HR: 0800h
AN: A41C-0054    [Abstracts]
TI: Application of Maximum Likelihood Ensemble Filter (MLEF) With a Parameterized Chemistry Transport Model (PCTM) to Optimize Surface CO2 Fluxes
AU: * Lokupitiya, R S
EM: ravi@atmos.colostate.edu
AF: Department of Atmospheric Science, Colorado State University, Fort Collins, CO 80523 United States
AU: Denning, S
EM: denning@atmos.colostate.edu
AF: Department of Atmospheric Science, Colorado State University, Fort Collins, CO 80523 United States
AU: Zupanski, D
EM: Zupanski@cira.colostate.edu
AF: Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO 80523 United States
AU: Gurney, K
EM: keving@atmos.colostate.edu
AF: Department of Atmospheric Science, Colorado State University, Fort Collins, CO 80523 United States
AU: Zupanski, M
EM: ZupanskiM@cira.colostate.edu
AF: Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO 80523 United States
AB: Inverse modeling is widely used to optimize surface CO2 fluxes using observed CO2 concentrations in the atmosphere. A number of methods have been employed in inverse problems in geophysical research. Toward this goal, we apply the Maximum Likelihood Ensemble Filter (MLEF) algorithm, developed at Colorado State University. This is a novel ensemble filter approach, defined taking into account the experience (in terms of advantages and disadvantages) from Variational Methods, the Iterated Kalman Filter, and the Ensemble Trasform Kalman Filter. The MLEF is used with an offline chemistry/transport model (PCTM) to optimize the surface fluxes. To evaluate the technique, we ran the transport model with artificially generated CO2 observations sampled at CMDL flask locations by running the PCTM forward for four years with hourly CO2 fluxes generated by a terrestrial biosphere model (SiB) and monthly mean prescribed ocean fluxes. A time shifted true flux map was considered as the prior flux field for the assimilation step. Assimilation was done by using 8-12 week flux moving window and observations were assimilated at the end of the window. Spatial and temporal covariance in the retrieved fluxes are estimated by the optimization scheme by propagating the forecast error covariance forward through successive assimilation cycles. In the future, this technique will be implemented with real atmospheric observations.
DE: 0312 Air/sea constituent fluxes (3339, 4504)
DE: 0322 Constituent sources and sinks
DE: 1631 Land/atmosphere interactions (1218, 1843, 3322)
SC: Atmospheric Sciences [A]
MN: Fall Meeting 2005