Recent advances in atmospheric composition data assimilation at JCSDA

The atmospheric composition team (COMPO) at JCSDA worked on developing the JEDI capability for numerous things. First, the 4DEnVar and its stability were tested for trace gas and aerosols. In parallel work, tracers were integrated in the FV3-linear model, allowing atmospheric composition 4DVar using the tangent linear and adjoint code. This work was followed by testing and developing the capability of the OOPS and FV3JEDI code to use novel methodologies such as the Hybrid tangent linear model, i.e. using the non-linear ensemble information to correct the linear model and adjoint tendencies to account for chemistry and physics. This work showed promising results and improved the 4DVar minimization results. Much effort has been put into cycling the JEDI 4DEnVar with the GEOS-CF model with the 4DVar and 4DEnVar. An EDA + high-res data assimilation configuration is currently designed for a scientific assessment and publication.

A generalized aerosol lookup table loader for optical properties has been developed into the CRTM code allowing user to come up with their own set of aerosol schemes and optical coefficients. This gives CRTM more genericity for aerosol assimilation. Also, a lot of developments have been made in the Vader JEDI repository to allow aerosol variable transforms for particulate matter surface observations work also has started to use VADER as a variable change tool to allow users to change their aerosol state to match certain aerosol look-up tables, e.g. CMAQ aerosol model could use the NASA GEOS5 lookup tables. This would make the interagency joint work on aerosol data assimilation much more transferable and inter-comparable. Part of our team is also looking at reflectance L1b aerosol data assimilation. Work has been done to enhance part of the CRTM code to compute H(x) and look at the Jacobian matrix at different locations and channels. Research is ongoing for scene classification to differentiate between cloud versus aerosol-contaminated pixels.

Last but not least the COMPO team has put a significant priority on the demonstration of the capability JEDI code to assimilate the new TEMPO instrument. Less than 24 hours after the official L2 public release we published a demonstration news web article showing data assimilation results of NO2 fields using TEMPO. This shows how well-fitted the JEDI code is for operations. Work is ongoing to look into more details of the 4DEnVar TEMPO data assimilation and QC strategies and more importantly, develop an emission inversion capability with JEDI.