Radar reflectivity and radial velocity forward operator comparisons in EAKF data assimilation

Assimilating radar reflectivity and radial velocity into storm-scale numerical weather prediction (NWP) models has proven to be a challenging task, but has also shown to have great potential at improving severe weather forecasting. The work presented here explores the analysis and forecast impacts of radar data assimilation using the ensemble Kalman filter. In particular, we demonstrate that the nonlinear reflectivity forward operator and adaptive inflation when applied at different stages of the assimilation cycle, can generate significant changes in the resulting ensemble state's analysis. Doppler radial velocity assimilation often uses a fall velocity adjustment to account for the speed of falling hydrometeors in the calculation of radial velocity in the forward operator. The fall velocity adjustment has also been included in the state vector along with reflectivity in many previous works. vWe use the Ensemble Adjustment Kalman Filter (EAKF) present in the Data Assimilation Research Testbed (DART) system in combination with the WRF-ARW model to compare the impacts of calculating reflectivity and fall weighted velocity within the forward operator from interpolated prognostic state variables vs. using reflectivity and fall weighted velocity interpolated directly from their corresponding diagnostic variables included in the state. The results show that the analysis is very sensitive to whether the reflectivity and/or fall weighted velocity is included as a component of the state vector (rather than being computed within the DART system via a forward operator). Due to the nonlinearity of the radar forward operator, the use of adaptive inflation on the observational state's priors also impacts the results significantly depending on which assimilation method is used.