| Summary: | Air-sea fluxes of CO$\sb2$ depend on the gas-transfer coefficient (K) and the air-sea difference in the partial pressure of CO$\sb2$ ($\Delta p$). If K and $\Delta p$ covary, the mean air-sea flux of CO$\sb2$ will differ from the flux computed from means of K and $\Delta p$. The difference is termed here the covariance term. Moreover, the partial pressure of CO$\sb2$ in seawater (p) is a nonlinear function of several seawater properties, such as temperature (T), salinity (S), concentration of dissolved inorganic carbon (C) and alkalinity (A). As a result of this nonlinearity, air-sea fluxes computed using mean values of p for a range of seawater conditions will differ from the fluxes calculated using p computed from corresponding means of T, S, C and A. The difference is termed here the carbonate nonlinearity term. Any study of air-sea fluxes of CO$\sb2$, whether based on observations or models, should, ideally, select time and space scales such as to minimize both these terms. In this thesis, I quantify the covariance and the nonlinearity terms at various spatial and temporal scales and explore implications of these results for studies of air-sea fluxes of CO$\sb2$. The spatial component of the nonlinearity term is examined using data collected during the Geochemical Ocean Sections Study (GEOSECS). Standard deviation of p is a good indicator of the magnitude of the nonlinearity term. Failing to consider the spatial fluctuations at the global scale may bias direct estimates of global oceanic uptake of CO$\sb2$ upward by 3.0 GtCy$\sp{-1}$ (=65% of total emissions of anthroprogenic CO$\sb2$ in 1973). Partitioning the global dataset into subsets representing high- and low-latitude waters reduces the bias to 1.4 GtCy$\sp{-1}$. To study the temporal components of the covariance and the nonlinearity terms a new ecosystem model of the Labrador Sea is developed. The model is then used to simulate the annual cycles of K, T, S, C and A. When the annual means of these properties are used to compute air-sea fluxes of CO$\sb2$, both the covariance and the nonlinearity terms are neglected. This results in the overestimation of the air-sea fluxes by 2.4 mol C m$\sp{-2}\rm y\sp{-1}$ (=300% of the estimated total annual uptake of anthropogenic CO$\sb2$ for the Labrador Sea) compared with the best estimate from the ecosystem model. The overstimulation would increase markedly in CO$\sb2$-rich environments. Partitioning the annual cycle into warm and cold seasons reduces the overestimation severalfold. The Labrador Sea model is used to rank the importance of various oceanic processes for air-sea CO$\sb2$ flux. Effects of changes in these processes on the CO$\sb2$ flux would be larger in CO$\sb2$-rich environments. Thesis (Ph.D.)--Dalhousie University (Canada), 1996.
|
|---|