Long-term memory dynamics of continental and oceanic monthly temperatures in the recent 125 years

For both Northern and Southern hemispheres, the long-term memory dynamics for continent and ocean temperature records in the recent 125 years is studied in this paper. It is found that the records exhibit long-range memory and multifractality characteristics where large temperature anomalies display a more random behavior than the overall time series. A 256-month moving window was used to compute the time evolution of the fractal scaling exponent, giving the following results: (i) Ocean temperatures are more persistent than land temperatures, a result already reported in recent publications, (ii) All records show multifractality features, reflecting the nonlinear behavior of the temperature dynamics. Continent temperatures present sharper multifractal spectra than ocean temperatures, (iii) The persistency, as revealed by the scaling exponent, for ocean temperatures displays a cyclic behavior around a nearly constant average value of 22 years, (iv) The persistency for the Northern Hemisphere land temperature is also cyclical but with an increasing trend, and (v) The time at which the Northern Hemisphere continent temperature persistency will converge into the Northern Hemisphere ocean behavior was estimated with linear and exponential extrapolation functions, showing hitting dates around 2050 +/- 20 A.D. Potential implications of these results concerning the nature of climate change are also discussed. (C) 2008 Elsevier B.V. All rights reserved.