Statistical aspects of global warming dynamics

Abstract EN

This communication aims to identify and utilize persistent features of global atmospheric temperature for predicting underlying fluctuating phenomena.

A graphical method, Loess plot is used to show the presence of some degree of non-linearity in global temperature dynamics.

Here we propose to split the average global temperature data set into three scale invariant temperature series on the basis of Hurst exponent in order to extract useful features of the physical system responsible for temperature anomalies.

Global warming dynamics is revealed as possible Martingale and Markov stochastic processes.

Different orders of Markov processes are examined for representative adequate models of the phenomenon.

The models considered can be used both for short term and long-term temperature forecasts.