Comparison of All-sky Radiance Assimilation with Different Cloud Control Variables for FV3GFS

A new global model using the Finite-Volume Cubed-Sphere Dynamical Core (FV3) coupled to the GFS physics packages was built at NOAA/GFDL. It was further embedded into the NOAA NEMS and coupled to the data assimilation system by NCEP/EMC. This next generation global forecast system called FV3GFS will be in operations in 2019. One important model upgrade is to use the GFDL cloud microphysics scheme to replace the Zhao-Carr grid scale condensation and precipitation scheme in the original GFS physics package.

In the planned 2019 implementation of the FV3GFS, the normalized cloud water is used as the cloud control variable in all-sky radiance assimilation. Since the hydrometeors, including cloud liquid water, cloud ice, rain, snow and graupel, are the prognostic variables with the GFDL microphysics scheme, an alternative choice is to use individual hydrometeors as cloud control variables. In this study, we performed several 4DEnVar experiments to evaluate the impact of changing cloud control variables in all-sky radiance assimilation. Radiance observations from AMSUA and ATMS are assimilated in all-sky approach.

The first step is to replace cloud water with cloud liquid water and cloud ice as the control variable. By doing so, the decomposition of cloud water into cloud liquid water and cloud ice, and its associated tangent linear and adjoint operators for the conversion between control and state variables are no longer required. This change allows more AMSUA observations to be assimilated and it also improves the first guess fit to the AMSUA radiance observations. The next step is to include all five hydrometeors as control variables. This change allows precipitation affected radiance observations to be assimilated. The use of individual hydrometeors as control variables has neutral to positive impact on 500 hPa ACC and positive impact on hurricane track forecast. More detailed analyses of the results as well the system tuning will be presented at the conference.