A Greenland ice core perspective on the dating of the Late Bronze Age Santorini eruption

Abstract: The Santorini eruption is one of the greatest in historic time and an important time marker in the regional archaeology. It has been dated to 1627-1600 BC (2σ error) using radiocarbon dating (Friedrich et al. 2006) and to 1642±5 BC by identifying a volcanic fallout layer in the Greenland ice cores (Vinther et al. 2006). However, archaeologists have estimated the eruption almost a century later than the other datings. The disagreement between the datings has caused a debate about the reliability of the different dating methods, with archaeologists questioning the radiocarbon and ice core datings as they do not agree within 2σ errors. Muscheler (2009) found that the discrepancy can also be seen when comparing records of cosmogenic radionuclides (10Be and 14C) in the GRIP ice core and IntCal04 tree ring data. In this study a new 10Be dataset from the NGRIP ice core is presented. It has a resolution of 7 years and spans the period 3752-3244 cal yr BP (1803-1295 BC). The NGRIP 10Be record and the previously published 10Be GRIP record were compared to IntCal datasets to investigate the discrepancy between the ice core and tree ring chronologies. By modelling the 14C production rate based on atmospheric 14C records a comparison could be made to the 10Be flux which is assumed to represent the 10Be production rate. This showed a time shift of ~23 years between the records. The sensitivity of the results to changes in important model parameters were evaluated. Uncertainties in the carbon cycle model may influence the inferred time shift in the order of 0-6 years. The potential influences of climate and atmospheric processes on the 10Be deposition were studied using δ18O from the respective cores and GISP2 ice core ion data. The comparison to δ18O revealed a negative correlation when the common production signal was removed from the 10Be curves by subtracting the 14C production rate curve. The ion data, as proxies for atmospheric circulation changes, did not show any correlations to the 10Be record or the 10Be/14C difference. When including possible data uncertainties there is still a minimum discrepancy of ~10 years between the 10Be ice core and the 14C tree ring record. Due to lack of alternative explanations it is concluded that the ice core and/or the tree ring chronologies contain unaccounted errors in this range.