# How do core surface flow models vary when inverted from IGRF-14 candidate field models?

## TL;DR

This paper examines how different candidate models for Earth's magnetic field affect predictions of core surface flows, finding that higher model resolution improves accuracy.

## Contribution

The study introduces a new analysis of core flow variability derived from IGRF-14 candidate models using fixed inversion parameters.

## Key findings

- Candidate models deviating from the median IGRF model show the largest flow differences.
- Flow discrepancies are most significant at small spatial scales and in the Pacific and polar regions.
- Increasing spherical harmonic degree truncation improves flow predictions and increases maximum flow speed by up to 31%.

## Abstract

The International Geomagnetic Reference Field (IGRF) model is a series of models that describe the large scale magnetic field measured at the Earth’s surface. It is used by a wide range of scientists and industries, including in navigation, space weather applications, and resource exploration. The 14th generation, IGRF-14, is the result of an international collaboration over 19 different lead research groups. This new generation provides a definitive model of the main magnetic field (MF) for the epoch 2020.0, a prediction of the MF for 2025.0, and a predicted average annual time variation of the magnetic field for 2025.0\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$-$$\end{document}-2030.0. The first time derivative of the magnetic field, known as secular variation, (SV) is linked to the flow at the top of Earth’s outer core. As such, the ensemble of IGRF-14 candidate models can be used to investigate predicted core flow variability between 2020.0 and 2030.0. We use the pygeodyn Python package using the AR-1 ‘dense’ methodology, the 71% geodynamo prior, and keep all inversion parameters fixed to ensure that differences in inferred flow arise solely from variations from the IGRF candidate models. Our results demonstrate that while all candidates produce broadly similar flow spectra for all degrees, candidates that deviate most from the median IGRF model also show the largest differences in flow. In all cases, the flow speed difference remains below 25% of the maximum flow speed of the Huber-weighted mean model in space, with most discrepancies occurring at small spatial scales. The greatest flow uncertainty appears over the Pacific and in the polar regions, where constraints are weaker. Finally, we assess the impact of the maximum spherical harmonic degree truncation of the SV candidate models by comparing two SV candidates truncated at spherical harmonic degree 13 instead of 8. Increasing the truncation degree of SV alters the flow spectral energy at all degrees and increases the maximum flow speed by up to 31%, despite the flow maps remaining highly correlated. This study supports the idea of raising the spherical harmonic truncation degree to 13 for the SV prediction component of the IGRF.