Articles | Volume 7, issue 2
Earth Syst. Dynam., 7, 419–439, 2016
Earth Syst. Dynam., 7, 419–439, 2016

Research article 28 Apr 2016

Research article | 28 Apr 2016

Are there multiple scaling regimes in Holocene temperature records?

Tine Nilsen, Kristoffer Rypdal, and Hege-Beate Fredriksen Tine Nilsen et al.
  • Department of Mathematics and Statistics, University of Tromsø, The Arctic University of Norway, Tromsø, Norway

Abstract. The concept of multiple scaling regimes in temperature time series is examined, with emphasis on the question whether or not a monoscaling model with one single scaling regime can be rejected from observation data from the Holocene. A model for internal variability with only one regime is simpler and allows more certain predictions on timescales of centuries when combined with existing knowledge of radiative forcing. Our analysis of spectra from stable isotope ratios from Greenland and Antarctica ice cores shows that a scale break around centennial timescales is evident for the last glacial period, but not for the Holocene. Spectra from a number of late Holocene multiproxy temperature reconstructions, and one from the entire Holocene, have also been analysed, without identifying a significant scale break. Our results indicate that a single-regime scaling climate noise, with some non-scaling fluctuations on a millennial timescale superposed, cannot be rejected as a null model for the Holocene climate. The scale break observed from the glacial time ice-core records is likely caused by the influence of Dansgaard–Oeschger events and teleconnections to the Southern Hemisphere on centennial timescales. From our analysis we conclude that the two-regime model is not sufficiently justified for the Holocene to be used for temperature prediction on centennial timescales.

Short summary
In this article it is discussed how temperature variability on centennial timescales and longer can be described in a simplistic way. By analysing the scaling in late Holocene temperature reconstructions and longer temperature records from Greenland and Antarctic ice cores, we find that the choice of model depends heavily on the data material and timescale one chooses to emphasize. Ignoring data beyond the Holocene seems plausible when predicting temperature, but not for other purposes.
Final-revised paper