Search

The Southern Ocean forms a vital component of the earth system as a sink of CO2 and heat, taking over 40% of the annual oceanic CO2 uptake (75% of global heat uptake), slowing down the accumulation of CO2 in the atmosphere and thus the rate of climate change. However, recent studies based on the Coupled Model Intercomparison Project version 5 (CMIP5) Earth System Models (ESMs) show that CMIP5 ESMs disagree on the phasing of the seasonal cycle of the CO2 flux (FCO2) and compare poorly with available observation estimates in the Southern Ocean. Notwithstanding these differences, the seasonal cycle is a dominant mode of CO2 variability in the Southern Ocean, and hence this is an important bias. Previous studies suggest that these biases of FCO2 in ESMs might be a significant limitation to the long-term simulation of CO2 characteristics in the Southern Ocean. Consequently, this study has three primary objectives: first, to develop a process-based diagnostic method to analyze and isolate key biases and their underlaying mechanisms in the model-observations seasonal cycle of FCO2 differences for forced ocean models and ESMs. Second, to use this framework to examine sources of biases responsible for the limited skill of CMIP5 models in simulating the seasonal cycle of FCO2 with respect to observed estimates. Thirdly, to investigate how these present-day biases in the seasonality and drivers of CO2 in CMIP5 ESMs affect modelled longterm changes in the mechanisms of CO2 uptake in the Southern Ocean. In the first part of the dissertation, an objective diagnostic framework was established to analyze model-observation biases in the seasonal scale of FCO2 using the NEMO PISCES ORCA2LP model output, and Takahashi et al. (2009) observed estimates. The diagnostic framework focuses on examining the relative contributions of the competing drivers (SST and DIC) and related processes (solubility, biological and mixing) to instantaneous monthly changes in surface pCO2 (and FCO2) at the seasonal scale. In the second part of the dissertation, this approach is applied to 10 CMIP5 models in the Southern Ocean, to investigate the mechanistic basis for the seasonal cycle of FCO2 biases. It was found that FCO2 biases in CMIP5 models can be grouped into two main categories, i.e. group-SST and group-DIC. Group-SST models are characterized by an exaggeration of the seasonal rates of change of Sea Surface Temperature (SST) in autumn and spring during the cooling and warming peaks, respectively. These faster-than-observed rates of change of SST tip the control of the seasonal cycle of pCO2 and FCO2 towards SST and result in divergence between the observed and modelled seasonal cycles, particularly in the Sub-Antarctic Zone. While almost all analyzed models show these SST-driven biases, 3 out of 10 (namely NorESM1-ME, HadGEM2-ES and MPI-ESM, collectively the group-DIC models) compensate the solubility bias because of their exaggerated primary production, such that biologically-driven DIC changes become the regulators of the seasonal cycle of FCO2. It was also found that despite significant differences in the spatial characteristics of the mean annual fluxes, CMIP5 models show a zonal homogeneity in the seasonal cycle of FCO2 at the basin-scale in contrast to observed estimates. In the final third of the dissertation, using five CMIP5 ESMs from the RCP8.5 scenario, it was found that CMIP5 models present climate biases in the seasonality and drivers of FCO2 are fundamental to how models simulate long-term changes in the mechanisms of CO2 uptake in the Southern Ocean. Although all five analyzed models show an increased annual mean CO2 uptake by the end of the century, they show significant differences in the mechanisms. The present-day temperature biased models (group-SST) generally maintain the dominance of the temperature driver in the seasonal variability of FCO2 to end of the century. But show enhanced CO2 uptake due to increased anthropogenic atmospheric CO2 and decreased surface CO2 buffering capacity but they display a weak to null role of biological activity in the increased CO2 sink. On the other hand, the increased CO2 uptake at the end of the century in group-DIC models is explained increased biological driven CO2 uptake in spring, linked to increased Revelle factor and solubility driven CO2 uptake in winter. Increased Revelle factor at the end of the century enhance pCO2 changes for even smaller DIC changes.

The Southern Ocean plays an important role in mitigating the effects of anthropogenically driven climate change. The region accounts for 43% of oceanic uptake of anthropogenic carbon dioxide (CO₂). This is foreseen to change with increasing greenhouse gas emissions due to ocean chemistry and climate feedbacks that regulate the carbon cycle in the Southern Ocean. Studies have already shown that Southern Ocean CO₂ is subject to interannual variability. Measuring and understanding this change has been difficult due to sparse observational data that is biased toward summer. This leaves a crucial gap in our understanding of the Southern Ocean CO₂ seasonal cycle, which needs to be resolved to adequately monitor change and gain insight into the drivers of interannual variability. Machine learning has been successful in estimating CO₂ in may parts of the ocean by extrapolating existing data with satellite measurements of proxy variables of CO₂. However, in the Southern Ocean machine learning has proven less successful. Large differences between machine learning estimates stem from the paucity of data and complexity of the mechanisms that drive CO₂. In this study the aim is to reduce the uncertainty of estimates, advance our understanding of the interannual drivers, and optimise sampling of CO₂ in the Southern Ocean. Improving the estimates of CO₂ was achieved by investigating: the impact of increasing the gridding resolution of input data and proxy variables, and Support vector regression (SVR) and Random Forest Regression (RFR) as alternate machine learning methods. It was found that the improvement gained by increasing gridding resolution was minimal and only RFR was able to improve on existing error estimates. Yet, there was good agreement of the seasonal cycle and interannual trends between RFR, SVR and estimates from the literature. The ensemble mean of these methods was used to investigate the variability and interannual trends of CO₂ in the Southern Ocean. The interannual trends of the ensemble confirmed trends reported in the literature. A weakening of the sink in the early 2000's, followed by a strengthening a strengthening of the sink into the early 2010's. Wind was the overall driver of dominant decadal interannual trends, being more important during winter due to the increased efficacy of entrainment processes. Summer interannual variability of CO₂ was driven primarily by chlorophyll, which responded to basin scale changes in drivers by the complex interaction with underlying physics and possibly sub-mesoscale processes. Lastly CO₂ sampling platforms, namely ships, profiling floats and moorings, were tested in an idealised simulated model environment using a machine learning approach. Ships, simulated from existing cruise tracks, failed to adequately resolve CO₂ below the uncertainty threshold that is required to resolve the seasonal cycle of Southern Ocean CO₂. Eight high frequency sampling moorings narrowly outperformed 200 profiling floats, which were both able to adequately resolve the seasonal cycle. Though, a combination of ships and profiling floats achieved the smallest error.

Includes bibliographical references, A recent study by Lenton et al., 2013, compared the mean seasonal cycle of air-sea CO₂ flux in the Southern Ocean(SO) over 1990 – 2009 period using five ocean biogeochemical models(BGMs) and atmospheric and ocean inversion models with monthly mean observations for the year 2000. This was done using a set of geographic boundaries to defined sub-domains of the SO consistent with the Regional Carbon Cycle and Assessment and Processes (RECCAP) protocol. Lenton et al., 2013 found that the seasonal cycle anomaly of the five BGMs better resolved observations of the air-sea CO₂ flux seasonal cycle in the SAZ, but was generally out phase with observations in the polar zone. In this study two setups of the ocean biogeochemical model NEMO PISCES was used to investigate the characteristics of the air-sea CO₂ flux seasonal cycle in the Southern Ocean in the period 1993- 2006. The study focused on two aspects i.e. (i) the sensitivity of air-sea CO₂ flux seasonal cycle to model resolution: comparing the ORCA2-LIM-PISCES (2° x 2° cos Ø) and PERIANT05 (NEMO-PISCES) (0.5° x 0.5° cos Ø) model configurations relative to climatological mean observations for the year 2000 (Takahashi et al., 2009) , and (ii) the sensitivity of air-sea CO₂ flux seasonal cycle to zonal boundary definition: comparing the air-sea CO₂ flux seasonal cycle and annual fluxes for three different boundaries i.e. Lenton 2013 RECCAP boundaries (44°S – 58°S and south of 58°S), geographic boundaries (40°S -50°S and south of 50°S) and dynamic boundaries (Sub-Antarctic Zone and Antarctic Zone, defined using climatological frontal positions). The seasonal cycle of the air-sea CO₂ flux in ORCA2 was found to be out of phase and overestimated the CO₂ flux compared to observations in almost all the sub-regions considered. The use of dynamic boundaries was found not to improve resolving observations seasonal cycle of air-sea CO₂ flux in both ORCA2 and PERIANT05. Boundary definition was found to affect the magnitude of ORCA2 annual air-sea CO₂ fluxes surface area based, where sub-regions of larger surface area gave larger annual CO₂ uptake and vice versa. This was mainly because ORCA2 air-sea CO₂ fluxes were found to show a general CO₂ in-gassing bias and spatially uniform in most parts of the SO and hence integration over a larger surface area gave larger annual fluxes. On the contrary PERIANT05 air-sea CO₂ fluxes spatial variability was not uniform in most parts of the SO however influenced by regional processes and hence annual fluxes were found not surface area based. The poor spatial representation and seasonal cycle sensitivity of ORCA2 air-sea CO₂ fluxes was found to be primarily due to lack or weak winter CO₂ entrainment and biological CO₂ draw down during the summer season. PERIANT05 on the contrary showed the effect of winter CO₂ entrainment, however maintains lack of or weak biological CO₂ draw down in the seasonal cycle. PERIANT05 was also found to show major weakness in the spatial representation of air-sea CO₂ fluxes north of the polar front with relative to T09 observations.

Includes bibliographical references., A model study was undertaken to investigate the optimization of sampling strategies for returning low-uncertainty sea-air CO₂ flux measurements in the Southern Ocean. Replicating Lenton et al. (2006) using the ORCA2/PISCES ocean biological model shows that sampling 4 times a year, every 2⁰ in latitude and every 40⁰ in longitude reduces the uncertainty of estimating annual CO₂ flux estimates such that sampling at a higher frequency does not reduce the total uncertainty in proportion to the increase in sampling effort.

Includes bibliographical references., An observational study was undertaken to determine the seasonality of the marine carbonate system of the southern Benguela focusing on three key points: the processes driving bulk stoichiometry, alkalinity production on the continental shelf, and the air-sea flux of CO2. Monthly samples were taken along the St. Helena Bay Monitoring Line in the southern Benguela for ten of the months in 2010. Samples were analysed for dissolved inorganic carbon (DIC) and total alkalinity (TA). Temperature, salinity, oxygen and nutrients were also measured.

Bibliography: leaves 33-34., An ADCP and water-column temperature study was conducted to determine the circulation aspect of subtidal-scale, density-driven exchange in Saldanha Bay, South Africa. Density-driven exchange conditions develop in response to synoptic-scale wind events in the southern Benguela region, even under light (