Ojima, Dennis: Colorado State University (Project Lead)
Estep, Don: (Co-Investigator)
Parton, William: Colorado State University (Co-Investigator)
Schimel, David (Dave): JPL (Co-Investigator)
Project Funding:
2007 - 2011
Funded by NASA
Abstract:
It is now widely accepted that, while daily and interannual variability in carbon fluxes are controlled by the biophysical environment, longterm uptake is determined mainly by disturbance history, land management and other factors affecting ecosystem structure. This makes projecting decadal future sequestration rates an initial value problem, determined, in part, by the distribution of carbon and nitrogen between pools at the beginning of the forecast. Satellite data and data assimilation techniques have been used in studies of 'fast fluxes': we propose to focus this investigation on constraining both fast fluxes and long term sequestration rates combining satellite and in situ observations in a data assimilation context. The focus on 'initial conditions' adds an additional set of scaling issues. While drivers of fast fluxes are often uniform or predictable over large scales (eg climate anomalies: Keyer et al 2000), disturbance and land management generally create complex landscapes with heterogeneity at very local scales. Biogeochemical modelers have used large-scale coherence in patterns of temperature, radiation and precipitation to estimate carbon responses globally and regionally using relatively simple models and data sets (Braswell et al 1997, Myneni et al 1997, Asner et al 2003). Estimating the effects of prior disturbance and management must focus on the opposite and finest end of the scale spectrum (Schimel et al 1997b). The very different scales of climate vs disturbance effects may, in fact, help to separate their signatures in satelite observational time series. The methodology we will use treats both initial values and fixed parameters (e.g., temperature dependences) as ?control parameters? of the solution in that they both govern the evolution of the system to the next timestep. The variational framework we use allows us to both constrain estimates of ecosystem state with observations during model execution, and provides error analyses to track the influence of uncertainty in estimates of ecosystem state on fluxes. This information then feeds into assessments of model uncertainty and allows the targeted improvement of observing systems. The project should greatly enhance our quantitative knowledge and predictive ability for disturbance driven carbon-water dynamics in complex landscapes.
More details may be found in the following project profile(s):
