A forest biogeochemistry model intercomparison on the East Coast of the United States

Recent advances in forest ecosystem modeling allow the simulation of a suite of dynamics relevant to management at the landscape scale. Yet, there is no single scale perfectly suited to the multitude of management-related questions. Here, we conducted a model intercomparison exercise using two forest biogeochemistry models, PPA-BGC and LANDIS-II. We simulated past-decade conditions at flux tower sites in Harvard Forest, MA and Jones Ecological Research Center, GA. We mined the wealth of field data available for both sites to perform model parameterization, validation, and intercomparison. We assessed model performance using the following time-series metrics: NEE, ANPP, aboveground biomass (B_AG), C (C_AG), and N (N_AG), belowground biomass (B_BG), C (C_BG), and N (N_BG), soil respiration (r_soil), and, species total biomass (B_Sp) and relative abundance (n_Sp). We also assessed static observations of soil organic C (SOC) and N (SON), and concluded with an assessment of general model usability, performance, and transferability. Despite substantial differences in design, both models achieved a high level of accuracy across the full range of metrics. While each model displayed strengths and weaknesses, the PPA-BGC model indicated the best overall correspondence to observational data for both sites across the 11 temporal and 2 static metrics tested (HF EMS R^2 = 0.73, +0.07, RMSE = 4.84, -0.39; JERC RD R^2 = 0.76, +0.04, RMSE = 2.69, -0.17).