In this work we show the capability of the ECMWF’s Integrated Forecasting System (IFS) to couple near realtime composition analyses and forecasts with surface flux inversion.
To achieve this, an Ensemble of Data Assimilations (EDA) is used, which is an ensemble of perturbed 4DVar independent realizations. The methodology is based on using the ensemble information to compute Jacobians relating a model parameter to another, here concentrations and surface fluxes. The first results on carbon monoxide (CO) show realistic and physical constraint patterns on the fluxes and the ensemble inversion spread can provide an evolution of uncertainty on the fluxes. Finally, evaluation with surface data is promising, showing improvement on the scores, and it is expected to provide stronger improvement with higher resolution and different species (e.g. NO2)