Parameterizing air-water gas exchange in the shallow, microtidal New River estuary

© The Author(s), 2019. This article is distributed under the terms of the Creative Commons Attribution-NonCommercial‐NoDerivs License. The definitive version was published in Van Dam, B. R., Edson, J. B., & Tobias, C. Parameterizing air-water gas exchange in the shallow, microtidal New River estuary. Journal of Geophysical Research-Biogeosciences, 124(7), (2019): 2351-2363, doi: 10.1029/2018JG004908.Estuarine CO2 emissions are important components of regional and global carbon budgets, but assessments of this flux are plagued by uncertainties associated with gas transfer velocity (k) parameterization. We combined direct eddy covariance measurements of CO2 flux with waterside pCO2 determinations to generate more reliable k parameterizations for use in small estuaries. When all data were aggregated, k was described well by a linear relationship with wind speed (U10), in a manner consistent with prior open ocean and estuarine k parameterizations. However, k was significantly greater at night and under low wind speed, and nighttime k was best predicted by a parabolic, rather than linear, relationship with U10. We explored the effect of waterside thermal convection but found only a weak correlation between convective scale and k. Hence, while convective forcing may be important at times, it appears that factors besides waterside thermal convection were likely responsible for the bulk of the observed nighttime enhancement in k. Regardless of source, we show that these day‐night differences in k should be accounted for when CO2 emissions are assessed over short time scales or when pCO2 is constant and U10 varies. On the other hand, when temporal variability in pCO2 is large, it exerts greater control over CO2 fluxes than does k parameterization. In these cases, the use of a single k value or a simple linear relationship with U10 is often sufficient. This study provides important guidance for k parameterization in shallow or microtidal estuaries, especially when diel processes are considered.We thank SERDP and DCERP for funding and support. Dennis Arbige assisted with EC tower construction, and Susan Cohen provided invaluable logistical support. I also thank Marc Alperin (UNC Chapel Hill) for his thoughtful guidance and encouragement with this project. All data sets for this manuscript are available at FigShare (https://doi.org/10.6084/m9.figshare.7276877.v1). Additional funding for this project was provided by DAAD (57429828) from funds of the German Federal Ministry of Education and Research (BMBF)