Ecosystem Demography model version 2 (ED2) Model structure Inputs and outputs Gap, big leaf, hybrid? Applications PEcAn framework for running ED2 and assimilating data
ED2: Overview Terrestrial biosphere model Mechanistic Size- and age-structured vegetation dynamics coupled with ecophysiology and biogeochemistry models Key sub-models: Leaf physiology (C3 from Farquhar, Bar-Berry, et al.) C allocation to leaves, fine roots, sapwood, stored leaf pool (from LEAF, Walko et al.) Active biomass (GPP – allocated pools + storage pool - respiration) Phenology (MODIS + equations) Soil biogeochemistry model (CENTURY)
ED2: Structure Calculated within each tile: Water (W), Internal energy (H) Carbon (C) Based on: Plant functional type (determines physiology) Size/age (determines access to water and light) Kim et al. 2012: Seasonal carbon dynamics and water fluxes in an Amazon rainforest Medvigy et al. 2009: Mechanistic scaling of ecosystem function and dynamics in space and time
ED2: Inputs & Outputs Key Inputs: Daily met data Plant functional type parameters Stand inventory data Soil type Land use Key Outputs: Ecosystem structure: Growth, mortality, AGB, etc. Ecosystem fluxes NPP, NEP, ET, etc.) Respiration/photosynthetic activity Hydrologic data Kim et al. 2012
ED2: Optimization Harvard Forest (optimized) Quebec (far from Harvard) Howland (near Harvard) Medvigy et al. 2009: Mechanistic scaling of ecosystem function and dynamics in space and time
ED2: Scaling heterogeneity Observed “Big-leaf” estimate ED2 estimate “ED2 captures subgrid scale biotic heterogeneity using a system of size- and age-structured partial differential equations that closely approximate the ensemble mean behavior of an individual-based stochastic gap model” Nov. 1994-Oct. 1996 Medvigy et al. 2009: JGR Biogeosciences
ED2: Predicting ecosystem response to climate change ED2 big leaf ED2 structured Kim et al. 2012: Seasonal carbon dynamics and water fluxes in an Amazon rainforest
ED2: Predicting and optimizing agricultural yields Dietz et al. 2013: Ecological Applications
ED2: Predicting post-disturbance carbon balance Frasson et al. 2015: JGR Biogeosciences
ED2: Testing the potential of new tools
Integrating data and models Continuous data, lab research Field measurements Models and modelers
PEcAn (Predictive Ecosystem Analyzer) Model-Data feedback: Uses Bayesian statistics to estimate parameters Quantifies contribution of each parameter to overall model variance Help to decide what variables are important to measure and assimilate LeBauer et al. 2013 Facilitating feedbacks between field measurements and ecosystem models
PEcAn: Informing parameter estimates with data Model-Data feedback: Uses Bayesian statistics to estimate parameters Quantifies contribution of each parameter to overall model variance Help to decide what variables are important to measure and assimilate
PEcAn: Constraining model parameters Coefficient of variation (CV): uncertainty associated with each parameter Elasticity: normalized sensitivity of modeled AGB to each parameter Percentage of total variance explained by each parameter
PEcAn: Comparing model parameters
PEcAn: Interface
PEcAn: Interface
PEcAn: Interface
Summary ED2 simulates an individual canopy-gap model at large scales Extensively validated in forest and ag. systems Can be applied broadly to test hypotheses, make predictions, etc. PEcAn runs ED2 and other models Provides a framework to assimilate data and refine parameters Enables broad community to use terrestrial biosphere (and other) models