Modelling of attitude error in vector magnetic data: application to Ørsted data

Analysis of data from the Ørsted satellite indicates that the largest source of error arises from an anisotropic attitude uncertainty, related to poorly determined rotation about the axis of the star imager. For two data sets from May and December 1999, I quantify this error, and review and apply a formalism designed to allow for this problem. I argue that, when modelled correctly, this attitude uncertainty should not significantly damage the main field models obtained from Ørsted data.