The 14Constraint team catalyzed a community effort to build the International Radiocarbon Database (ISRaD; www.soilradiocarbon.org) and build a completely open-source repository for published soil radiocarbon data. The current version of ISRaD summaries 423 studies, with close to 15,000 radiocarbon measurements made at nearly 1000 different sites. Importantly, the data include not only bulk radiocarbon measurements but also measurements of physically, chemically and thermally separated fractions, soil CO2 fluxes and metadata including a range of soil properties that provide the opportunity to link the age distribution of carbon in soil to various stabilization mechanisms.
One of our goals was to fill gaps in global data coverage. We did this through collaboration with international researchers, especially those from regions like Africa and Latin America without easy access to 14C measurements. In addition we focused our efforts on expanding time series of radiocarbon data to provide strong constraints for modeling carbon cycling in soils and how this is affected by management (e.g. of carbon inputs) or by disturbances like deforestation.
Much effort went into advancing methods to facilitate data-model intercomparison efforts and communicating how radiocarbon data can be most effectively used to evaluate models. Many studies in the past have used a single metric - the mean age of carbon in soil organic matter derived from a bulk radiocarbon value. This provides a measure of the average time since the C in all soil organic matter was originally fixed from the atmosphere. However, it also assumes that soil organic matter is homogeneous, while soil organic matter is highly heterogeneous. We therefore proposed the use of an additional metric - the transit time, a measure of the mean time elapsed between when C was fixed from the atmosphere to when it is returned by respiration or decomposition back to the atmosphere as CO2. Mean soil carbon transit times range from years (tropical forests) to centuries (boreal and arctic regions), much shorter than the mean ages of organic matter (generally hundreds to thousands of years). For models attempting to predict changes in C stores over the next century, the transit time is a more constraining metric than the age as it is a measure of faster-cycling carbon that can respond to global change.
However, we also moved beyond the single metrics of mean age and transit time to take advantage of our ability to use a variety of methods to determine the distributions of ages and transit times. For example, we developed a method to determine the age distribution of C in mineral soil using a physical separation of particulate organic matter followed by thermal ramped oxidation. We used incubation of soils from different depth intervals to help define the shape of transit time distributions. Where available, we could also take advantage of time series of radiocarbon for constraining the age and transit time distributions of carbon. These can then be compared directly with age and transit time distributions predicted by models. Alternatively, model predictions can be used to generate predictions of radiocarbon distributions for any given year since 1960 for comparison with measurements and especially time series. Model-data comparisons then can be focused on what parts of the distribution are not well simulated - and which processes might be responsible.
