Impacts of Assimilating Vertical Velocity, Latent Heating, or Hydrometeor Water Contents Retrieved From a Single Reflectivity Data Set

Assimilation of observation data in cloudy regions has been challenging due to the unknown properties of clouds such as cloud depth or cloud drop size distributions (DSD). Attempts to assimilate data in cloudy regions generally assume a DSD, but most assimilation systems fail to maintain consistency between models and the observation data, as each has its own set of assumptions. This study tries to retain the consistency between the forecast model and the retrieved data by developing a Bayesian retrieval scheme that uses the forecast model itself for the a priori database. Through the retrieval algorithm, vertical profiles of three variables related to the development of tropical cyclones, including vertical velocity (VV), latent heating (LH), and hydrometeor water contents (HYDRO), are derived from the same reflectivity observation. Each retrieved variable is assimilated in the data assimilation system using a flow‐dependent forecast error covariance matrix. The simulations are compared to evaluate the respective impact of each variable in the assimilation system. In this study, three assimilation experiments were conducted for two hurricane cases captured by the Global Precipitation Measurement satellite: Hurricane Pali (2016) and Hurricane Jimena (2015). Analyses from these two hurricane cases suggest that assimilating LH and HYDRO have similar impacts on the assimilation system while VV has less of an impact than the other two variables. Using these analyses as an initial condition for the forecast model reveals that the assimilations of retrieved LH and HYDRO were able to improve the track forecast as well.