Soil carbon is a recent but critical sub-component of Earth system models that contributes significant uncertainty to carbon cycle simulations in the current Intergovernmental Panel on Climate Change (IPCC) assessment. Understanding and reducing this uncertainty is a high priority for the next round of IPCC assessments. To this end, experts in soil carbon dynamics gathered in Breckenridge, Colo., for 3 days to discuss next steps for global soil carbon modeling.
The workshop brought together a diverse group of specialists in soil carbon modeling, model-data assessment and integration, data product development, and soil carbon measurements. The meeting was structured around informative morning presentations about current research and afternoon discussion sessions. Participants generated recommendations for future research directions around 3 broad needs in the community: data products, model development, and data-model integration.
Data products are useful both for calibrating and evaluating models, and participants agreed that recently available biome and global soil carbon maps are fantastic first steps. Both models and data sets, however, need to explicitly report soil depth to be rigorously comparable. There was significant optimism among participants about the development of a global soil respiration database that could lead to a new data product for the modeling community, although models will have to be modified to report aggregated heterotrophic and below-ground autotrophic respiration. Attendees stressed the need for formal uncertainty analyses to be included with all future data products, as well as a continued need to update soil survey information to improve accuracy, especially in the high northern latitudes.
A wide range of processes could be incorporated into models to improve the representation of decomposition. Top recommendations from participants included the effect of depth on decomposition, permafrost freeze-thaw dynamics, microbial kinetics, sub-grid scale hydrology and topography, nutrient dynamics, land use change, and fire.
Model evaluation recommendations centered around model-data comparison metrics, scaling, and understanding model behavior. There was much debate among the participants concerning appropriate metrics and scale to evaluate models. In general, those attending agreed that pedon, regional, biome, and global totals are likely more informative than grid-by-grid comparisons, given current data product accuracy.
Although participants acknowledged that data-model integration techniques can improve model-data fit, without formal uncertainties in the data product, it remains unclear whether there is an improvement in real-world representation. There was some interest by participants in integrating site-specific data sets (i.e., flux tower observations and climate change experiments) into models. However, there was also debate over whether issues of scaling from measurement to simulation have been adequately addressed.
Finally, there was an emerging consensus at the meeting that it is critical to understand the drivers of soil carbon dynamics in the models, as well as the real world. This insight can be used to target the design of field experiments and future model development.
This promises to continue to be a dynamic and exciting field in the future for modelers, data product developers, and experimentalists. This workshop was funded by a National Science Foundation Research Coordination Network (RCN) Forecast Of Resources and Environmental Changes using data Assimilation Science and Technology (FORECAST) grant.
—K. Todd-Brown and Y. Luo, Department of Microbiology and Plant Biology, University of Oklahoma, Norman; [email protected]