The BGS candidate models for IGRF-13 with a retrospective analysis of IGRF-12 secular variation forecasts

Abstract

The three candidate models submitted by the British Geological Survey for the 13th generation International Geomagnetic Reference Field are described. These DGRF and IGRF models are derived from vector and scalar magnetic field data from the European Space Agency Swarm satellites and ground observatories, covering the period 2013.9 to 2019.7. The internal field model has time dependence for degrees 1 to 15, represented by order 6 B-splines with knots at six monthly intervals. We also solve for a degree 1 external field time dependence describing annual and semi-annual signals with additional dependence on a bespoke Vector Magnetic Disturbance index. Satellite data are weighted by spatial density, along-track standard deviations, and a larger-scale noise estimator defined in terms of a measure of Local Area Vector Activity at the geographically closest magnetic observatories to the sampled datum. Forecasting of the magnetic field secular variation for 2020–2025 is by advection of the main field using steady core surface flows with steady acceleration applied. We also investigate the performance of the previous generation of candidate secular variation models, for IGRF-12, analysing the agreement of the candidates between 2015 and 2020 with the retrospective IGRF-13. We find that there is no clear distinction between the performance of mathematically and physically extrapolated forecasts in the period 2015–2020. We confirm that the methodology for the BGS IGRF-12 predictions performed well, despite observed secular accelerations that are highlighted by our analysis, and thus justify the methodology used for our IGRF-13 SV candidate.

Evaluation of candidate models for the 13th generation International Geomagnetic Reference Field

“…For the main field at epochs 2015.0 and 2020.0, some teams derived their candidate models from parent models describing the magnetic field over multi-decadal periods by combining datasets from several satellite missions and ground observatories Huder et al 2020;Ropp et al 2020;Sabaka et al 2020;Wardinski et al 2020). Several teams built parent models covering the full Swarm satellite mission era (November 2013 to present) (Brown et al 2020;Rother et al 2020;Vigneron et al 2020). Other teams built dedicated main field candidate models for each of the epochs requested by the call, thus using data within smaller time windows centered on 2015.0, or immediately preceding 2020.0 (Pavón-Carrasco et al 2020;Yang et al 2020;Alken et al 2020a;Petrov and Bondar 2020), which required less complex parameterization in time.…”

“…Other teams built dedicated main field candidate models for each of the epochs requested by the call, thus using data within smaller time windows centered on 2015.0, or immediately preceding 2020.0 (Pavón-Carrasco et al 2020;Yang et al 2020;Alken et al 2020a;Petrov and Bondar 2020), which required less complex parameterization in time. Some teams co-estimated a low-degree external field model representing magnetospheric sources (Brown et al 2020;Yang et al 2020;Pavón-Carrasco et al 2020;Rother et al 2020;Ropp et al 2020;Huder et al 2020;Sabaka et al 2020;Vigneron et al 2020). Two teams co-estimated additional geomagnetic source fields in their parent models (Ropp et al 2020;Sabaka et al 2020).…”

“…For the predictive secular variation models covering 2020.0 to 2025.0, the candidates broadly fall into two main categories: (1) computing SV solely from the latest available satellite and ground data (Alken et al 2020a;Pavón-Carrasco et al 2020;Huder et al 2020;Petrov and Bondar 2020;Rother et al 2020), or (2) applying physics-based modeling, combined with recent satellite and ground data to forecast future field changes based on underlying core dynamics (Brown et al 2020;Fournier et al 2020;Minami et al 2020;Metman et al 2020;Sanchez et al 2020;Tangborn et al 2020 et al 2020). Accurately forecasting the temporal evolution of the main geomagnetic field is nontrivial, due to challenges such as the low resolution of the recoverable magnetic field at the core-mantle boundary, the occurrence of unpredictable geomagnetic jerks or the uncertainty associated with diffusion of the field over short timescales (e.g.…”

Physics-based secular variation candidate models for the IGRF

“…But before congratulating ourselves further, let us first recall the mixed conclusions drawn from the hindcasting experiments described above, and second note that the global analysis reveals that all forecasts missed the geomagnetic jerk that started early in 2014, following a pulse of secular acceleration that had peaked at the core surface in 2012-2013 (Torta et al 2015;Finlay et al 2016;Kotzé 2017;Soloviev et al 2017). The findings of Alken et al (2021b) were corroborated by Brown et al (2021) who performed a similar analysis of the same period; their conclusion was that physical predictions did not do better or worse than mathematical extrapolations over the 5-year time span of the IGRF-12 release. How can we improve on this state of affairs?…”

“…The two families of physical models (based on core surface flows or on dynamos) are themselves uncertain; in an attempt to take, at least partly, this uncertainty into account, most groups now resort to ensemble approaches (Minami et al 2020;Sanchez et al 2020;Brown et al 2021;Fournier et al 2021;Tangborn et al 2021) for steps (2) and (3) in Fig. 4.…”