The GEOMAGIA50 database - Available Models
This section gives a general overview of the available global geomagnetic field models that can be calculated using the GEOMAGIA web site. Monika Korte and Cathy Constable provided the codes and coefficients for the calculations.
ARCH3k.1 and CALS3k.3

These two models were generated using the compilation of archeomagnetic and lake sediments data covering the past 3000 years as available in 2009. In particular, ARCH3K.1 uses archeomagnetic data only. The global distribution of this data type is strongly biased towards the northern hemisphere, and Europe in particular. The model gives reasonable field values for these regions, but should not be used for global studies or southern hemisphere field predictions.
The CALS3k.3 model was generated using both archeomagnetic and lake sediments data, therefore the geographic coverage of the data is better than for ARCH3K.3. It can be used for calculations of the field over the entire globe. Note that model estimates from regions not well covered by data might have uncertainties that are even larger than the estimates provided by the bootstrap method.
Both models are constrained to agree loosely with the historical gufm1 model by Jackson et al. (2000) for the recent 400 years. Note that it recently turned out that the applied constraint is too weak to overcome a problem with some North American lake sediment data in CALS3k.3 and the use of CALS3k.4 is now recommended. For more information please refer to this article.

This model was generated in 2011 using a combination of the available archeomagnetic and lake sediments data covering the past 3000 years (Korte and Constable, 2011). Compared to CALS3k.3, 163 additional archeomagnetic data are included and 13 newly published sediment records have been included, which cause differences between the two models mainly in the South-East Asian, Alaskan and Siberian regions. The model is constrained to agrees very closely to gufm1 (Jackson et al., 2000) for the past 4 centuries.

This was one of the first models published (Korte and Constable, 2005), generated using archeomagnetic full vector data and lake sediments directional data. It covers the past 7 millenia. Due to the fact that no sedimentary intensity data were used and archeomagnetic intensity data are sparse before around 1000 BC the dipole moment seems underestimated somewhat in this model. The CALS7K.2 model is superseded by the recent CALS10k.1b model, which should be used instead.

This model is the latest available and covers the past 10000 years. It supersedes CALS7K.2. It is based on both lake sediments and archeological artifacts data available up to 2011, but is strongly controlled by sediment data. The model is an average obtained from bootstrap sampling to take account of uncertainties in magnetic components and ages in the data. Consequently, it is strongly smoothed compared to the CALS3k.3 and CALS3k.4 models, in order to provide a robust reconstruction for the earlier epochs. The model is described in the following paper.