Abstract
In this study, the usefulness of the Adjoint Sensitivity-based Data Assimilation (ASDA) method in assimilating radar data was investigated by comparing it with the Three/Four Dimensional Variational (3/4D-Var) method. A total of 10 heavy rainfall cases over the Korean Peninsula were selected and classified as one of four Heavy Precipitation Systems (HPSs) according to their phenomenological properties.
The Quantitative Precipitation Forecasting (QPF) skill is evaluated by computing the threat and bias scores, and the Root Mean Square Errors (RMSEs) of the simulated radial velocity are also calculated. The forecast skill of the ASDA method is comparable to that of the 4D-Var method in most of cases. This is because the ASDA method has some of the advantages of the 4D-Var method such as flow-dependent background error covariance and well-balanced analysis. In addition, the dependence of the QPF skill of the ASDA method on the characteristics of heavy rainfall cases is analyzed by calculating time-lagged autocorrelations of the observed radar data.
