Direct Assimilation of Radar Reflectivity Data Using Ensemble Kalman Filter Based on a Two-Moment Microphysics Scheme for the Analysis and Forecast of Typhoon Lekima (2019)

We investigate the impact of directly assimilating radar reflectivity data using an ensemble Kalman filter (EnKF) based on a double-moment (DM) microphysics parameterization (MP) scheme in GSI-EnKF data assimilation (DA) framework and WRF model for a landfall typhoon Lekima (2019). Observations from a single operational coastal Doppler are quality-controlled and assimilated. Compared with the baseline experiment initialized by GFS analysis, the reflectivity data assimilation experiment (Z-DA) resulted in an obvious improvement in both structural analysis and typhoon forecast skills in terms of intensity, precipitation, and track. Sensitivity experiments were conducted to evaluate the ability of EnKF to update certain state variables considering that the degree of freedom of analytical variables increased with a DM MP scheme. When either the total number concentration or other large-scale state variables that are not directly linked to reflectivity observations via the observation operator are not updated, the tendency of RMSIs and PS to be imbalanced is significantly increased during DA cycles compared to those of Z-DA with updating a full set of state variables, resulting in increased intensity and track forecast errors. These results indicate that the reliable ensemble covariance could handle the underconstraint issue associated with the DM scheme, and helps in obtaining more physically balanced analytical fields.