Perpendicular error effect in the DGRF model proposals

Abstract

It is known that when spherical harmonic models of the geomagnetic field are derived from data sets having insufficient component (vector) data these models are liable to pseudo-systematic errors, particularly large for the sectorial coefficients. These coefficient errors correspond to error fields having foci on and near the dip equator, and which are everywhere roughly perpendicular to the main (approximately dipole) field—the perpendicular error effect. This paper demonstrates the existence of such an effect, with foci of up to 300 nT, in the differences between the IGS, NASA, and USGS spherical harmonic models proposed for the DGRF 1965–1975, and deduces that it probably comes almost entirely from the NASA models. It is shown that, although the NASA data set contained a large amount of observatory component data, the method of analysis used by NASA meant that this data was given little weight in the main field model. It appears that ∼ 15% of component data is needed to avoid significant perpendicular error effect.