Subantarctic pCO2 estimated from a biogeochemical float: comparison with moored observations reinforces the importance of spatial and temporal variability

Abstract

Understanding the size and future changes of natural ocean carbon sinks is critical for the projection of atmospheric CO2 levels. The magnitude of the Southern Ocean carbon flux has varied significantly over past decades but mechanisms behind this variability are still under debate. While high accuracy observations, e.g. from ships and moored platforms, are important to improve models they are limited through space and time. Observations from autonomous platforms with emerging biogeochemical capabilities, e.g. profiling floats, provide greater spatial and temporal coverage. However, the absolute accuracy of CO2 partial pressure (pCO2) derived from float pH sensors is not well constrained. Here we capitalize on data collected for over a year by a biogeochemical Argo float near the Southern Ocean Time Series observatory to evaluate the accuracy of pCO2 estimates from floats beyond the initial in water comparisons at deployment. A latitudinal gradient of increasing pCO2 southward and spatial variability contributed to observed discrepancies. Comparisons between float estimated pCO2 and mooring observations were therefore restricted by temperature and potential density criteria (~ 7 µatm difference) and distance (1° latitude and longitude, ~ 11 µatm difference). By utilizing high quality moored and shipboard underway pCO2 observations, and estimates from CTD casts, we therefore found that over a year, differences in pCO2 between platforms were within tolerable uncertainties. Continued validation efforts, using measurements with known and sufficient accuracy, are vital in the continued assessment of float-based pCO2 estimates, especially in a highly dynamic region such as the subantarctic zone of the Southern Ocean.

Drivers of Air‐Sea CO2 Flux in the Subantarctic Zone Revealed by Time Series Observations

“…This has also been pointed out by Wynn‐Edwards et al. (2023), who closely compared the estimates from mooring and float. The computed MLR is smoother at all points in the time series compared to observations, which is a result of the monthly and regionally averaged predictor variables used as input.…”

“…The SOCCOM estimates are higher than the MLR model by 20–40 μatm, especially outside the productive season (Figure 3a and Figure S2 in Supporting Information S1). Several previous studies have pointed out that they did not capture the strong winter outgassing estimated by floats (discussed further in Section 4.1; Sutton et al., 2021; Wynn‐Edwards et al., 2023). It is likely that the MLR model misses some short‐lived and weak outgassing periods in winter, but the impact of these on the net annual fluxes is small.…”

Observed amplification of the seasonal CO2 cycle at the Southern Ocean Time Series

“…Long-time series are important for the validation of float based observations (e.g. Fay et al, 2018;Wynn-Edwards et al, 2023) and for understanding air-sea interactions on decadal timescales in the Southern Ocean and elsewhere (e.g., Sutton et al, 2017;Brandon et al, 2022;Leseurre et al, 2022), as well as to improve model representation of these processes for reliable forecasting of the evolving ocean and global carbon cycle. Distinguishing anthropogenic signals from natural interannual variability requires observations sustained over many years.…”