Scaling Properties of Climate Variability as Reconstructed from Different Paleo-Indicators : an Analysis of Holocene Time Series Based on Haar Fluctuations

Thursday, 18 December 2014
Raphaël Hébert1, Shaun Lovejoy1 and Anne de Vernal2, (1)McGill University, Montreal, QC, Canada, (2)University of Quebec at Montreal UQAM, Montreal, QC, Canada
Holocene climate changes in Arctic and subarctic areas were investigated using

time series resulting from the analyses of different biological indicators in marine

(dinocysts, foraminifers, alkenones) and lake (pollen, chironomids, diatoms)

sediment cores (see compilation by Sundqvist et al. Climate of the Past, 2014).

These indicators were used to reconstruct climate-related parameters such as

temperatures, precipitation, salinity and/or sea-ice cover. Dinocyst, foraminifer

and pollen series were usually calibrated with the Modern Analogue Technique

(MAT) whereas those from chironomids and diatoms mainly used the Weighted Average Partial

Least Square (WAPLS) calibration approach. Hence, the available times series used to document long

term climate changes are heterogeneous since they are based on different indicators

and different reconstruction methods.

In order to document the climate variability captured by the time series, we have

analyzed the scaling behavior of Holocene records (last 12,000 years) by the mean

of the Haar fluctuations (for a given time interval, these are simply the differences

between the means of the first and second halves of the interval). Over various

ranges of time scale, they can be characterized by exponents H (e.g. [Lovejoy and

Schertzer, 2012]). When H<0, fluctuations tend to cancel, the series appears

stable”, averaging fluctuations over longer and longer intervals typically yields

smaller values. When H>0 on the contrary, fluctuations tend to grow with larger time

intervals as the series “wanders” like a drunkard’s walk, i.e. it appears “unstable”. We

show how to perform the analyses in a robust fashion, avoiding biases due to

irregular sampling and/or variable temporal resolutions. On this basis, we

were able to evaluate whether the signals of different paleo-indicators for the

same location converge at low frequency, or whether they diverge. Even when

considering the same core, different indicators sometimes yield diverging signals. This

suggests variations which could be due to methods of reconstruction or to the

climatic signal recorded by the indicators. Despite divergences in some of the regional

records, results consistently show that signals from marine cores tend to be

more stable than those from continental environments.