A Simple Stochastic Model for Long-term Climatic Variability

Thursday, 18 December 2014
Ioannis Markonis, National Technical University of Athens (NTUA), Marousi Athens, Greece and Demetris Koutsoyiannis, National Technical University of Athens (NTUA), Zographou Athens, Greece
In its fifth Assessment Report, IPCC (Hartmann et al. 2013, Climate Change 2013, Contribution of Working Group I to the Fifth Assessment Report) referred for the first time to the statistical implications of the natural long-term climatic variability. In our recent work (Markonis and Koutsoyiannis, Surveys in Geophysics, 2013) we showed that natural temperature variability exhibits Hurst-Kolmogorov (HK) behaviour (also known as long term persistence) for time scales spanning nine orders of magnitude with Hurst coefficient near 0.92. However, this is a negatively biased estimate, as no bias correction technique was applied; this implies that the true value should be higher. Here, we adapt the estimation to account for bias and provide a possible stochastic model for an important variable of the climatic system: temperature. We also compare the behaviour of this model with other models which are widely used in climatology, including AR and ARMA models.