the surface-atmosphere exchange of momentum, heat, H2O, CO2 and other trace gases. A typical setup consists of a sonic anemometer and an infrared gas analyzer (IRGA). The data streams from these two sensors are combined for the eddy-covariance flux computation. To capture all contributing scales of turbulent transport, both instruments have typical response times ≤0.1 s. However, additional infrastructure is often required for sampling air at a pre-defined location and transporting it to an IRGA for analysis. Depending on the design and dimensions of such gas sampling system (GSS), high-frequency spectral corrections can exceed 20%. The objective of this study is to produce an optimal combination of IRGA and GSS that (i) maximizes system practicability, and (ii) minimizes high-frequency spectral losses.
For this purpose an enclosed-path LI-COR LI7200 IRGA was selected. In contrast to open-path IRGAs, this type of instrument can determine H2O and CO2 dry mole fractions directly, while compared to closed-path IRGAs the required tube length is minimal. We tested combinations of this IRGA and GSSs for fulfilling our objectives in laboratory and in field settings. Focused laboratory tests were performed to determine the frequency response of individual GSS infrastructure components, such as intake tube, particle filter and screened inlet, as well as combinations thereof. This allowed identifying and improving the bottlenecks in the GSS with regard to frequency response. Comprehensive field experiments were then performed at the Niwot Ridge AmeriFlux site in July 2013, which included conditions of condensing humidity. This enabled confirming the integral performance of IRGA and GSS under field conditions, and determining the optimal setting of intake tube and particle filter heating.
From the laboratory tests we determine the frequency at which signal attenuation reaches 20% for individual parts of the GSS. For different models of screened inlets and particle filters, this frequency falls into a range of 0.8 Hz 5 Hz for H2O, 0.9 Hz >10 Hz for CO2, and 0.6 Hz 6 Hz for H2O, 4 Hz >10 Hz for CO2, respectively. A 0.7 m long stainless steel tube with 4.8 mm inner diameter is not found to limit frequency response, with 20% attenuation occurring at frequencies >10 Hz for both, H2O and CO2. From the field experiment we find that heating the intake tube and particle filter continuously with 4 W is effective, and reduces the occurrence of problematic humidity levels (RH>60%) by 50% in the IRGA cell. No further improvement of H2O frequency response is found for heating powers in excess of 4 W.
The final GSS was developed as a result of the tests. It consists of the stainless steel tube, a pleated mesh particle filter, and a proprietary screened inlet in combination with 4 W of heating and insulation. This design maximizes practicability due to minimal flow resistance and maintenance needs, and supports the routine application of adaptive correction procedures due to minimal high-frequency spectral loss.