Using geomagnetic secular variation to separate remanent and induced sources of the crustal magnetic field

Magnetic fields originating from magnetized crustal rocks dominate the geomagnetic spectrum at wavelengths of 0.1-100 km. It is not known whether the magnetization is predominantly induced or remanent, and static surveys cannot discriminate between the two. Long-running magnetic observatories offer a chance, in principle, of separating the two sources because secular variation leads to a change in the main inducing field, which in turn causes a change in the induced part of the short-wavelength crustal field. We first argue that the induced crustal field, b(I)(t), is linearly related to the local core field, B(t), through a symmetric, trace-free matrix A: b(I)(t)=AB(t). We then subtract a core field model from the observatory annual means and invert the residuals for three components of the remanent field, b(R)(t), and the five independent elements of A. Applying the method to 20 European observatories, all of which have recorded for more than 50 years, shows that the most difficult task is to distinguish b(R) from the steady part of b(I). However, for nine observatories a time-dependent induced field fits the data better than a steady remanent field at the 99 per cent confidence level, suggesting the presence of a significant induced component to the magnetization.