Sensitivity of Flux Accuracy to Setup of Fossil Fuel and Biogenic CO2 Inverse System in an Urban Environment

Abstract:

The Indianapolis Flux Experiment (INFLUX) aims to utilize a variety of measurements and a high resolution inversion system to estimate the spatial distribution and the temporal variation of anthropogenic greenhouse gas (GHG) emissions from the city of Indianapolis. We separated biogenic and fossil fuel CO₂ fluxes and tested the sensitivity of inverse flux estimates to inverse system configurations by performing Observing System Simulation Experiments (OSSEs). The a priori CO₂ emissions from Hestia were aggregated to 1 km resolution to represent emissions from the Indianapolis metropolitan area and its surroundings. With the Weather Research and Forecasting (WRF) model coupled to a Lagrangian Particle Dispersion Model (LPDM), the physical relations between concentrations at the tower locations and emissions at the surface were simulated at 1 km spatial resolution, hourly. Within a Bayesian synthesis inversion framework, we tested the effect of multiple parameters on our ability to infer fossil fuel CO₂ fluxes: the presence of biogenic CO₂ fluxes in the optimization procedure, the use of fossil fuel CO₂ concentration measurements, the impact of reduced transport errors, the sensitivity to observation density, and the spatio-temporal properties of prior errors. The results indicate that the presence of biogenic CO₂ fluxes obviously weakens the ability to invert for the fossil fuel CO₂ emissions in an urban environment, but having relatively accurate fossil fuel CO₂ concentration measurements can effectively compensate the interference from the biogenic flux component. Reduced transport error and more intensive measurement networks are two possible approaches to retrieve the spatial pattern of the fluxes and decrease the bias in inferred whole-city fossil fuel CO₂ emissions. The accuracy of posterior fluxes is very sensitive to the spatial correlation length in the prior flux errors which, if they exist, can enhance significantly our ability to recover the known fluxes and reduce the posterior uncertainties.