Uncertain depth-age relationships in paleomagnetic records: Assessing data quality and geomagnetic content

We present a statistical model for two types of errors that cause uncertainties in the depth-age relationship for any stratigraphic core. In an application of this model, we assess the consequences that uncertain age assignments have on combining globally distributed relative pal\-eo\-in\-ten\-sit\-y records from sediment cores. Such records have been stacked in attempts to isolate the record of geomagnetic dipole intensity variations. First, the model addresses errors in the ages of the control points that establish a first order age-scale. These inferred ages are generally gathered from a complementary source, such as, in the case of paleomagnetism, radiometric methods and accepted oxygen isoptope age-scales. Second, we model errors in the ages of physical measurements due to interpolation of an assumed accumulation rate between control points. These inferred ages are in error due to the relatively short time-scale stochastic variations in the actual accumulation rates. Given appropriate parameters, our model provides estimates of uncertainties in inferred age at any depth for individual cores and allows simulations of plausible depth-age relationships. Isolation of geomagnetic dipole intensity variations is complicated by other uncertainties, particularly inaccuracies in the relative paleointensity record and non-dipole geomagnetic field contributions. We use simulations from the statistical age model and from detailed numerical dynamo simulations to explore a method of data quality control and to evaluate geomagnetic content in stacked paleomagnetic records. Both are accomplished by examining the spectral content of the paleomagnetic signal resulting from the simulations. Quality control tests using records from nearby cores show that introducing age errors acts to decorrelate geomagnetic signals, primarily at shorter periods, although it is possible for nearby records to be uncorrelated in most, if not all, period bands. Modest age errors can account for the poor correlation between neighboring paleomagnetic records often observed in real data. When we consider the global stack our results show that, with the largest errors, dipole variations with periods longer than about 20 kyr are present in the stacked record of dipole variations. Modest errors in the ages of control points can account for most of the spectral degradation, while errors due to variations in accumulation rate may have a similar effect. With adequate knowledge of age errors, this method offers the possibility of quantifying uncertainties due to other noise sources.