On NAO's predictability through the DFA method

Abstract

In this paper, the authors address a predictability study of the NAO index by using the time series analysis method known as “detrended fluctuation analysis” (DFA). This technique provides a quantitative measure of predictability by computing several piecewise fits (either linear or higher degree polynomial ones) to the cumulative series of fluctuations associated to the original series. The measurements, made on several examples of data available in the Internet, agree with other previous analyses (Fernández et al, 2003) assessing that the NAO signal is a slightly red one, whose prediction needs a deeper understanding of the underlying physics.