Multi-parameter crossdating for sub-fossil and historical samples

Crossdating of sub-fossil or historical samples almost exclusively involves the use of a single tree-ring (TR) parameter such as ring-width (RW), Blue Intensity (BI) or stable isotopes. This paper details a new method that combines the information from multiple TR parameters – in this case specifically using RW, earlywood BI and latewood BI – to enhance our ability to identify a robust correct date of a single timber (i.e. a sub-fossil sample) or a set of internally crossdated timbers (i.e. an historical building phase chronology). The approach builds on standard approaches where sliding correlations are used to identify the position of maximal correlation to denote the correct temporal date. For the multi-parameter dating approach, the time-series of these sliding correlations are averaged to maximise the correct highest correlation temporal position, which should be common for all parameters, while minimising spurious correlations that individual parameter may express through time. The correlation time-series for the individual parameters, as well as their average (the COMBO variant), are then transformed to t-values with the degrees of freedom being adjusted lower by considering the autocorrelation structure of the Reference and Undated chronologies while the inter-correlation of the parameter chronologies is further used to adjust the COMBO degrees of freedom down. The multi-parameter COMBO method shows great promise for combining information from individual TR parameter series (single or multiple timbers) by increasing the chance of dating timbers that individually may express relatively weak signals, or for dating short-sequenced samples by increasing the effective degrees of freedom.