Statistical approaches to soil carbon dynamics

Terrestrial ecosystems absorb ≈30% of anthropogenic CO2 emissions in a process termed the land sink. This process thus mitigates a large fraction of current and future climate change, and Earth’s future climate depends greatly on whether or not the land sink continues. Accumulation of soil organic carbon (SOC), is responsible for a large fraction of carbon absorbed by the land sector, yet we currently lack sufficient observational constraints on changes in SOC at the global scale. Moreover, the computational models that we rely on to simulate SOC dynamics are too complex to effectively use the limited available data, leading to very large uncertainty in their projections. To help address these challenges, we develop simple-yet-powerful statistical models of soil organic carbon degradation that use available observations of carbon turnover time and radiocarbon dating to constrain the ~10-100 year dynamics of SOC, the relevant time scale over which societies can plan for climate change. We show that these models can be independently parameterized from available data, and have predictive performance on par or exceeding state-of-the-art models, with many fewer parameters.