Beyond linear correlation: Strong nonlinear structures in diurnal temperature range variability over southern China

Abstract

The correlation structure and dynamical properties in diurnal temperature range (DTR) variability over southern China during 1979–2018 are studied in this study. Detailed results show that the linear correlation of DTR variability can be well captured by the first-order autoregressive (AR (1)) process over most of the southern stations. Since the standard detrended fluctuation analysis (DFA) based on power law assumption is designed to determine the long-range correlation strength but not the short-term correlation strength, a novel way of modified DFA is proposed to estimate the parameter of AR (1) process well over the range of short time scales. Moreover, beyond the linear correlation, the strong nonlinear structures are also detected in DTR variability with marked residual structures, i.e. significant time-reversal asymmetry and/or residual delay maps (RDM) with trough-ridge asymmetry. The mixed structures of linear correlation and strong nonlinear features improve the understandings on the prediction and predictability of DTR variability.