Exploring an ensemble approach to estimating skill in multiproxy palaeoclimate reconstructions

This paper describes an approach that gives an estimate of the reconstructive skill of a proxy-based palaeoclimatological reconstruction. Uncertainties in proxy data are likely to be reflected in variations in reconstructive skill of the proxy-based climatic reconstructions. The method is based on making an ensemble of reconstructions, providing a probability distribution of each reconstruction estimate. The relative breadth of the distribution for a particular reconstructed value should give an indication of the reconstructive skill. The ensemble reconstruction approach draws on the ensemble prediction system as used for operational weather forecasting. The ensemble reconstruction approach is tested using a recent multiproxy reconstruction of the winter North Atlantic Oscillation (NAO) index. It is shown that the ensemble approach provides a representation of the degree of certainty associated with the reconstructed NAO-index values.