Within TRICUSO’s Work Package 3 (Observing system design), an established methodology has been modified to characterise surface ocean pCO2 variability using observations from multiple Research Infrastructures. The approach disentangles, for the first time, the anthropogenic and circulation-driven contributions to observed pCO2 changes and has been successfully validated at reference ocean time-series stations and extended to the Southern Ocean.
This milestone achievement was led by partner National Oceanography Centre (NOC) in support of TRICUSO’s objective of defining a roadmap for delivering the optimal surface ocean pCO2 observing system.
The challenge and motivation: assessing pCO₂ drivers
TRICUSO brings together three major Research Infrastructures, combining ship-based repeat hydrography (GO-SHIP), autonomous profiling floats (Argo floats), and surface pCO2 observations to build a more complete and more coherent picture of Southern Ocean carbon uptake. A central challenge in leveraging these diverse datasets together is that each platform samples different temporal and spatial scales, making it difficult to attribute observed changes in the partial pressure of CO2 (pCO2) to their true physical and biogeochemical drivers.
The pCO2 at the ocean surface governs air-sea CO2 exchange, a flux central to the global carbon budget and to the WMO Global Greenhouse Gas Watch that TRICUSO supports. However, observed pCO2 trends reflect a mixture of long-term anthropogenic carbon accumulation, seasonal biological activity, and the continuous reorganisation via ocean circulation and physical processes. Without a framework to separate these contributions, integrating and interpreting observations across RIs remains unclear. Indeed, existing approaches, such as the classical thermal / non-thermal decomposition proposed by Takahashi et al. (1993), do not adequately isolate the anthropogenic signal from physical variability, leaving a critical gap.
The methodological framework
The method employed in this work relies on the temperature decomposition approach developed by Turner (2021), which exploits a fundamental feature of ocean physics: the preindustrial distributions of temperature and dissolved inorganic carbon (DIC) are broadly anticorrelated because CO2 solubility decreases with increasing temperature. As atmospheric CO2 rises, both excess heat and anthropogenic carbon accumulate in a correlated manner. However, changes in ocean circulation also have to be considered as they redistribute the preindustrial temperature and carbon fields without adding to them globally. Hence, the applied methodology relies on the following statement: any observed change in ocean temperature, and by extension DIC, can be decomposed into two components:
-
Excess component: driven by surface heat gain from increased radiative forcing, this parameter increases monotonically everywhere and represents the direct imprint of anthropogenic warming on the ocean.
-
Redistributed component: driven by changes in ocean circulation, which reorganises the existing heat and carbon without net addition at the global scale, this component integrates to zero globally and can drive both warming and cooling anomalies locally.
In practice, temperature and DIC profiles from each observing platform are first gridded vertically in order to handle the irregular sampling characteristic of repeat hydrographic data. At each depth the changes in DIC and temperature can be decomposed into an excess and a redistributed component via a pair of simultaneous equations. This decomposition relies on high-quality collocated measurements of DIC and temperature, such as those made as part of the GO-SHIP repeat hydrographic programme. Based on the existing method described above, in TRICUSO, we have developed an additional step: by using the carbonate system calculator CO2SYS, propagation of the decomposed temperature and DIC terms into corresponding “excess pCO2” and “redistributed pCO2” has been established, providing a direct, physically grounded link between interior ocean changes and the surface CO2 signal accessible to autonomous sensors, ships, and satellites.
Validation of the updated methodology using reference ocean time-series
The modified methodology was tested at two of the most thoroughly observed open-ocean sites, where records spanning more than 30 years provide the ideal testbed for a rigorous evaluation:
-
BATS (Bermuda Atlantic Time-series Study, North Atlantic),
-
ALOHA (Hawaii Ocean Time-series, North Pacific).
Operational since 1988, these two sites have been sampled monthly throughout the water column using discrete water sampling for chemical parameters (DIC, total alkalinity, and nutrients) and biological variables (e.g., biomass, dissolved organic carbon, phosphorus, and nitrogen).
At both sites, the decomposition yields physically coherent results consistent with our understanding of these basins. The excess components of temperature and DIC increase monotonically through time, reflecting the progressive penetration of anthropogenic heat and carbon into the subsurface ocean. The redistributed components reveal the signature of physical reorganisation. At BATS, for instance, the application of the decomposition method has revealed a strong temperature redistribution signal and alternating bands of warming and cooling extending deeper, highlighting the impact of wintertime convection and restratification. Hence, the outputs indicate that, over this time period, temperature changes at BATS are driven by redistribution, whereas DIC changes are due to excess changes representing long-term warming and anthropogenic carbon inputs (Figure 2).
The key outcome of the validation is the decomposition of surface pCO2 variability. In the upper 0–50 m layer (i.e., the layer directly exchanging CO2 with the atmosphere), the excess pCO2 term explains approximately 90% and 96% of the variance in observed pCO2 changes at ALOHA and BATS, respectively (Figure 3). In contrast, the redistributed pCO2 term accounts for only 25–30% of the variance and exhibits a weak, noisy relationship with the observed changes. When combined, the decomposed terms explain 83% and 97% of the variance in pCO2 changes at ALOHA and BATS, respectively. These values are comparable to those obtained from the classical thermal/non-thermal decomposition, which explains 100% and 92% of the variance at ALOHA and BATS, respectively. However, unlike the thermal/non-thermal framework, the new decomposition explicitly isolates the anthropogenic imprint from circulation effects. This confirms that surface pCO2 trends are primarily controlled by long-term anthropogenic carbon accumulation. Indeed, thermal pCO2 changes explain only about 50% of the variance at both sites, whereas non-thermal processes account for 90% and 76% of the variance at ALOHA and BATS, respectively.
Moreover, while the Takahashi decomposition identifies the contribution of temperature to pCO2 variability, it should not be interpreted as evidence that future changes will be predominantly thermally driven. Indeed, comparison with the Turner decomposition suggests that a substantial fraction of the observed evolution may instead reflect changes in carbon redistribution associated with physical circulation processes.
Extension to the Southern Ocean: preliminary results
The Southern Ocean is the focus of the TRICUSO project and even though it accounts for a disproportionately large fraction of global ocean carbon uptake yet remains critically undersampled. With the methodology validated at well-characterised reference stations, the approach has been extended to this region. Two complementary Southern Ocean datasets have been used: repeated cruises conducted in the Indian sector (OISO) from which time-series stations were derived, and a full meridional GO-SHIP repeat hydrographic section in the Pacific/Indian sector (SR03). Together, they illustrate the methodology’s ability to operate across the range of platforms and sampling strategies that TRICUSO integrates.
Indian sector: the OISO time-series programme
The OISO programme (Océan Indien Service d’Observation) is a French Research Infrastructure providing long-term repeat hydrographic sections in the Indian sector of the Southern Ocean. Since 1998, OISO cruises have resampled a network of stations spanning approximately 30°S to 60°S, collecting temperature, salinity, DIC, and total alkalinity profiles across contrasting oceanographic regimes, from the subtropical zone to the Antarctic Circumpolar Current. To circumvent the irregular sampling in space associated with the OISO research cruises, seven regularly sampled sites were excerpted and considered as time-series stations in this study. Additionally, along the repeated transects of the R/V Marion Dufresne, continuous surface measurements (5 m depth) of temperature, salinity, and atmospheric pressure are collected at 20-minute intervals, while pCO2 is calculated from discrete DIC and total alkalinity samples. In future work, these direct observations will be used to validate the inferred pCO2 values.
The application of the decomposition to the selected seven OISO sites produces physically coherent results. Indeed, the excess components of temperature, DIC, and pCO2 show systematic increases over time at most sites, consistent with the progressive imprint of anthropogenic forcing (Figure 4). However, strong spatial variability is also evident: non-thermal processes such as upwelling of rich-DIC deep waters, lateral advection, and biological activity dominate the short-term variability at several locations, particularly while going southwards. The decomposition disentangles these contributions in a way that classical approaches cannot, demonstrating its added value in this challenging environment.
Pacific/Indian sector: the GO-SHIP SR03 repeated section
The GO-SHIP SR03 section runs meridionally from Tasmania (~43°S) to the Antarctic continental margin (~67°S), crossing all major Southern Ocean fronts in the Pacific/Indian sector (Figure 1). Unlike the almost annually repeated OISO stations, SR03 captures the full latitudinal structure of the region in a single section, providing a complementary view of how temperature, carbon, and pCO2 have changed across the frontal system over nearly three decades. Located along the SR03 repeat section, the SOTS site (Southern Ocean Time Series; 47°S, 142°E) provides a unique link between long-term time-series observations and repeated ship-based measurements, with surface (5 m depth) pCO2 observations available since 2011, offering an additional independent observational constraint for the reconstructed pCO2.
Application of the decomposition to SR03 reveals clear latitudinal structure in the pCO2 signal (Figure 5). The excess pCO2 component is consistently the dominant term across all occupations and latitudes, confirming that the long-term anthropogenic signal is robust even across the highly dynamic frontal system. The redistributed component shows stronger variability between years, particularly at high latitudes where upwelling brings deep, carbon-rich waters close to the surface. Similarly, the non-thermal component varies noticeably while going southwards. Importantly, the excess pCO2 term tracks the total observed pCO2 increase far more closely than the classical thermal decomposition, which fails to capture the monotonic anthropogenic trend along the section.
The depth structure of the observed and derived pCO2 signals has been further examined by averaging each component over four pressure layers spanning the full water column (Figure 6). Across all depth ranges and occupation years, the non-thermal and excess components are the two dominant contributors to the observed ΔpCO2, and their magnitudes track each other closely. This observation is consistent with the interpretation that anthropogenic carbon accumulation is the primary driver of long-term pCO2 change along SR03 as well as in other places (i.e., BATS and ALOHA). Both components increase systematically from 2001 to 2018 at all depths, confirming that the anthropogenic signal has penetrated well below the surface layer. The largest absolute changes are found in the 150-500 m layer, a depth range corresponding to the ventilation of Subantarctic Mode Water and Antarctic Intermediate Water, hence highlighting the key subduction pathways for anthropogenic carbon in the Southern Ocean. In comparison, the thermal contribution is smaller and changes sign between years and depth layers, reflecting the competing effects of sea surface warming and the advection of temperature anomalies along isopycnal surfaces. The redistributed component remains minor throughout, indicating that lateral and vertical redistribution of pre-existing carbon plays a secondary role on decadal timescales relative to the direct accumulation of anthropogenic CO2.
Next steps: towards a global pCO₂ field
The establishment of this methodology represents the foundation for the next phase of WP3 and 5, which will integrate observations from across TRICUSO’s three Research Infrastructures at increasing spatial and temporal resolution:
-
GO-SHIP global repeat hydrography sections will be used to map the excess temperature and DIC fields at basin scales, characterising the full depth structure of the anthropogenic signal along key Southern Ocean transects.
-
BGC-Argo profiling floats will provide the high spatio-temporal resolution needed to interpolate both the pCO2 and excess temperature field globally between ship-based sections, leveraging the autonomous and continuous sampling capability that floats uniquely provide, while allowing comparisons between the derived products.
-
Surface observations from autonomous vehicles (saildrones, wave gliders), moorings, and/or satellite-derived sea surface temperature products will ultimately be combined with the decomposed pCO2 framework to produce constrained, high-resolution pCO2 fields across the Southern Ocean.
Together, these steps will deliver improved estimates of air-sea CO2 fluxes in the Southern Ocean, thus reducing one of the largest remaining uncertainties in the global carbon budget.
Milestone story written by Cathy Wimart-Rousseau (NOC) and Elaine McDonagh (NOC/NORCE).
