Assessing the reliability of Holocene relative palaeointensity estimates: a case study from Swedish varved lake sediments

SUMMARY We present palaeomagnetic secular variation and relative palaeointensity estimates from multiple sediment cores recovered from a varved lake sediment sequence in Kälksjön, a lake in central west Sweden. Alternating field demagnetization of discrete samples reveals natural remanent magnetizations dominated by a stable, single component. Magnetic grain size indicators—which include Day plots and first‐order reversal curves—indicate that the main carrier of the natural remanence is single‐domain magnetite throughout the varved sequence, which extends to ∼9200 cal. yr BP. The millennial‐scale features of the Kälksjön relative palaeointensity, inclination and declination data are well matched against the Fennoscandian master curves FENNORPIS and FENNOSTACK, and a global dipole moment reconstruction curve. There is evidence, however, of environmental bias in the relative palaeointensity estimates during specific time intervals. We find that bias is particularly pronounced during the early Holocene, specifically between ∼9000 and 7800 cal. yr BP. During this period the bias is apparent as a series of relative palaeointensity cycles, with a wavelength of ∼300 years. Some bias is also evident between 7800 and 3200 cal. yr BP. By comparing Kälksjön's smoothed relative palaeointensity data with the reconstructed geomagnetic field intensity from 14 C on timescales longer than 500 years—thereby removing the effect of relatively high frequency environmental bias in the early Holocene—we suggest that a distinct peak in intensity at centred on ∼8700 cal. yr BP is a true feature of the geomagnetic field. A minimum at ∼7400 cal. yr BP and a maximum at ∼2500 cal. yr BP are also robust features. Our main conclusion is that sedimentary‐based relative palaeointensity reconstructions must be carefully assessed if short‐term, centennial variations are to be examined.