| Abstract |
Accurate atmospheric state analysis is a difficult scientific problem due to its chaotic nature. In global atmospheric state analysis for numerical weather prediction (NWP), data assimilation (DA) does this by unifying information of the atmosphere including observation data. However, the approximation accuracy of error covariance matrices (ECMs) for observation data and model forecasts are not sufficient, and this strongly limits the accuracy and validity of the analysis field. The purpose of this study is to overcome this problem by estimating entire ECMs with high accuracy using multiple ECM estimation methods. In order to see effects of ECMs on observation data, a new observation impact estimation method that can analyze the spatiotemporal structure of the impact is also constructed. This is the first study targeting the entire ECM estimation and refinement, which shows that the highly accurate estimation of ECMs based on the DA theory and the detailed observation impact analysis enable highly accurate atmospheric state analysis. |
