Embedding reality in a numerical simulation with data assimilation

Data assimilation (DA) is a synthesis technique based on the Bayesian filtering method by embedding observation/experiment data in a numerical simulation. It yields an accommodation ability to make a simulation real, and the better initial and boundary conditions can be automatically obtained. In statistical methodology, DA can be formulated in the state space model that draws much interest of the researchers in various domains such as the time series analysis, signal processing, and control theory. There are two types of DA in terms of a methodology; sequential DA and variational (non-sequential) DA. An ensemble-based sequential DA (EnSDA) has an advantage in terms of less human resources which is achieved by plugging into the existing ”forward” simulation codes. We briefly explain a recent advancement in EnSDA, and give a simple description on the relationship among the nonlinear non-Gaussian filters.