Large Uncertainty in EstimatingpCO2From Carbonate Equilibria in Lakes

Abstract Most estimates of carbon dioxide (CO 2 ) evasion from freshwaters rely on calculating partial pressure of aquatic CO 2 ( p CO 2 ) from two out of three CO 2 ‐related parameters using carbonate equilibria. However, the p CO 2 uncertainty has not been systematically evaluated across multiple lake types and equilibria. We quantified random errors in pH, dissolved inorganic carbon, alkalinity, and temperature from the North Temperate Lakes Long‐Term Ecological Research site in four lake groups across a broad gradient of chemical composition. These errors were propagated onto p CO 2 calculated from three carbonate equilibria, and for overlapping observations, compared against uncertainties in directly measured p CO 2 . The empirical random errors in CO 2 ‐related parameters were mostly below 2% of their median values. Resulting random p CO 2 errors ranged from ±3.7% to ±31.5% of the median depending on alkalinity group and choice of input parameter pairs. Temperature uncertainty had a negligible effect on p CO 2 . When compared with direct p CO 2 measurements, all parameter combinations produced biased p CO 2 estimates with less than one third of total uncertainty explained by random p CO 2 errors, indicating that systematic uncertainty dominates over random error. Multidecadal trend of p CO 2 was difficult to reconstruct from uncertain historical observations of CO 2 ‐related parameters. Given poor precision and accuracy of p CO 2 estimates derived from virtually any combination of two CO 2 ‐related parameters, we recommend direct p CO 2 measurements where possible. To achieve consistently robust estimates of CO 2 emissions from freshwater components of terrestrial carbon balances, future efforts should focus on improving accuracy and precision of CO 2 ‐related parameters (including direct p CO 2 ) measurements and associated p CO 2 calculations.