Measuring Dissolved Carbon Dioxide in the Field
Carbon dioxide (CO2) is an important, well-known greenhouse gas that has been studied extensively. As methodologies and sampling technologies improve, researchers are able to develop a fuller understanding of carbon dioxide dynamics and their response to climate change, allowing for more robust climate model predictions. A large piece missing from these climate models are aquatic CO2 dynamics. Spatial and temporal carbon concentration and flux dynamics are currently lacking in the broader study of carbon cycling, which can be partly attributed to the inability to collect continuous measurements in the field, over extended periods of time (Hari et al. 2008; Johnson et al. 2010; Butman and Raymond 2011; Fiedler et al. 2012; Striegl et al. 2012; Huang et al. 2015; Leith et al. 2015).
Classical methods of measuring in situ CO2 concentrations in water have proven to be labour intensive, especially when covering large areas, and typically consist of either direct point samples (eg: light and dark bottle method (Gaarder and Gran 1927), headspace equilibration (Hope et al. 1995)) or indirect estimates from field-based pH and alkalinity measurements (Hari et al. 2008; Johnson et al. 2010). Although they may be cumbersome, these methods have been important in establishing baselines for large areas. For example, researchers have used in situ direct measurements of pH and alkalinity to calculate efflux of all streams and rivers in the contiguous United States (Butman and Raymond 2011), as well as direct concentration measurements to calculate annual gas emissions from the Yukon River Basin (Striegl et al. 2012). Despite the utility of these studies, indirect measurements used to calculate dissolved CO2concentrations are based on assumptions about chemical and temporal dynamics, potentially limiting their reliability. Crucial, high-variability portions of the timeseries are likely to be missed by these labour-intensive, low-temporal resolution methods (Hari et al. 2008).
More recently, non-dispersive infrared (NDIR) sensors have been used for a wide range of applications, including aquatic carbon cycling research. These sensors collect in situ, continuous data autonomously, and have allowed for a higher spatio-temporal resolution of CO2 dynamics with considerably less labour. Researchers have found that NDIR sensing technology is able to monitor diurnal CO2 patterns (Hari et al. 2008; Johnson et al. 2010) with accuracy comparable to that of conventional methods, and with the added bonus of being manageably sized, making deployment easier (Fiedler et al. 2013). Even in situations where the sensors experienced large changes in temperature and pressure over short (1.5 h) time intervals, sensor drift was found by Fiedler et al. (2013) to be negligible. Studies in boreal aquatic systems, where temporal changes are highly variable, also highlight the importance of measuring CO2 concentrations directly, and at high frequencies in order to capture the variability in detail (Leith et al. 2015).
While classical methods have provided useful insights, they are limited by chemical and temporal dynamic assumptions that are made in order to understand CO2 dynamics. Newer methods, like NDIR sensors, are beginning to fill some of these gaps due to their ability to collect direct, continuous measurements in the field. Their manageable size and shape allows for a wider range of study sites to be selected, making them easier to deploy in remote areas, while still maintaining a high-level of accuracy.
Supplementary Citations (not on the blog):
Gaarder, T., and H. H. Gran (1927), Investigations of the production of plankton in the Oslo Fjord, Rapp. Proc. Verb. Cons. Int. Explor. Mer. 42, 1-48.
Hope, D., J. J. C. Dawson, M. S. Cresser, and M.F. Billett (1995), A method for measuring free CO₂ in upland streamwater using headspace analysis, Journal of Hydrology 166, 1-14.