Ensuring the long-term sustainability of eastern US forests in the face of climate variability and change will require that forest managers have the best available climate change research to make sound management decisions. Ecosystem process models are now able to project forest landscape conditions in response to anticipated climate, natural disturbance, forest management, and their interactions; these projections can inform forest management decisions. However, there is no single scale which is perfectly suited to addressing all questions about climate change and management. Critical patterns which emerge at fine-scales may be over-averaged at larger scales and vice-versa. The objectives of this project were to: a) compare model outcomes from two modeling frameworks against empirical data and to each other, b) examine climate change, disturbance, and management interactions at Ft. Bragg, North Carolina; translate these procedures; and prepare a roadmap for deployment across other forested military installations.

Technology Description

The project team executed a two-stage approach for integrating climate, disturbance, and management projections at multiple scales. First, they calibrated and compared two models, a stand-scale model Perfect Plasticity Approximation-Simple Biogeochemistry and a landscape-scale model, LANdscape DIsturbance and Succession model, variant 2/Net Ecosystem Carbon and Nitrogen model, (LANDIS-II/NECN) against empirical data collected from two sites in the eastern US. Second, they applied the LANDIS-II/NECN and Photosynthesis and EvapoTranspiration model of plant growth operating within LANDIS-II (LANDIS-II/PnET) models to the Ft. Bragg landscape in central North Carolina under multiple projections of climate change, prescribed fires, and hurricanes. They assessed the strengths and weaknesses of each model and their respective capacity to project a suite of ecosystem processes, including succession, disturbance and nutrient cycling, given current and potential management practices and anticipated climate change.

The project team worked closely with the Ft. Bragg natural resource management team to refine data inputs and to develop scenarios via an iterative process that identified goals and scenarios, data needs, and desired outputs. The models will be delivered to Ft. Bragg fully parameterized and prepared for subsequent use, including full documentation and access to the open-source code for all models.

Interim Results

For the Test Phase, the project team compared perfect plasticity approximation-simple biogeochemistry (PPA-SiBGC) and LANDIS-II/NECN against empirical data and therefore the success criterion for each objective was an R2 value across the vector of annual simulated and empirical values. Performance metrics focused on near-term projections of ecological processes (not including climate change, management, or disturbance) that could be compared against 5-25 years of empirical data collected at Harvard Forest and the Jones Center.

The overall performance for both models was comparable. At Harvard Forest, both PPA-SiBGC and LANDIS-II NECN accurately predicted aboveground biomass (R2=0.98 and 0.98, respectively). LANDIS-II/NECN better predicted soil respiration (0.23 vs. 0.17) and aboveground net primary production (ANPP) (0.62 vs. 0.23), and PPA-SiBGC better predicted Net Ecosystem Exchange (NEE, a metric of overall Carbon balance, 0.67 v. 0.64). Only LANDIS-II/NECN predicted Nitrogen mineralization with an R2 value of 0.54. Mean R2 values across all the parameters were similar: 0.52 (PPA-SiBGC) vs. 0.6 (LANDIS-II/NECN). At Jones Farm, model fit was consistently much lower than at Harvard Forest with R2 values of 0.48 (PPA-SiBGC) and 0.17 (LANDIS-II/NECN). In PPA-SiBGC, model fit was better for aboveground biomass (0.54), ANPP (0.45), and NEE (0.74) than soil respiration (0.19). In LANDIS-II/NECN, model fit was consistently poor across all parameters (ranged from 0.03 to 0.36), indicating that additional calibration is necessary at Jones Farm.

For the Demonstration Phase, the project team simulated 50 years of forest change based on scenarios developed in collaboration with the Ft. Bragg Directorate of Public Works, Environmental Division. Because the scenarios represent future potential conditions, there was no empirical data available for comparison. Performance metrics therefore focused on the capacity to successfully integrate multiple processes and to produce reasonable projections that reflect expected changes given the scenarios. These scenarios were collaboratively developed scenarios with Ft. Bragg natural resource managers and represent their primary concerns about future management on the base. Therefore, the performance assessment included: climate change, prescribed fire, and hurricanes.

The preliminary results suggest that LANDIS-II/NECN and LANDIS-II/PnET are sensitive to different climatic drivers and different forest types (conifer vs. deciduous) have differing sensitivity to climate. Regarding disturbance and management, neither model estimates of aboveground biomass was particularly sensitive to variation in the prescribed fire management regime. This is in part because prescribed fire targets sub-canopy deciduous trees that have low biomass. Both models are highly sensitive to simulated hurricanes, highlighting model differences in regeneration and growth.


Successful demonstration and validation of models will help decision-makers integrate a multitude of management strategies into the context of the military mission and installation specific natural resources management plans. Forest managers will be able to use either LANDIS-II/NECN or LANDIS-II/PnET to estimate the effects of different management practices on the local installations over varying time horizons and spatial scale resolutions. Upon completion, this technology can be applied immediately at Ft. Bragg’s more than 89,000 acres of longleaf pine forests and at other Department of Defense installations with forested habitats.