Impact of archeointensity data quality on geomagnetic field modeling

The accuracy and reliability of the intensity paleosecular variation curves and models, which provide the behavior of the past changes of the Earth's magnetic field strength, rely on the quality, the quantity and the spatiotemporal distribution of the data used for their construction. The archeomagnetic database is in continuously improving and, consequently, models and secular variations curves should be more accurately defined. However, the scatter of the archeointensity data is very puzzling and clearly suggests that some of the intensity data presented in the databases might not be reliable. In this work, we use the European and Western Asian archaeointensity data (the best covered region in terms of archeomagnetic data) to study how the quality of the different data can affect geomagnetic field model results. As the establishment of well-defined objective criteria to infer the different reliabilities of archeointensity data is not obvious, we use very basic palaeomagnetic criteria to rank the archeointensity data in different categories depending on the methodology used during the laboratory treatment of the samples and on the number of specimens retained to calculate the mean intensities. Four different intensity regional models were obtained using the different intensity quality datasets. Models tend to converge for the last 1500 year, but very different predictions are observed for the first millennium BC. Two clear maximum in archeointensity are observed around 600 and 800 AD, preceded by a minimum around 400-500 AD only well detected when high quality data are used. Some high frequency variations observed in the first millennium BC are clearly controlled by the low quality data (more numerous).