The Archaeomagnetic Field and the Historical Field: Never the Twain Shall Meet?
We compare predictions of the geomagnetic field in Europe from 1590 to 1800 AD from a field model from archaeomagnetic data in Europe, developed for archaeomagnetic dating and from the historical field model gufm1 (Jackson et al., 2000). A consistent discrepancy pre-1800 AD is observed in inclination (I), with gufm1 producing higher values of I than those predicted from archaeomagnetic data. Between 1590 and 1800 AD there are 354 archaeomagnetic data and 133 historical data; in general, the archaeomagnetic data are in Eastern Europe and the historical data in Western Europe (and the Atlantic). Thus if the two data sets were combined, they would provide much better spatial resolution; however, the apparent discrepancy between the datasets makes simple simultaneous modelling problematic. Is this offset to higher I values an indication of a systematically lower value of I measured in archaeomagnetic materials? Alternatively is this discrepancy caused because of inadequate error estimates? When long time series of historical data at single locations, e.g. London and Paris are plotted, it can be clearly seen that the scatter in I is substantially higher pre-1800 AD than post-1800 AD, but the quality is still likely to be higher than from archaeomagnetic measurements. We are currently testing improvements to the handling of errors in our models by considering correlation of temporal errors and also using the gradient of the current model iteration to convert temporal errors into measurement errors. We present comparisons of models produced using these different error calculations and discuss their effect on the discrepancy between archaeomagnetic models and gufm1.