Sediment and archeomagnetic data spanning the Holocene enable us to reconstruct the evolution of the geomagnetic field on time scales of centuries to millennia. In global field modeling the reliability of data is taken into account by weighting according to uncertainty estimates. Uncertainties in sediment magnetic records arise from (1) imperfections in the paleomagnetic recording processes, (2) coring and (sub) sampling methods, (3) adopted averaging procedures, and (4) uncertainties in the age-depth models. We take a step toward improved uncertainty estimates by performing a comprehensive statistical analysis of the available global database of Holocene magnetic records. Smoothing spline models that capture the robust aspects of individual records are derived. This involves a cross-validation approach, based on an absolute deviation measure of misfit, to determine the smoothing parameter for each spline model, together with the use of a minimum smoothing time derived from the sedimentation rate and assumed lock-in depth. Departures from the spline models provide information concerning the random variability in each record. Temporal resolution analysis reveals that 50% of the records have smoothing times between 80 and 250 years. We also perform comparisons among the sediment magnetic records and archeomagnetic data, as well as with predictions from the global historical and archeomagnetic field models. Combining these approaches, we arrive at individual uncertainty estimates for each sediment record. These range from 2.5 degrees to 11.2 degrees (median: 5.9 degrees; interquartile range: 5.4 degrees to 7.2 degrees) for inclination, 4.1 degrees to 46.9 degrees (median: 13.4 degrees; interquartile range: 11.4 degrees to 18.9 degrees) for relative declination, and 0.59 to 1.32 (median: 0.93; interquartile range: 0.86 to 1.01) for standardized relative paleointensity. These values suggest that uncertainties may have been underestimated in previous studies. No compelling evidence for systematic inclination shallowing is obtained from the analysis of the available database of Holocene sediment magnetic records. The analysis highlights the importance of collecting oriented cores, publishing and archiving unprocessed raw paleosecular variation determinations, and presenting a detailed chronology so that changes in the sedimentation rate can be assessed. With regard to future field models, workers should consider rejection of anomalous cores through comparisons to other sources and ensure that realistically large uncertainties are allocated to high-latitude declination records.
Panovska, S., Finlay, C., Donadini, F., & Hirt, A. M. (2012). Spline analysis of Holocene sediment magnetic records: Uncertainty estimates for field modeling. Journal of Geophysical Research, 117, B02101. https://doi.org/10.1029/2011JB008813