Regression-based characterization of the marine carbonate system across shelf and nearshore waters of Queen Charlotte Sound

Marine carbonate system measurements are essential for understanding ocean acidification and CaCO 3 saturation states, and their response to oceanographic and anthropogenic processes. Acquiring such measurements in remote coastal areas is limited by challenges in the development and deployment of autonomous sensors for these parameters, and by the complexity and costs of directly measuring them. We address this challenge by extending an established method of estimating carbonate system parameters through proxy variables to the remote waters of Queen Charlotte Sound, British Columbia. Paired regressions are developed from bottle samples and common hydrographic measurements for both offshore and coastal waters that intersect in this region but retain distinct relationships between total alkalinity and salinity. Application of these regression models to regional autonomous glider data demonstrates that established features of this shelf system can be characterized from a carbonate system perspective along a contiguous transect from land to the shelf break, despite greater regression uncertainties in coastal water. These results also provide new insight into the dynamics of aragonite saturation on the shelf in relation to regional winds. This approach extends a tool to gain knowledge of the marine carbonate system to a coastal region of the northeast Pacific Ocean where hydrographic data used as proxies are more commonly available. • Carbonate system parameters were predicted by linear regressions in coastal waters. • Oxygen concentrations were key predictive variables in most linear regressions. • Regression uncertainty was always greater in coastal waters than in offshore waters. • The aragonite saturation horizon initially shoaled at the onset of winter winds. • Persistent downwelling favourable winds subsequently deepened the saturation horizon.