The comparison of statistical algorithms for constructing age-depth models of speleothems

초록

is very important to understand the philosophy and operation properties of statistical algorithms for constructing realistic age-depth models of speleothems. This study presents the characteristics of statistical algorithms designed for speleothems. We have selected previously published five Korean stalagmites with distinct characteristics of the changes in growth rates. Then, the age-depth models of these stalagmites were constructed using 3 statistical algorithms (StalAge, COPRA, MOD-AGE) which have been widely applied to the recent cave records. The age-depth models resulted from COPRA and MOD-AGE show the lowest deviations from median Th-230 ages when the stalagmite samples have relatively constant growth rates (the cases of BN-1 and SSN1). In contrast, for the stalagmites (DY-1, SGS1 and ED-1) including significant hiatuses with large time gaps, the most convinced model can be selected depending on both Th-230 ages and petrographic data. COPRA simulates age-depth models with good agreements to Th-230 ages, but it is sometimes impossible to construct final models when the dataset includes Th-230 ages with large errors and/or inversions in the stratigraphic order. StalAge could be preferred to simulate the age-depth relationships in the cases of the dataset with large errors or age inversions but this provides the largest deviations from Th-230 ages. The results of this study emphasize the needs of criteria to construct the most reliable age-depth model for speleothems using the statistical algorithms. Our results can be equally applied to various paleoclimatic archives (e.g., terrestrial sediments) that require similar age-depth models.

키워드