How Does Variability in Aragonite Saturation Proxies Impact Our Estimates of the Intensity and Duration of Exposure to Aragonite Corrosive Conditions in a Coastal Upwelling System?

Determining aragonite saturation state (Ω) in seawater requires measurement of two parameters of the carbonate system: most commonly dissolved inorganic carbon (DIC) and total alkalinity (TA). The routine measurement of DIC and TA is not always possible on frequently repeated hydrographic lines or at moored-time series that collect hydrographic data at short time intervals. In such cases a proxy can be developed that relates the saturation state as derived from one time or infrequent DIC and TA measurements (Ω_meas) to more frequently measured parameters such as dissolved oxygen (DO) and temperature (Temp). These proxies are generally based on best-fit parameterizations that utilize references values of DO and Temp and adjust linear coefficients until the error between the proxy-derived saturation state (Ω_proxy) and Ω_meas is minimized. Proxies have been used to infer Ω from moored hydrographic sensors and gliders which routinely collect DO and Temp data but do not include carbonate parameter measurements. Proxies can also calculate Ω in regional oceanographic models which do not explicitly include carbonate parameters. Here we examine the variability and accuracy of Ω_proxy along a near-shore hydrographic line and a moored-time series stations at Trinidad Head, CA. The saturation state is determined using proxies from different coastal regions of the California Current Large Marine Ecosystem and from different years of sampling along the hydrographic line. We then calculate the variability and error associated with the use of different proxy coefficients, the sensitivity to reference values and the inclusion of additional variables. We demonstrate how this variability affects estimates of the intensity and duration of exposure to aragonite corrosive conditions on the near-shore shelf and in the water column.