Recent Trends in the Land Carbon Cycle

Abstract

Land ecosystems absorb about a quarter of all human emissions of carbon (C) by fossil fuel burning and land use change. This percentage varies greatly within years due to the land ecosystem response to climate variability and disturbance. Significant uncertainties remain in our knowledge of the magnitude and spatio-temporal changes in the land C sinks. The aims of my thesis are 1) to evaluate the capacity of different dynamic global vegetation models (DGVMs) to reproduce the fluxes and stocks of the land C cycle and 2) to analyse the drivers of change in the land C over the last two decades (1990-2009). In the first part of this thesis I evaluated the DGVM results over two regions: the Northern Hemisphere (NH) and the Tropics. Over the NH DGVMs tend to simulate longer growing seasons and a greater positive leaf area index trend in response to warming than that observed from satellite data. For the tropical region we found a high spatial correlation between the DGVMs and the observations for C stocks and fluxes, but the models produced higher C stocks over the non-forested areas. In the second part I studied the processes controlling the regional land C cycle. The findings can be summarized as: (1) the land CO2 sink has increased over the study period, through increases in tropical and southern regions with negligible change in northern regions; (2) globally and in most regions, the land sinks are not increasing as fast as the growth rate of excess atmospheric CO2 and (3) changes in water availability, particularly over the dry season, played a fundamental role in determining regional trends in NPP. My work seeks to improve our understanding of the relationship between the C cycle and its drivers, however considerable research is needed to understand the role of additional processes such as land use change, nitrogen deposition, to mention just a few.