Scale-dependent Reconstruction of An Winter Index For The North Atlantic Oscillation
Abstract
The European winter climate is under considerable influence of the North Atlantic Oszillation (Hurrell and van Loon, 1997) . For example dryer winters in Spain and Portugal and wet and mild winters in Scandinavia coincide with a high pressure gradi- ent between Iceland and the Azores. Furthermore, the climatic influence of the NAO is well known for the North American East Coast and Greenland. This dominant role of the NAO has attracted much interest among climatologists.
The predominant strength of the north-south pressure gradient during the 1990ies has led to speculation about the future development of the NAO. Model results indicate an intensification and eastward shift of the low pressure center for greenhouse gas warming scenarios (Ulbrich et al., 1999). However, evaluation of the past climate is indispensable for understanding the dynam- ics of the NAO and its evolution in coming times. Various paleoclimate studies have exploited different sources of proxy records aiming for reconstructions of an win- ter NAO index. Comparisons of those proxy-based reconstructions revealed that they can be significantly different (Schmutz et al, 2000). Recently, new reconstructions based on multi-proxy records promise that the uncertainties can be further minimized (Cullen et al., 2000; Cook et al., 2001). The purpose of our study is to improve the quality of reconstructed NAO indices by introducing a new method of extracting the climate information from proxies. Tra- ditionally, reconstructions use unfiltered, annually resolved (or even more detailed) proxy records for the calibration and verification. Then it follows an analysis of the variability of the reconstructed index on different time scales. Instead, we decompose the time series by means of discrete wavelet transformation before the calibration step. The various wavelet scales are then used for the calibration of transfer models. The transfer models are derived by means of linear regression techniques. During the verification step a special treatment is neccessary for statistical testing. A proper Monte Carlo test is used to estimate the statistical significance of the regression results. The method is applied to Greenland ice core proxies and European tree ring chronolo- gies. Our results for the winter NAO index show that slightly better reconstructions 1 can be achieved by this method if one is interested in the temporal variations on dis- tinct scales. However, comparisons with other reconstructions emphasise the fact that an unequivocal reconstruction of the NAO's history remains a demanding task in the field of paleoclimate research. 2- Publication:
-
EGS General Assembly Conference Abstracts
- Pub Date:
- 2002
- Bibcode:
- 2002EGSGA..27.4190T