Abstract

We present a reanalysis of ground-measured geomagnetic field data at 5-minute resolution, spanning 11 years (a full solar cycle). Our aim is to resolve the dominant structures (i.e. modes of variability) in the Earth’s external magnetic fields, which are generated by interaction between the terrestrial and solar magnetic fields in near-Earth space. The external magnetic fields are the single largest source of error in efforts to model the Earth’s internal magnetic field – essential for navigation and resource exploration.

A major focus of our reanalysis is to avoid making assumptions about the nature of the external fields, and let the data ‘tell us’ how to best resolve them. Hence, we employ a (commonly-used) pattern-finding method called Empirical Orthogonal Functions (EOFs), also called Principal Component Analysis (PCA). EOF is typically used as a ‘breadth-first’ tool in initial assessments of the structure of a given dataset. Here we show that with appropriate processing, EOF analysis can be used to good effect in the completion and decomposition of a sparse set of magnetic field data, aiding subsequent interpretation and prediction.