Scale and phase characteristics of temperature series in China

The scaling behavior and phase probability distribution of temperature return series of Beijing station is investigated using Hilbert Huang transform (HHT) method. Results show that return series have nice scaling behavior, characterized by an exponent α (2<;α<; 4), well outside the stable Lévy regime 0<;α<;2. Besides, the phase probability distribution of temperature return series shows that abruptly changing behaviors indicate non-predictable and stochastic features on long scales, while gradually it is convergent to a spike value at the point φ=0 indicating a comparatively better prediction on short scales.