Climatological Statistics of Extreme Geomagnetic Fluctuations With Periods From 1 s to 60 min

Abstract

Using a global database of 125 magnetometers covering several decades, we present occurrence statistics for fluctuations of the horizontal geomagnetic field (dBh/dt) exceeding the 99.97th percentile (P99.97) for both ramp changes (Rn) and the root‐mean‐square (Sn) of fluctuations over periods, τ, from 1 to 60 min and describe their variation with geomagnetic latitude and magnetic local time (MLT). Rates of exceedance are explained by reference to the magneto‐ionospheric processes dominant in different latitude and MLT sectors, including ULF waves, interplanetary shocks, auroral substorm currents, and traveling convection vortices. By fitting generalized Pareto tail distributions above P99.97, we predict return levels (RLs) for Rn and Sn over return periods of between 5 and 500 years. P99.97 and RLs increase monotonically with frequency (1/τ) (with a few exceptions at auroral latitudes) and this is well modeled by quadratic functions whose coefficients vary smoothly with latitude. For UK magnetometers providing 1‐s cadence measurements, the analysis is extended to cover periods from 1 to 60 s and empirical Magnetotelluric Transfer functions are used to predict percentiles and return levels of the geoelectric field over a wide frequency range (2 × 10−4 to 4 × 10−2 Hz) assuming a sinusoidal field fluctuation. These results help identify the principal causes of field fluctuations leading to extreme geomagnetically induced currents (GIC) in ground infrastructure over a range of timescales and they inform the choice of frequency dependence to use with dBh/dt as a GIC proxy.

An Extreme Auroral Electrojet Spike During 2023 April 24th Storm

“…In fact, the time derivative is comparable to the predicted once‐in‐10‐year value by Rogers et al. (2020, 2021). The amplitude and time derivative were based on 1‐min sampled data because 1‐s data are unfortunately not available at this station (generally speaking, 1‐s data would show larger time derivatives).…”

Distribution and Occurrence Frequency of dB/dt Spikes During Magnetic Storms 1980–2020

“…In future studies, we plan to use magnetic data with a higher time resolution (10 s or even 1 s data), to detect any spike events, which in our case may have fallen into two adjacent 1 min bins. Such higher resolution could also address more spectral details in the dB/dt variability, which have a strong impact on GIC occurrences as already pointed out by several studies (Grawe et al, 2018;Heyns et al, 2021;Marshall et al, 2010;Pulkkinen & Kataoka, 2006;Pulkkinen et al, 2008;Rogers et al, 2021). Both in the US and in Europe (through the European Space Agency that funded the MAG-SWE-DAN magnetometer project) such improved observational networks are being developed and their data will soon become available within the SuperMAG services.…”

“…Any GIC caused by a dB/dt spike depend on its intensity, direction, spectral characteristics, the nature of the underlying three-dimensional conductivity in the ground and the geometry of any electric power network above the ground. Several studies recommend a frequency-weighted dB/dt in order to have a more precise proxy of GICs (e.g., Grawe et al, 2018;Heyns et al, 2021;Marshall et al, 2010;Pulkkinen et al, 2008;Rogers et al, 2021). The amplitude and phase of the geomagnetic spectrum has been shown to be important when determining the potential difference in oil pipeline (Marshall et al, 2010).…”

Determining ULF Wave Contributions to Geomagnetically Induced Currents: The Important Role of Sampling Rate

“…Rogers et al. (2021) found for three locations in the United Kingdom that B fluctuations with roughly 1,200 s period ( f ≪ f Nyquist60 ) produced the most intense E when considering events occurring several times per year, whereas the 1‐in‐100 years return levels were greatest for 30–120 s period fluctuations ( f ≥ f Nyquist60 ). Trichtchenko (2021) found that a 0.1 Hz sampling rate is generally needed to understand power system responses and posed the question, “What should be the sampling rate (or Nyquist frequency) in order to provide an adequate representation of the fast geomagnetic variations (such as storm sudden commencement (SSC), substorms, pulsations, and rapid variations during the main phase of the storm) for use in the calculations of the extreme geoelectric field values and in the GIC modeling?” Motivated by this question, the present study will focus on the recommended sampling rate for one particular source of B listed by Trichtchenko (2021), pulsations, considering several factors: expected pulsation frequencies, the amplitude of B / E related to pulsations at different frequencies, and the ability of pulsations to cause potentially hazardous GIC.…”

“…Some studies found that sampling intervals of 60 s can adequately capture B/E/GIC variations (e.g., A. , while others (e.g., Grawe & Makela, 2021;Grawe et al, 2018;Trichtchenko, 2021) found that 60 s measurements are not adequate due to the presence of variations with frequencies (f) higher than the Nyquist frequency (f Nyquist60 = 0.0083 Hz for measurements sampled every 60 s). Rogers et al (2021) found for three locations in the United Kingdom that B fluctuations with roughly 1,200 s period (f ≪ f Nyquist60 ) produced the most intense E when considering events occurring several times per year, whereas the 1-in-100 years return levels were greatest for 30-120 s period fluctuations (f ≥ f Nyquist60 ). Trichtchenko (2021) found that a 0.1 Hz sampling rate is generally needed to understand power system responses and posed the question, "What should be the sampling rate (or Nyquist frequency) in order to provide an adequate representation of the fast geomagnetic variations (such as storm sudden commencement (SSC), substorms, pulsations, and rapid variations during the main phase of the storm) for use in the calculations of the extreme geoelectric field values and in the GIC modeling?"…”