Variability and Trends in the Carbon Cycle

Abstract

This thesis examines the mechanisms that control the removal of anthropogenic CO2 from the atmosphere on timescales of decades to centuries. I focus on applying Bayesian statistical methods to constrain models of CO2 uptake, detect trends from observations and provide uncertainty analysis for integrated assessment models for climate change.

I first apply a data assimilation technique to a newly-released database of sur- face ocean carbon measurements. This application uses model-based information to compensate for data sparsity in a novel way and globally maps trends in the partial pressure of surface ocean pCO2. The results provide a new estimate of the growth in the total air-sea flux of CO2 and regional trend estimates. In particular, the result in the Southern Ocean provides new insights into the physical and chemical controls on CO2 uptake in that region of the ocean.

I then review model and data-based estimates of the Southern Ocean CO2 uptake and how it may change under climate change. I use an observational system observing experiment to show that the standing uncertainty in Southern Ocean CO2 uptake could be resolved with a float-based sampling network with between 150-200 members. I also show that changes to Southern Ocean CO2 fluxes predicted by models as a result of climate change will take decades to detect directly.

The last chapter focuses on the implications of uncertainty in the present and future carbon cycle for models that are used to price CO2 emissions as an externality of economic growth. The scientific contribution quantifies the implications of the large uncertainty in estimates of historical land use emissions of CO2 to future projections. This study novelly highlights the critical role that uncertainty in the response of the natural carbon sinks plays in evaluating the economic impacts of climate change and quantifies how reductions in that uncertainty can be used to improve the efficiency of climate policy design.