G05_04

Evaluating the new generation of soil organic carbon models using radiocarbon

Brunmayr A¹,  Graven H¹, Moreno Duborgel M^2,3

¹Imperial College London, London, United Kingdom, ²Eidg. Forschungsanstalt WSL, Birmensdorf, Switzerland, ³ETH Zurich, Zurich, Switzerland

In recent decades, soil carbon models have moved away from using ill-defined conceptual carbon pools and instead started using operationally defined pools which can be individually extracted from soil samples and measured for their carbon and ¹⁴C content.  One major advantage of simulating measurable pools is the ability of directly assimilating not only bulk soil data but also pool-specific carbon and ¹⁴C measurements, which greatly aids with model calibration and validation.  However, only a handful of these new-generation models with measurable pools has been tested with ¹⁴C data from field samples.  As the dataset of ¹⁴C measurements for bulk soil and individual soil pools is expanding ever more rapidly, we should utilize the power of ¹⁴C as a carbon cycle tracer and take full advantage of pool-specific ¹⁴C data when calibrating new-generation models.  In this study, we evaluate various new-generation models (including Millennial, MEND, SOMic, …) with ¹⁴C measurements for individual soil pools across diverse geoclimatic regions.  While some models with measurable pools perform rather well, a few seem to strongly underestimate the time scales of carbon storage in soils and thus produce unrealistic ¹⁴C values which miss the datapoints by large margins.  These results raise questions about the validity of these models’ carbon turnover predictions and demonstrate the importance of verifying the consistency of model output with measured ¹⁴C data.