Uncertainty in initial forest structure and composition when predicting carbon dynamics in a temperate forest

The initial or current ecosystem state is a necessary factor in forecasting how terrestrial ecosystems will respond to changes in climate, CO2, and other environmental forcings over the upcoming decades. Terrestrial biosphere models are important forecasting tools, but to improve our understanding of large-scale terrestrial ecosystem function, we need to consider data from a number of sources and scales. Today remote sensing is improving our ability to derive information on forest structure and composition at a variety of scales, but their uncertainties in deriving these products to predicted carbon fluxes have not been investigated. This study investigated how uncertainties in forest structure and composition initialized at a temperate North American forest using a state-of-the-art terrestrial biosphere model affect predictions of carbon dynamics in the short and decadal time-frame. Uncertainties in net carbon predictions are compared to a ±20% uncertainty value estimated for the terrestrial sink component of the global carbon budget (Pan et al., 2011). For short-term simulations, a ±20% uncertainty in both forest structure and composition is enough to predict a net carbon flux variation of ±21.5% (±0.077 kg C/m2/year) and ±20.5% (±0.092 kg C/m2/year) respectively. For medium-term (11–40 years) simulations, only a 50% uncertainty in the initial forest structure predicts a net carbon flux variation beyond the ±20% threshold, but uncertainties in net carbon flux variation beyond ±20% were predicted from a 5% change in initial composition after 25 years of simulation. Remote sensing-derived forest structure and composition using LiDAR and imaging spectroscopy were also initialized, with all short term simulated net carbon fluxes within the ±20% cutoff. These results indicate that an accurate and full description of forest structure and composition from remote sensing within a ±20% uncertainty can be adequate in producing improved forecasts of terrestrial ecosystems for the next few decades.