Application of the Quasi Static Variational Assimilation (QSVA) Method to Heavy Rainfall Cases over the Korean Peninsula

Abstract

In this study, the usefulness of the Quasi Static Variational Assimilation (QSVA) method in assimilating radar data for heavy rainfall cases, which are related to a high degree of nonlinearity, is investigated by comparing it with the single or multiple outer-loop Four Dimensional Variational (4D-Var) method. In the QSVA method, the length of the assimilation window is gradually increased, and the starting point of the current minimization task comes from the minimizer of the previous one.
In some cases, the rainfall and wind forecasts are improved to the greatest extent through the use of the multiple outer loops. On the contrary, in the other cases, the use of the QSVA method improves the rainfall and wind forecasts the most significantly. According to the analyses of the nonlinearity of the minimization problem, the cases, in which the QSVA method is the most effective in assimilating radar data, are related to a relatively high degree of nonlinearity. Owing to the quasi-static adjustment of the QSVA method, the possibility of getting trapped near the local minimum is reduced, and the QSVA method results in better analyses and forecasts.