WOMBAT v1.0: a fully Bayesian global flux-inversion framework

Abstract. WOMBAT (the WOllongong Methodology for Bayesian Assimilation of Trace-gases) is a fully hierarchical statistical framework flux inversion trace gases from flask, in situ, and remotely sensed data. extends the conventional synthesis through consideration correlated error term, capacity online bias correction, provision uncertainty quantification on all unknowns that appear model. We show, an observing system simulation experiment (OSSE), these extensions are crucial when data indeed biased have errors spatio-temporally correlated. Using GEOS-Chem atmospheric transport model, we show able to obtain posterior means variances non-fossil-fuel CO2 fluxes Orbiting Carbon Observatory-2 (OCO-2) comparable those Model Intercomparison Project (MIP) reported Crowell et al. (2019). also find WOMBAT's predictions out-of-sample retrievals obtained Total Column Observing Network (TCCON) are, most part, more accurate than made by MIP participants.