Toward Reduced Transport Errors in a High Resolution Urban CO2 Inversion System

Atmospheric transport is one of the key components in estimating posterior surface fluxes of greenhouse gases using an atmospheric inversion system. Assimilation systems have the capability to combine advanced atmospheric models with various meteorological observation systems and therefore improve the current modeling performances of transport models. In this paper, we evaluate the impact of transport solutions on inverse CO₂ emissions when assimilating various meteorological observations in our high resolution atmospheric assimilation framework. We compare the impact of assimilating standard surface stations, wind profiles from a Doppler lidar, and observations from regular commercial flights, to assess the improvement of the modeling performances. The different simulations were coupled into our urban inversion system for the Indianapolis Flux Experiment (INFLUX) project, using the Weather Research and Forecasting (WRF) mesoscale atmospheric model, with nested grids covering the city of Indianapolis and the surrounding region at high resolution. We continuously assimilated the available observations into the WRF simulations using a Four Dimensional Data Assimilation (FDDA) technique. The WRF-FDDA system was also coupled to the high resolution CO₂ emission product Hestia to provide three-dimensional fields of atmospheric CO₂ mixing ratios that can be compared to the CO₂ observations from the 12 operational INFLUX towers measuring continuously over the entire period.

To understand the sensitivity of meteorological data assimilation on the posterior fluxes, we used the WRF-FDDA simulations to general multiple transport scenarios. The adjoint of the transport, as described in the inverse system, is represented by influence functions which are computed using a Lagrangian Particle Dispersion Model (LPDM) in backward mode driven by the WRF solutions. We finally present the effect of assimilating various observations on the inverse CO₂ emissions by performing multiple inverse simulations. We show that assimilating additional meteorological measurements has a significant impact on our CO₂ emission estimates at high resolution, in both time and space.