A perspective on carbon dioxide flux measurements using an open-path infrared gas analyzer in cold environments

Open-path infrared gas analyzers, such as the LI7500 (LICOR Inc, Lincoln, NE), offer tremendous promise for measuring fluxes of water and carbon dioxide. The advantage is fast frequency response, lower power requirements and simplified system design because a pump is not needed to flow air through the sensor. Hence, they have become a common instrument for measurements in remote locations where power requirements are a major consideration. Over the past few years, we have been puzzled by apparent downward fluxes of carbon dioxide during winter at our sites in northern forests. The observations are inconsistent with our knowledge of biological activity in frozen landscapes, and are different from measurements made with a closed path sensor, where small net respiration is observed. Several other researchers have also observed this at their sites, although there does not appear to be a problem at warmer sites, and summer-time comparisons between closed- and open-path sensors are usually satisfactory. Recently, scientists at LICOR have illustrated that the sensor head is temperature controlled and that the heating/cooling power increases with temperature deviations from an ideal condition of about 30C. We believe that this heating is causing air density fluctuations within the sensor head path during cold conditions, and these density fluctuations are not being accounted in the Webb/Pearman/Leuning (WPL) adjustment for density to determine the correct quantity in the eddy flux calculation. Although attempts are being made to resolve heat transfer calculations, it is possible that a full adjustment will be elusive. Here, we present a simplified bracketing of the problem to be used in quality control. We do this through a calculation of the local sensible heat flux density created by the heated sensor, which increases with decreasing temperature. We then apply the correction as an added heat flux term to the WPL adjustment to show the magnitude of potential error. This is less satisfactory than a full heat transfer correction, but allows for data quality control until a rigorous solution becomes available. We illustrate the effect using data through seasonal cycles from forested sites in central Saskatchewan, Canada, which clearly shows that the data are questionable through much of the winter.