Ecosystem Model Comparison at Multiple Scales and Sites

The U.S. Department of Defense (DoD) manages 4.6 million hectares of land in the conterminous United States and an additional 32 thousand hectares in territories or overseas. DoD land managers face the challenging task of optimizing management to enhance the capabilities of current and future warfighters while meeting economic, social, and environmental goals. Recent advances in forest ecosystem modeling presently allow the simulation of a suite of dynamics relevant to management at the landscape scale, yet, there is no single scale perfectly suited to all management-related questions. Here, we compare results from two forest ecosystem models- Sortie-PPA (individual-tree-scale) and LANDIS-II (landscape-scale). We relied on a wealth of field data for model parameterization, calibration, and intercomparison and assessed model performance using the following metrics-NEE, ANPP, long-term successional dynamics, N cycling, and general model usability, performance, and transferability at the flux tower sites located at Harvard Forest and Jones Ecological Research Center. We also compared the ability of different single-objective optimization algorithms to improve model performance with net ecosystem exchange (NEE) and tested the ability of robust loss functions to improve convergence during optimization, to limit the influence of outliers. Finally, we compared management impacts on NEE, biomass and productivity between the two models at the largest military installation in the world, Fort Bragg, North Carolina, USA and prepared a roadmap for model deployment across forest military installations.