Predicting the partial pressure of carbon dioxide in boreal lakes

Abstract

Sixteen boreal lakes in northern Alberta were sampled for a suite of water chemistry parameters, including dissolved carbon dioxide (CO2), using a headspace gas analysis technique. The lakes encompassed a wide range of pH and alkalinity but had very high dissolved organic carbon (DOC) levels (11-36 mg L-1) and were supersaturated in CO2 with respect to the atmosphere. While the partial pressure of carbon dioxide (pCO2) is regularly estimated from pH and dissolved inorganic carbon (DIC), pH was related to pCO2 at only 13 of 16 lakes and overall pH in combination with DIC was a poor predictor of pCO2. Similarly, despite very high DOC levels, pCO2 was unrelated to the DOC concentration of the lakes. Stepwise multiple linear regressions improved the prediction capability for the entire data set, when compared to simple regressions. Both physicochemical (alkalinity, temperature) and landscape descriptors (lake area, peatland relative area) were important predictors of pCO2.The best regression model included lake area, peatland relative area, and water temperature and was better able to predict pCO2 than relationships based on DOC, and pH and alkalinity, but lakes with high pCO2 (> 1000 μatm) remain under-predicted and are likely subject to additional factors controlling pCO2 that were not considered in this analysis. © 2009 Canadian Water Resources Association.