Variability at low frequencies with wavelet transform and empirical mode decomposition: Aplication to climatological series

The aim of this study is to detect variability at low frequencies and trend of time series connected with climate using two different processing techniques. In previous work the wavelet transform and models of pure oscillations with statistical parameter setting were applied to the series of surface temperatures of the Orcadas Antarctic Station (Argentina) over 110 years. Periods of about 20 and 50 years were detected. The analysis highlighted the limitations of the usual calculations of trend involving a few decades if there is present a significant variability. Periods of the order of 150–200 years or more were also obtained, although they do not represent scales with physical meaning but the best simple oscillation which fits the nonlinear tendency. To improve the understanding of the long term behavior of the temperature series, the empirical mode decomposition method was applied in the present work to the same data and the trend or stationary component was obtained with more precision. The result of the comparison of trends was promising. It is advantageous to apply different methods to the same series in order to reveal complementary characteristics.