You are here

Harvard Forest >

Harvard Forest Symposium Abstract 2008

  • Title: Reducing uncertainty about the effects of climatic variation on forest ecosystems by measuring, modeling, and analyzing intermediate-turnover carbon pools
  • Primary Author: Andrew Richardson (Northern Arizona University)
  • Additional Authors: Bryan Dail (University of Maine); Eric Davidson (University of Maryland - Center for Environmental Science); David Hollinger (USDA Forest Service); J. William Munger (Harvard University); Paul Schaberg (USDA Forest Service); Steven Wofsy (Harvard University)
  • Abstract:

    Existing ecosystem models fail to predict measured interannual variation in net ecosystem CO2 exchange (NEE), limiting their value for predicting potential climate change effects. Our objective is improved understanding of the causes of interannual variation in NEE in three important forest types of the NICCR NE Region, represented by the Howland, Bartlett, and Harvard Forest AmeriFlux sites. We believe that key processes involving significant interannual variation in intermediate-term (months–years) dynamic carbon pools are missing in most models. New field measurements (estimates of temporal variation of stemwood non-structural carbohydrates [TNC] and transient C sinks and sources in the forest floor litter layer) and ongoing CO2 flux and ecosystem measurements will be used in a novel modeling framework to evaluate mechanisms that might be included to improve climate and carbon cycle models.

    TNC pools (sugars and starches) will be measured (four times/year) in sapwood samples obtained by increment cores of the three dominant tree species at each of the three sites. Interannual variation in FF decomposition will be measured in multi-year O horizon decomposition studies at Bartlett and Howland. Inverse analysis methods (Bayesian, non-linear optimization using simulated annealing) will be used to parameterize a forest C cycle model with dynamic carbon pools. Ongoing measurements of whole-ecosystem and soil CO2 flux, soil C pool sizes and turnover rates, stand inventories, leaf area index and foliar chemistry will be used to further constrain model optimizations. Our modeling approach balances model complexity sufficient to accurately represent system dynamics against the requirement that the model be well-constrained by the available data. Information theoretic methods will be applied to quantitatively evaluate tradeoffs between model performance and complexity.

  • Research Category: Forest-Atmosphere Exchange