Carbon Data Assimilation Using Maximum Likelihood Ensemble Filter (MLEF)
Dusanka
Zupanski, Colorado State University, Zupanski@cira.colostate.edu
(Presenting)
Scott
Denning, Colorado State University, denning@atmos.colostate.edu
Marek
Uliasz, Colorado State University, marek@atmos.colostate.edu
Andrew
E.
Schuh, Colorado State University, aschuh@atmos.colostate.edu
Milija
Zupanski, Colorado State University, ZupanskiM@cira.colostate.edu
Ensemble based data assimilation approaches are promising techniques for carbon science problems, especially because of the their capability to calculate realistic flow-dependent uncertainties of the estimated variables (e.g., carbon fluxes). We are developing a generalized framework for carbon flux estimation from multiple streams of carbon observations for the purposes of a NASA supported North American Carbon Program (NACP) research. For this research we employ Maximum Likelihood Ensemble Filter (MLEF), an ensemble-based data assimilation approach recently developed at Colorado State University. The MLEF approach is especially suitable for the carbon data assimilation problems because of its capability to address non-linear data assimilation problems, involving biased atmospheric and carbon transport models.
In this presentation, we address the carbon inversion problem using an offline Lagrangian Particle Dispersion Model (LPDM) forced by SiB-RAMS fluxes. Experimental results estimating biases in respiration and photosynthesis carbon fluxes using simulated tall tower CO2 observations will be presented and discussed.