Surface Ocean pCO2 Research Articles

The importance of adding unbiased Argo observations to the ocean carbon observing system

The current coverage of direct, high-quality ship-based observations of surface ocean pCO2 includes large gaps in time and space, and has been declining since 2017. These ocean observations provide the basis for the data products that reconstruct surface ocean pCO2 and estimate ocean carbon uptake. Improved data coverage is needed to advance our understanding of the ocean carbon sink and air–sea CO2 exchange. Targeted sampling from autonomous platforms, such as biogeochemical floats, combined with traditional shipboard measurements represents a promising path forward to improve surface ocean pCO2 reconstructions. However, floats provide indirect pCO2 estimates derived from pH, and thus have higher uncertainty and are biased compared to direct shipboard measurements. Here, we use a Large Ensemble Testbed (LET) of Earth System Models and the pCO2-Residual method to reconstruct surface ocean pCO2 globally to test the impact of additional float observations, both with and without measurement uncertainties. Through comparison to the ‘model truth’, the LET allows for robust evaluation of the reconstructions. With only shipboard sampling, surface ocean pCO2 is overestimated, and the 2000–2016 global ocean carbon sink is underestimated by 0.1 Pg C year−1. Additional float observations significantly reduce this underestimation, and deviate from the ‘model truth’ by as little as 0.01 Pg C year−1, even when floats have random uncertainties of ± 11 μatm. However, systematic bias in the float observations significantly degrades the accuracy of pCO2 reconstructions, leading to an even stronger underestimation of the global ocean carbon sink of up to 0.32 Pg C year−1. We conclude that adding float-based observations to the global observing system can significantly improve reconstructions of global surface ocean pCO2, but only if these data are unbiased.

The Southern Ocean carbon sink has been overestimated in the past three decades

Employing machine learning methods for mapping surface ocean pCO2 has reduced the uncertainty in estimating sea-air CO2 flux. However, a general discrepancy exists between the Southern Ocean carbon sinks derived from pCO2 products and those from biogeochemistry models. Here, by performing a boosting ensemble learning feed-forward neural networks method, we have identified an underestimation of the surface Southern Ocean pCO2 due to notably uneven density of pCO2 measurements between summer and winter, which resulted in about 16% overestimating of Southern Ocean carbon sink over the past three decades. In particular, the Southern Ocean carbon sink since 2010 was notably overestimated by approximately 29%. This overestimation can be mitigated by a winter correction in algorithms, with the average Southern Ocean carbon sink during 1992-2021 corrected to −0.87 PgC yr−1 from the original −1.01 PgC yr−1. Furthermore, the most notable underestimation of surface ocean pCO2 mainly occurred in regions south of 60°S and was hiding under ice cover. As the surface ocean pCO2 under sea ice coverage in the winter is much higher than the atmosphere, if sea ice melts completely, there could be a further reduction of about 0.14 PgC yr−1 in the Southern Ocean carbon sink.

Zinc stimulation of phytoplankton in a low carbon dioxide, coastal Antarctic environment.

Zinc (Zn) is a key micronutrient used by phytoplankton for carbon (C) acquisition, yet there have been few observations of its influence on natural oceanic phytoplankton populations. In this study, we observed Zn limitation of growth in the natural phytoplankton community of Terra Nova Bay, Antarctica, due to low (~220 μatm) pCO2 conditions, in addition to primary iron (Fe) limitation. Shipboard incubation experiments amended with Zn and Fe resulted in significantly higher chlorophyll a content and dissolved inorganic carbon drawdown compared to Fe addition alone. Zn and Fe response proteins detected in incubation and environmental biomass provided independent verification of algal co-stress for these micronutrients. These observations of Zn limitation under low pCO2 conditions demonstrate Zn can influence coastal primary productivity. Yet, as surface ocean pCO2 rises with continued anthropogenic emissions, the occurrence of Zn/C co-limitation will become rarer, impacting the biogeochemical cycling of Zn and other trace metal micronutrients.

Assessing improvements in global ocean pCO2 machine learning reconstructions with Southern Ocean autonomous sampling

Abstract. The Southern Ocean plays an important role in the exchange of carbon between the atmosphere and oceans and is a critical region for the ocean uptake of anthropogenic CO2. However, estimates of the Southern Ocean air–sea CO2 flux are highly uncertain due to limited data coverage. Increased sampling in winter and across meridional gradients in the Southern Ocean may improve machine learning (ML) reconstructions of global surface ocean pCO2. Here, we use a large ensemble test bed (LET) of Earth system models and the “pCO2-Residual” reconstruction method to assess improvements in pCO2 reconstruction fidelity that could be achieved with additional autonomous sampling in the Southern Ocean added to existing Surface Ocean CO2 Atlas (SOCAT) observations. The LET allows for a robust evaluation of the skill of pCO2 reconstructions in space and time through comparison to “model truth”. With only SOCAT sampling, Southern Ocean and global pCO2 are overestimated, and thus the ocean carbon sink is underestimated. Incorporating uncrewed surface vehicle (USV) sampling increases the spatial and seasonal coverage of observations within the Southern Ocean, leading to a decrease in the overestimation of pCO2. A modest number of additional observations in Southern Hemisphere winter and across meridional gradients in the Southern Ocean leads to an improvement in reconstruction bias and root-mean-squared error (RMSE) of as much as 86 % and 16 %, respectively, as compared to SOCAT sampling alone. Lastly, the large decadal variability of air–sea CO2 fluxes shown by SOCAT-only sampling may be partially attributable to undersampling of the Southern Ocean.

Assessment of Global Ocean Biogeochemistry Models for Ocean Carbon Sink Estimates in RECCAP2 and Recommendations for Future Studies

AbstractThe ocean is a major carbon sink and takes up 25%–30% of the anthropogenically emitted CO2. A state‐of‐the‐art method to quantify this sink are global ocean biogeochemistry models (GOBMs), but their simulated CO2 uptake differs between models and is systematically lower than estimates based on statistical methods using surface ocean pCO2 and interior ocean measurements. Here, we provide an in‐depth evaluation of ocean carbon sink estimates from 1980 to 2018 from a GOBM ensemble. As sources of inter‐model differences and ensemble‐mean biases our study identifies (a) the model setup, such as the length of the spin‐up, the starting date of the simulation, and carbon fluxes from rivers and into sediments, (b) the simulated ocean circulation, such as Atlantic Meridional Overturning Circulation and Southern Ocean mode and intermediate water formation, and (c) the simulated oceanic buffer capacity. Our analysis suggests that a late starting date and biases in the ocean circulation cause a too low anthropogenic CO2 uptake across the GOBM ensemble. Surface ocean biogeochemistry biases might also cause simulated anthropogenic fluxes to be too low, but the current setup prevents a robust assessment. For simulations of the ocean carbon sink, we recommend in the short‐term to (a) start simulations at a common date before the industrialization and the associated atmospheric CO2 increase, (b) conduct a sufficiently long spin‐up such that the GOBMs reach steady‐state, and (c) provide key metrics for circulation, biogeochemistry, and the land‐ocean interface. In the long‐term, we recommend improving the representation of these metrics in the GOBMs.

A regional pCO2 climatology of the Baltic Sea from in situ pCO2 observations and a model-based extrapolation approach

Abstract. Ocean surface pCO2 estimates are of great interest for the calculation of air–sea CO2 fluxes, oceanic uptake of anthropogenic CO2, and eventually the Global Carbon Budget. They are accessible from direct observations, which are discrete in space and time and thus always sparse, or from biogeochemical models, which only approximate reality. Here, a combined method for the extrapolation of pCO2 observations is presented that uses (1) model-based patterns of variability from an empirical orthogonal function (EOF) analysis of variability with (2) observational data to constrain EOF pattern amplitudes in (3) an ensemble approach, which locally adjusts the spatial scale of the mapping to the density of the observations. Thus, data-constrained, gap- and discontinuity-free mapped fields including local error estimates are obtained without the need for or dependence on ancillary data (e.g. satellite sea surface temperature maps). This extrapolation approach is generic in that it can be applied to any oceanic or coastal region covered by a suitable model and observations. It is used here to establish a regional pCO2 climatology of the Baltic Sea (Bittig et al., 2023: https://doi.org/10.1594/PANGAEA.961119), largely based on ICOS-DE ship of opportunity (SOOP) Finnmaid surface pCO2 observations between Lübeck-Travemünde (Germany) and Helsinki (Finland). The climatology can serve as improved input for atmosphere–ocean CO2 flux estimation in this coastal environment.

Spatiotemporal reconstruction of global ocean surface pCO2 based on optimized random forest

The partial pressure of ocean surface CO2 (pCO2) plays an important role in quantifying the carbon budget and assessing ocean acidification. For such a vast and complex ocean system as the global ocean, most current research practices tend to study the ocean into regions. In order to reveal the overall characteristics of the global ocean and avoid mutual influence between zones, a holistic research method was used to detect the correlation of twelve predictive factors, including chlorophyll concentration (Chlor_a), diffuse attenuation coefficient at 490 nm (Kd_490), density ocean mixed layer thickness (Mlotst), eastward velocity (East), northward velocity (North), salinity (Sal), temperature (Temp), dissolved iron (Fe), dissolved silicate (Si), nitrate (NO3), potential of hydrogen (pH), phosphate (PO4), at the global ocean scale. Based on measured data from the Global Surface pCO2 (LDEO) database, combined with National Aeronautics and Space Administration (NASA) Ocean Color satellite data and Copernicus Ocean reanalysis data, an improved optimized random forest (ORF) method is proposed for the overall reconstruction of global ocean surface pCO2, and compared with various machine learning methods. The results indicate that the ORF method is the most accurate in overall modeling at the global ocean scale (mean absolute error of 6.27μatm, root mean square error of 15.34μatm, R2 of 0.92). Based on independent observations from the LDEO dataset and time series observation stations, the ORF model was further validated, and the global ocean surface pCO2 distribution map of 0.25° × 0.25° for 2010 to 2019 was reconstructed, which is of significance for the global ocean carbon cycle and carbon assessment.

Calibrating Non‐Thermal Effects on Planktic Foraminiferal Mg/Ca for Application Across the Cenozoic

AbstractForaminiferal Mg/Ca has proven to be a powerful paleothermometer for reconstructing past sea‐surface temperature, which, among other applications, is a critical parameter for boron isotope reconstructions of past surface ocean pH and PCO2. However, recent laboratory culture studies indicate seawater pH and the total dissolved inorganic carbon content (DIC) may both exert a significant additional control on foraminiferal Mg/Ca, likely influencing paleotemperature records as a result of seawater chemistry evolution on geologic timescales. In addition, the seawater Mg/Ca composition (Mg/Casw) has been shown to reduce the sensitivity of foraminiferal Mg/Ca to temperature and possibly its sensitivity to the carbonate system as well. Here we present new Mg/Ca data from laboratory culture experiments with living planktic foraminifera— Globigerinoides ruber (p), Trilobatus sacculifer, and Orbulina universa— grown under a range of different pH and/or seawater DIC conditions and in low Mg/Casw to mimic the chemical composition of the Paleocene ocean. We also conducted targeted [Ca] experiments to help define Mg/Cacalcite–Mg/Casw relationships for each species and conducted new pH experiments with G. bulloides. We find that pH effects on foraminiferal Mg/Ca are reduced or absent at Mg/Casw = 1.5 mol/mol in all three species, and that T. sacculifer is generally insensitive to variable DIC and pH, making it the ideal species for Mg/Ca SST reconstructions back to 20 Ma. We apply our new T. sacculifer calibration to a Middle Miocene Mg/Ca record and provide recommendations for interpreting Mg/Ca records from extinct species.

Estimating marine carbon uptake in the northeast Pacific using a neural network approach

Abstract. The global ocean takes up nearly a quarter of anthropogenic CO2 emissions annually, but the variability in this uptake at regional scales remains poorly understood. Here we use a neural network approach to interpolate sparse observations, creating a monthly gridded seawater partial pressure of CO2 (pCO2) data product from January 1998 to December 2019, at 1/12∘ × 1/12∘ spatial resolution, in the northeast Pacific open ocean, a net sink region. The data product (ANN-NEP; NCEI Accession 0277836) was created from pCO2 observations within the 2021 version of the Surface Ocean CO2 Atlas (SOCAT) and a range of predictor variables acting as proxies for processes affecting pCO2 to create nonlinear relationships to interpolate observations at a spatial resolution 4 times greater than leading global products and with better overall performance. In moving to a higher resolution, we show that the internal division of training data is the most important parameter for reducing overfitting. Using our pCO2 product, wind speed, and atmospheric CO2, we evaluate air–sea CO2 flux variability. On sub-decadal to decadal timescales, we find that the upwelling strength of the subpolar Alaskan Gyre, driven by large-scale atmospheric forcing, acts as the primary control on air–sea CO2 flux variability (r2=0.93, p<0.01). In the northern part of our study region, divergence from atmospheric CO2 is enhanced by increased local wind stress curl, enhancing upwelling and entrainment of naturally CO2-rich subsurface waters, leading to decade-long intervals of strong winter outgassing. During recent Pacific marine heat waves from 2013 on, we find enhanced atmospheric CO2 uptake (by as much as 45 %) due to limited wintertime entrainment. Our product estimates long-term surface ocean pCO2 increase at a rate below the atmospheric trend (1.4 ± 0.1 µatm yr−1) with the slowest increase in the center of the subpolar gyre where there is strong interaction with subsurface waters. This mismatch suggests the northeast Pacific Ocean sink for atmospheric CO2 may be increasing.

Seasonal Variability of the Surface Ocean Carbon Cycle: A Synthesis

AbstractThe seasonal cycle is the dominant mode of variability in the air‐sea CO2 flux in most regions of the global ocean, yet discrepancies between different seasonality estimates are rather large. As part of the Regional Carbon Cycle Assessment and Processes Phase 2 project (RECCAP2), we synthesize surface ocean pCO2 and air‐sea CO2 flux seasonality from models and observation‐based estimates, focusing on both a present‐day climatology and decadal changes between the 1980s and 2010s. Four main findings emerge: First, global ocean biogeochemistry models (GOBMs) and observation‐based estimates (pCO2 products) of surface pCO2 seasonality disagree in amplitude and phase, primarily due to discrepancies in the seasonal variability in surface DIC. Second, the seasonal cycle in pCO2 has increased in amplitude over the last three decades in both pCO2 products and GOBMs. Third, decadal increases in pCO2 seasonal cycle amplitudes in subtropical biomes for both pCO2 products and GOBMs are driven by increasing DIC concentrations stemming from the uptake of anthropogenic CO2 (Cant). In subpolar and Southern Ocean biomes, however, the seasonality change for GOBMs is dominated by Cant invasion, whereas for pCO2 products an indeterminate combination of Cant invasion and climate change modulates the changes. Fourth, biome‐aggregated decadal changes in the amplitude of pCO2 seasonal variability are largely detectable against both mapping uncertainty (reducible) and natural variability uncertainty (irreducible), but not at the gridpoint scale over much of the northern subpolar oceans and over the Southern Ocean, underscoring the importance of sustained high‐quality seasonally resolved measurements over these regions.

Anthropogenic Attribution of the Increasing Seasonal Amplitude in Surface Ocean pCO2

AbstractA positive trend in time has been noted in the seasonal amplitude of surface ocean pCO2 over much of the oceans, which is expected to have detrimental impacts on marine ecosystems. To determine whether or not this has an anthropogenic cause, this study investigates historical climate simulations from the Detection and Attribution Model Intercomparison Project with and without anthropogenic forcing. The simulations with anthropogenic forcing show clear evidence of positive trends, whereas the simulations with constant preindustrial atmospheric CO2 and natural external forcing give only negligible trends. A statistical analysis of five zonal latitudinal bands reveals that the trends detected over 1990–2014 in an ensemble of six observational products are attributable to anthropogenic forcing in mid‐latitudes (40°N–10°N, 10°S–40°S), while no trends are detected and modeled in the tropics and the Southern Ocean. Most models fail to represent the sign of the observed climatological mean seasonal cycle difference in high latitudes.

High interannual surface pCO2 variability in the southern Canadian Arctic Archipelago's Kitikmeot Sea

Abstract. Warming of the Arctic due to climate change means the Arctic Ocean is now free from ice for longer, as sea ice melts earlier and refreezes later. Yet, it remains unclear how this extended ice-free period will impact carbon dioxide (CO2) fluxes due to scarcity of surface ocean CO2 measurements. Baseline measurements are urgently needed to understand spatial and temporal air–sea CO2 flux variability in the changing Arctic Ocean. There is also uncertainty as to whether the previous basin-wide surveys are representative of the many smaller bays and inlets that make up the Canadian Arctic Archipelago (CAA). By using a research vessel that is based in the remote Inuit community of Ikaluqtuutiak (Cambridge Bay, Nunavut), we have been able to reliably survey pCO2 shortly after ice melt and access previously unsampled bays and inlets in the nearby region. Here we present 4 years of consecutive summertime pCO2 measurements collected in the Kitikmeot Sea in the southern CAA. Overall, we found that this region is a sink for atmospheric CO2 in August (average of all calculated fluxes over the four cruises was −4.64 mmol m−2 d−1), but the magnitude of this sink varies substantially between years and locations (average calculated fluxes of +3.58, −2.96, −16.79 and −0.57 mmol m−2 d−1 during the 2016, 2017, 2018 and 2019 cruises, respectively). Surface ocean pCO2 varied by up to 156 µatm between years, highlighting the importance of repeat observations in this region, as this high interannual variability would not have been captured by sparse and infrequent measurements. We find that the surface ocean pCO2 value at the time of ice melt is extremely important in constraining the magnitude of the air–sea CO2 flux throughout the ice-free season. However, further constraining the air–sea CO2 flux in the Kitikmeot Sea will require a better understanding of how pCO2 changes outside of the summer season. Surface ocean pCO2 measurements made in small bays and inlets of the Kitikmeot Sea were ∼ 20–40 µatm lower than in the main channels. Surface ocean pCO2 measurements made close in time to ice breakup (i.e. within 2 weeks) were ∼ 50 µatm lower than measurements made > 4 weeks after breakup. As previous basin-wide surveys of the CAA have focused on the deep shipping channels and rarely measure close to the ice breakup date, we hypothesize that there may be an observational bias in previous studies, leading to an underestimate of the CO2 sink in the CAA. These high-resolution measurements constitute an important new baseline for gaining a better understanding of the role this region plays in the uptake of atmospheric CO2.

Heat and carbon uptake in the Southern Ocean: the state of the art and future priorities.

Heat and carbon uptake in the Southern Ocean: the state of the art and future priorities.

Air‐Sea Fluxes of CO2 in the Indian Ocean Between 1985 and 2018: A Synthesis Based on Observation‐Based Surface CO2, Hindcast and Atmospheric Inversion Models

AbstractThe Indian Ocean significantly influences the global carbon cycle but it is one of the undersampled regions with reference to surface ocean pCO2. As a part of the Regional Carbon Cycle Assessment and Processes‐2 (RECCAP2) project, several approaches, such as interpolated observational climatology, hindcast model, observation‐based surface CO2 (empirical models), and atmospheric inversion models have been employed for estimating net sea‐to‐air CO2 fluxes between 1985 and 2018. The seasonal, spatial and long‐term variability in sea‐to‐air fluxes of CO2 were compared with observational climatology. The mean value of CO2 in the Indian Ocean (north of 37.5°S) for the period of 1985–2018 using all models is estimated to be −0.19 ± 0.1 PgC yr−1 and it is consistent with the observational climatology (−0.07 ± 0.14 PgC yr−1). The Indian Ocean north of 18°S is found to be the mean annual source (0.04 ± 0.05 PgC yr−1) whereas a net sink (−0.23 ± 0.11 PgC yr−1) in the south of 18°S. All models captured observed spatial patterns but underestimated the net source of CO2 in the Oman/Somalia upwelling, the Equatorial Indian Ocean and the Bay of Bengal whereas CO2 sink is overestimated in the South Indian Ocean. Overall, all models captured the seasonality in pCO2 levels and CO2 fluxes but overestimated the amplitude of their variability. All models suggested the strengthening of the sink over the period between 1985 and 2018 by 0.02 PgC yr−1 decade−1. A significant increase in the collection of surface ocean pCO2 and atmospheric CO2 measurements improves the model simulations in the Indian Ocean.

Response of Surface Ocean pCO2 to Tropical Cyclones in Two Contrasting Basins of the Northern Indian Ocean

AbstractThis study examines the impact of two very severe tropical cyclonic events, Phailin (over the Bay of Bengal) and Ockhi (over the Arabian Sea), on surface ocean pCO2 and the associated changes in the upper ocean structure using a coupled biogeochemical ROMS model. The primary productivity averaged over the mixed layer is increased from 6.9 to 12.0 (7.2–45.6) mgCm−3d−1 in response to Phailin (Ockhi). A decomposition analysis reveals that the mean contribution of temperature‐driven changes in inducing pCO2 variability in response to pre‐ and post‐cyclonic conditions, in case of Phailin (Ockhi), are 5.4 (−8.8) and −7.6 (−58.8) μatm whereas dissolved inorganic carbon (DIC) driven changes are 23.6 (19.5) and 27.8 (78.0) μatm. Although salinity and total alkalinity have a relatively lesser control in inducing pCO2 variability, salinity's response to the post‐Phailin conditions is significant owing to strong salinity stratification in the Bay of Bengal. The enhancement of DIC is more in the near‐surface waters than its removal by net biological processes resulting in the dominance of cyclone‐induced upwelling and associated vertical mixing driven changes over the enhanced biology‐driven changes in controlling pCO2 variability during both cyclones. Despite comparable magnitudes of environmental forcing during both cyclones, the oceanic response to Phailin is comparatively short‐lived and subdued due to stronger stratification in the Bay of Bengal. A relatively large ratio of DIC to total alkalinity in the upper layers of Arabian Sea facilitates a higher pCO2 response to Ockhi. This makes Ockhi a greater source of CO2 to the atmosphere.

The increasing big gap of carbon sink between the western and eastern Pacific in the last three decades

The Pacific Ocean is one of the important carbon sink regions, and there is a significant west-east difference in sea-air CO2 flux. However, the influence of the long-standing greater CO2 uptakes in the western Pacific than in the east and the dynamic change of this west-east difference remain unclear. In this paper, using the gridded surface ocean pCO2 product constructed by the stepwise FFNN algorithm, we reported an increasing west-east CO2 flux difference from 0.41 PgC yr-1 in 1992 to 0.73 PgC yr-1 in 2020. This increase was mainly attributed to the strengthening western Pacific carbon sink and relatively stable eastern Pacific carbon source. During El Nino events, the west-east CO2 flux difference decreased significantly in a few years, and it then rose back rapidly when El Nino events ended. In addition, the increasing west-east difference in CO2 uptakes during the last three decades did not lead to a higher acidification speed in the western surface temperate Pacific than the east. The greater CO2 absorbed in the west was mainly transported to the deeper waters and caused a more significant carbon inventory change at 200-600 m than the eastern Pacific.

Equatorial Pacific pCO2 Interannual Variability in CMIP6 Models

AbstractThe El Niño‐Southern Oscillation (ENSO) in the equatorial Pacific is the dominant mode of global air‐sea carbon dioxide (CO2) flux interannual variability (IAV). Air‐sea CO2 fluxes are driven by the difference between atmospheric and surface ocean pCO2, with variability of the latter driving flux variability. Previous studies found that models in Coupled Model Intercomparison Project Phase 5 (CMIP5) failed to reproduce the observed ENSO‐related pattern of CO2 fluxes and had weak pCO2 IAV, which were explained by both weak upwelling IAV and weak mean vertical dissolved inorganic carbon (DIC) gradients. We assess whether the latest generation of CMIP6 models can reproduce equatorial Pacific pCO2 IAV by validating models against observations‐based data products. We decompose pCO2 IAV into thermally and non‐thermally driven anomalies to examine the balance between these competing anomalies, which explain the total pCO2 IAV. The majority of CMIP6 models underestimate pCO2 IAV, while they overestimate sea surface temperature IAV. Insufficient compensation of non‐thermal pCO2 to thermal pCO2 IAV in models results in weak total pCO2 IAV. We compare the relative strengths of the vertical transport of temperature and DIC and evaluate their contributions to thermal and non‐thermal pCO2 anomalies. Model‐to‐observations‐based product comparisons reveal that modeled mean vertical DIC gradients are biased weak relative to their mean vertical temperature gradients, but upwelling acting on these gradients is insufficient to explain the relative magnitudes of thermal and non‐thermal pCO2 anomalies.

Modified future diurnal variability of the global surface ocean CO2 system.

Our understanding of how increasing atmospheric CO2 and climate change influences the marine CO2 system and in turn ecosystems has increasingly focused on perturbations to carbonate chemistry variability. This variability can affect ocean-climate feedbacks and has been shown to influence marine ecosystems. The seasonal variability of the ocean CO2 system has already changed, with enhanced seasonal variations in the surface ocean pCO2 over recent decades and further amplification projected by models over the 21st century. Mesocosm studies and CO2 vent sites indicate that diurnal variability of the CO2 system, the amplitude of which in extreme events can exceed that of mean seasonal variability, is also likely to be altered by climate change. Here, we modified a global ocean biogeochemical model to resolve physically and biologically driven diurnal variability of the ocean CO2 system. Forcing the model with 3-h atmospheric outputs derived from an Earth system model, we explore how surface ocean diurnal variability responds to historical changes and project how it changes under two contrasting 21st-century emission scenarios. Compared to preindustrial values, the global mean diurnal amplitude of pCO2 increases by 4.8μatm (+226%) in the high-emission scenario but only 1.2μatm (+55%) in the high-mitigation scenario. The probability of extreme diurnal amplitudes of pCO2 and [H+ ] is also affected, with 30- to 60-fold increases relative to the preindustrial under high 21st-century emissions. The main driver of heightened pCO2 diurnal variability is the enhanced sensitivity of pCO2 to changes in temperature as the ocean absorbs atmospheric CO2 . Our projections suggest that organisms in the future ocean will be exposed to enhanced diurnal variability in pCO2 and [H+ ], with likely increases in the associated metabolic cost that such variability imposes.

Estimation of sea surface pCO2 and air–sea CO2 flux in the East China Sea using in-situ and satellite data over the period 2000–2016

The partial pressure of carbon dioxide (pCO2) is an important variable to characterize the state of the seawater carbonate system. It is required to calculate the air-sea CO2 flux (FCO2) and for understanding ocean acidification. Surface ocean pCO2 is influenced by thermodynamic process, physical mixing, biological activity and air-sea gas exchange. Knowledge of the surface pCO2 in the East China Sea (ECS) is limited due to a lack of observations and well-informed models. This study employed in situ hydrographic (sea surface temperature and salinity) and carbonate system observations combined with satellite observations of chlorophyll a to develop a regional multiple non-linear regression (MNR) model to estimate pCO2 for three different physical-biogeochemical domains of the ECS. Comparison of the pCO2 from in situ observations with the MNR model had a root mean square error (RMSE) of 45.19 μatm and coefficient of determination (R2) of 0.87 in the transitional domain, an RMSE of 11.59 μatm and R2 of 0.92 in the shelf-dominated domain and an RMSE of 3.73 μatm and R2 of 0.97 in the open water domain. The reliability of the MNR model was also validated using an independent dataset with a good accuracy (RMSE of 9.15 μatm for domain II and RMSE of 7.71 μatm for domain III). The MNR model was applied to estimate the surface pCO2 and FCO2 for ten cruises on the period of 2013–2018, and also used to predict monthly surface pCO2 by using Changjiang Biology Finite-Volume Coastal Ocean Model (FVCOM) outputs for the period of 2000–2016 on the ECS inner shelf, and analyzed the seasonal and interannual variability of surface pCO2 and FCO2. The seasonal dynamics of sea surface pCO2 were mainly attributed to temperature changes, biological activities and water mixing process over the seasons. The time series of surface pCO2 and FCO2 on the ECS inner shelf showed an increasing trend of 3.32 ± 0.79 μatm yr−1 and 0.018 ± 0.0125 mol CO2 m−2 yr−2, and the region was a CO2 sink with an CO2 uptake of 0.12 ± 0.48 Tg C yr−1.

Explicit Physical Knowledge in Machine Learning for Ocean Carbon Flux Reconstruction: The pCO2‐Residual Method

AbstractThe ocean reduces human impacts on global climate by absorbing and sequestering CO2 from the atmosphere. To quantify global, time‐resolved air‐sea CO2 fluxes, surface ocean pCO2 is needed. A common approach for estimating full‐coverage pCO2 is to train a machine learning algorithm on sparse in situ pCO2 data and associated physical and biogeochemical observations. Though these associated variables have understood relationships to pCO2, it is often unclear how they drive pCO2 outputs. Here, we make two advances that enhance connections between physical understanding and reconstructed pCO2. First, we apply pre‐processing to the pCO2 data to remove the direct effect of temperature. This enhances the biogeochemical/physical component of pCO2 in the target variable and reduces the complexity that the machine learning must disentangle. Second, we demonstrate that the resulting algorithm has physically understandable connections between input data and the output biogeochemical/physical component of pCO2. The final pCO2 reconstruction agrees modestly better with independent data than most other approaches. Uncertainties in the reconstructed pCO2 and impacts on the estimated CO2 fluxes are quantified. Uncertainty in piston velocity drives substantial flux uncertainties in some regions, but does not increase globally integrated estimates of uncertainty in CO2 fluxes from observation‐based products. Our reconstructed CO2 fluxes show larger interannual variability than smoother neural network approaches, but a lesser trend since 2005. We estimate an air‐sea flux of −1.8 PgC/yr (anthropogenic flux of −2.3 ± 0.5 PgC/yr) for 1990–2019, agreeing with other data products and the Global Carbon Budget 2020 (−2.3 ± 0.4 PgC/yr).