An optimal merger of data and models for carbon sequestration assessment at Ameriflux sites 
John D. Albertson, Duke University
Co-investigator: Amilcare Porporato, Duke University
Abstract
We address the need for reliable estimates of actual carbon uptake by terrestrial ecosystems and
the parallel need for robust predictions of how the fluxes will respond in the face of changing
environmental conditions. Currently, flux measurements at any point in time are treated as being
either completely reliable or completely unreliable, with a binary reliability index defined by
indicators such as the friction velocity (u*). When data are unreliable, empirical gap-filling
techniques are employed. We seek to reduce the uncertainty of both the measurement and model
estimates, through an optimal merger of the two, while providing a general framework for
comparisons across sites and regions.
The project results will be demonstrated on Duke Forest data, but will be apply in general to any
well-instrumented AMERIFLUX site.
A large investment is being made in measuring carbon flux from the atmosphere to vegetation.
However, due to practical issues the data only cover about 65% of each year for each site. The
approaches used to fill these gaps are questionable. This project seeks to combine the best of the
models that we have been developing and the best of the data we are collecting to provide
optimal and defensible estimates of carbon uptake.
The Data Assimilation System provides an explicit inclusion of the structure of both
measurement error and model error, with errors being treated as continuous variables rather than
binary. Such approaches are becoming routine in operational meteorological and climate studies.
With the maturation of the Ameriflux monitoring network and the continued effort at model
development and use, it is a natural progression to merge the data and models toward optimal
estimates. It is intended that the DAS developed in this project would be run for the full record
at any site to provide improved total carbon uptake estimates, provide explicit uncertainty
bounds on the estimates, and identify persistent weaknesses and pathologies in the model. The
end result is both an improved understanding of the fluxes and the underlying biophysical
system. In summary, DOE and NIGEC have spoken clearly of the need to reduce uncertainty in
estimates of present and future terrestrial carbon sequestration. This can only be accomplished if
uncertainties are explicitly accounted for, as this project proposes to begin to do.
Publications
Cassiani M, Franzese P. and Albertson J. D. 2007. A mixing model for intermittent concentration time series. Physics of Fluids, under review.
Cassiani M, Katul G. and Albertson J. D. 2007. The influence of canopy leaf area index on the airflow across forest edges: Large Eddy Simulation and analytical results. Boundary Layer Meteorology, under review.
Cassiani M., Albertson J. D. and Franzese P. 2007. Probability density function (PDF) and filtered density function (FDF) methods for turbulent scalar dispersion in incompressible flows. ETC11 - 11th European Turbulence Conference, Porto (Portugal) 25-28 June 2007. Published in Advance in Turbulence XI, Springer-Verlag.
Cassiani M., Radicchi A., Albertson J. D. and Giostra U. 2007. An efficient algorithm for scalar PDF modeling in incompressible turbulent flows; numerical analysis with evaluation of IEM and IECM micro-mixing models. Journal of Computational Physics, 223 (2): 519-550.
Cassiani M., Radicchi A. and Albertson J. D. 2007. Concentration fluctuations in canopy turbulence. Boundary Layer Meteorology, 122 (3): 655-681.
