Assimilating 3D Cloud Radar Reflectivity Retrieved Based on MODIS for Typhoon Forecast

3D clouds (W-band cloud radar reflectivity) of Typhoon Talim-2023 at its initial development stage were retrieved based on MODIS observations, using the CGAN-EBPF (Conditional Generative Adversarial Networks – Ensemble Bin Probability Fusion) technique. The 3D cloud radar reflectivity is assimilated into WRF with the hydrometeor and latent heat nudging four-dimensional data assimilation (HLHN-FDDA) scheme developed in NCAR. The retrieved 3D W-band cloud radar reflectivity is firstly projected to S-band radar reflectivity equivalent with an empirical projection formulation, developed based on collocated S-band radar observations and the 3D W-band radar reflectivity for previous landfall Typhoon cases, and then assimilated with the hydrometeors and latent heat retrieved from the 3D S-band radar reflectivity equivalent. A group of numerical experiments are conducted to assess the impact of the MODIS-derived 3D radar data. It is found that assimilating the MODIS 3D cloud radar reflectivity significantly helps initialize the cloud and precipitation structures of Typhoon Talim and 3-hour forecasts. It is also found that the data assimilation impact is sensitive to the microphysical parameterization scheme chosen for the WRF model. The impact of the data assimilation diminishes beyond 6 – 9-hour forecast for the current case. Thus, further research should be performed to improve the dynamical and thermodynamic aspects of the model with the MODIS 3D cloud radar reflectivity.