Quantification of Transport Model Error Impacts on CO2 Inversions Using NASA's GEOS-5 GCM

Abstract:

Remote sensing observations of CO2 offer the opportunity to reduce uncertainty in global carbon flux estimates. However, a number of studies have shown that inversion flux estimates are strongly influenced by errors in model transport. We will present results from modeling studies designed to quantify how such errors influence simulations of surface and column CO2 mixing ratios. These studies were conducted using the Goddard Earth Observing System, version 5 (GEOS-5) Atmospheric General Circulation Model (AGCM) and the implementation of a suite of tracers associated with errors in boundary layer, convective, and large scale transport. Unlike traditional tagged tracers which are emitted by a certain process or region, error tracers are emitted as air parcels are transported through the atmosphere. The magnitude of error tracer emissions is based on previously published ensembles of AGCM simulations with perturbations to subgrid convective and boundary layer transport, and on comparisons of several reanalysis products to estimate errors in large scale wind fields. Transport error tracers are simulated with several different e-folding lifetimes (e.g. 1, 4, 10, and 30 day) to examine differences between transient and persistent model errors. This quantification of transport error is then used in an illustrative Bayesian synthesis inversion to demonstrate how transport errors influence surface CO2 mixing ratios and how this translates into inferred biosphere flux error.