AGE ESTIMATES FOR THE LATE QUATERNARY HIGH SEA-STANDS

A database of more than 300 published alpha-counted uranium-series ages has been compiled for coral reef terraces formed by Late Pleistocene high sea-stands. The database was screened to eliminate unreliable age estimates (T-230/Th-232 Th < 20, calcite > 5%) and those without quoted errors, and a distributed error frequency curve was produced. This curve can be considered as a finite mixture model comprising k component normal distributions each with a weighting alpha. By using an expectation maximising algorithm, the mean and standard deviation of the component distributions, each corresponding to a high sea level event, were estimated. Eight high sea-stands with mean and standard deviations of 129.0 +/- 33.0, 123.0 +/- 13.0, 102.5 +/- 2.0, 81.5 +/- 5.0, 61.5 +/- 6.0, 50.0 +/- 1.0, 40.5 +/- 5.0 and 33.0 +/- 2.5 ka were resolved. The standard deviations are generally larger than the values quoted for individual age estimates. Whilst this may be due to diagenetic effects, especially for the older corals, it is argued that in many cases geological evidence clearly indicates that the high stands are multiple events, often not resolvable at sites with low rates of uplift. The uranium-series dated coral-reef terrace chronology shows good agreement with independent chronologies derived for Antarctic ice cores, although the resolution for the latter is better. Agreement with orbitally-tuned deep-sea core records is also good, but it is argued that Isotope Stage 5e is not a single event, as recorded in the cores, but a multiple event spanning some 12 ka. The much earlier age for Isotope Stage 5e given by Winograd et al. (1988) is not supported by the coral reef data, but further mass-spectrometric uranium-series dating is needed to permit better chronological resolution.