Challenges in simulating the historical trajectory of carbon stocks on land

The strength of the land carbon sink is a critical and uncertain aspect of the global earth system, exerting a strong influence on the airborne fraction of CO2 emissions and therefore having strong implications on permissible emissions to reach 1.5C or 2C temperature targets. Estimates of the cumulative land carbon sink over the historical period, though also uncertain, indicate broadly that land areas were a net carbon source up until about the 1940’s due to human driven land use change, including strong carbon losses due to deforestation and wood harvest. After the 1940’s the ongoing land use change effects are compensated for by carbon accumulation associated largely with the CO2 fertilization effect. Earth system models and the land models within them, both in fully coupled and land-only modes, show a wide range of behavior in terms of their simulated carbon stock trends. While some models can reproduce estimates of the cumulative land carbon flux from 1850 to present day, most models still do not reasonably simulate the historical trajectory of the cumulative land carbon flux. We will review the current understanding of how CO2 fertilization, climate-carbon feedbacks (including the permafrost climate-carbon feedback), and various forms of land use and land management change are represented in models and hypothesize about where the biggest limitations are. We will also explore how parametric uncertainty of current generation parameter heavy land models is contributing to the ability of models to replicate the historical trends. (Note: If there are other abstracts on this topic from other contributors, we could refine what we present to focus more on the land use and land management change aspect and/or the parameter uncertainty aspect of this problem to complement these other abstracts).