A0633
Title: Modelling atmospheric transport error in 4-D using the Vecchia approximation
Authors: Michael Bertolacci - The University of Western Australia (Australia) [presenting]
Abstract: The exchanges of CO2 between the atmosphere and ecosystems are a vital and difficult-to-quantify component of Earth's carbon cycle. A widely used Bayesian technique for estimating these exchanges is flux inversion, where observations of atmospheric CO2 concentrations are traced upwind to infer the fluxes. This requires a transport model that simulates the motion of CO2 in the atmosphere. The quality of the resulting estimates is limited by the fidelity of the transport model, and current approaches to quantifying transport error are limited by computational and statistical constraints. A new method for modeling the transport error in 4-D (3-D space + 1-D time) is discussed using the Vecchia approximation. This is implemented in an existing inversion framework called WOMBAT (the WOllongong Methodology for Bayesian Assimilation of Trace-gases). Results show that this approach is practical, and simulations show that this improves the accuracy and uncertainty quantification of the flux estimates.
