Estimate of atmospheric predictability and development of prediction model using ensemble forecast assimilation in nonlinear dynamical system

摘要

In this study, a limit of predictability for the atmosphere is estimated based on analog weather maps in the historical data, and a new type of ensemble forecast assimilation technique is developed in order to improve the forecast skill in the nonlinear dynamical system. The limit of the predictability (denoted as P) is defined as the time taken for the initial difference (E0) of the analog pair to reach the climate noise level which is defined by one standard deviation from the long term mean of the fluctuation in the observed atmosphere. Although a total of 185,547,600 pairs of the weather maps are searched, there are no good analog pairs to investigate the difference growth rate for a sufficiently small E0 of the analog pairs. For this reason, the behavior of E0 is explained by a quadratic error growth model. Regressing the quadratic error growth model to the scattergrambetween P and E0, it is estimated that P would extend 2.88 days when E0 is reduced to 1[?]e for sufficiently small E0. The limit of predictability P varies depending on variable bythe atmospheric boundary condition such as El Ni~no, La Ni~na, Pacific/North American (PNA), and North Atlantic Oscillation (NAO). In the case of PNA+ and NAO-, the differences of the analog pairs grow slower than the average showing the e-folding time of 3.17 and 3.07 days, respectively. Conversely, in the case of La Ni~na and PNA-, thedifferences grow faster than the average showing the e-folding time of 2.72 and 2.69 days,respectively. These results are also verified by hindcast datasets. ･･･