14 September 2015

Observations in Tundra Imply Needed Improvements in Land Models

Carbon cycle responses to warming from two versions of CLM (orange and red) are compared to observations (blue). CLM performed poorly against many of these observations.
Science

We describe an observational and modeling meta-analysis to benchmark land models and identify needed improvements. We applied the method to the Community Land Model (CLM) with two versions of belowground biogeochemistry (CN and Century).

Approach
  • We extracted benchmark metrics (e.g., belowground respiration, soil organic matter content) from 53 manipulation experiment studies across 17 high-latitude ecosystems.
  • We calculated a response ratio of a metric relative to the control.
  • We performed complimentary CLM4.5 simulations and analyzed discrepancies.
Impact
  • We identified poor representation of microbial activity, above- and belowground coupling, and nutrient cycling as the primary reasons for the discrepancies.
  • Identifying deficiencies in the model structure can motivate future experiments and focus model development efforts.
Contact
Nicholas J. Bouskill
Lawrence Berkeley National Laboratory (LBNL)
Acknowledgments

We acknowledge the constructive comments of Steven Allison and two anonymous reviewers that improved this manuscript. This research was supported by the Director, Office of Science, Office of Biological and Environmental Research of the US Department of Energy under contract no. DE-AC02-05CH11231 as part of the Next-Generation Ecosystems Experiments (NGEE Arctic) and the Regional and Global Climate Modeling (RGCM) program.