Home > Research > Publications & Outputs > Modelling directionality, seasonality, and loca...
View graph of relations

Modelling directionality, seasonality, and local time dependences in extreme geomagnetic field fluctuations

Research output: Contribution to conference - Without ISBN/ISSN Poster

Published
Publication date22/11/2019
Number of pages1
Pages1
<mark>Original language</mark>English
Event16th European Space Weather Week - Palais des Congrès, Liège, Belgium
Duration: 18/11/201922/11/2019
Conference number: 16
http://www.stce.be/esww2019

Conference

Conference16th European Space Weather Week
Abbreviated titleESWW
CountryBelgium
CityLiège
Period18/11/1922/11/19
Internet address

Abstract

In this poster we describe a statistical analysis of extreme temporal changes in the horizontal component of the geomagnetic field (dB/dt) – an important indicator of geomagnetically induced currents. Extreme value theory is applied to data from 125 magnetometers in the global SuperMAG archive – with an average of 28 years of measurements per site – to determine return levels (RL) of |dB/dt| expected over periods of 100 years or more. This is achieved by fitting generalized Pareto (GP) distributions to declustered measurements of |dB/dt| above a 99.97-percentile threshold. Since large fluctuations are driven by diverse magneto-ionospheric driving processes (substorm expansions, sudden commencements, Pc5 ULF waves, etc.), the occurrence rate and high percentiles of |dB/dt| vary with geomagnetic latitude, magnetic local time (MLT), season, and with the compass direction of the fluctuation, dB. The interplanetary magnetic field orientation also exerts a strong influence on these patterns of occurrence, which we present alongside examples of fitted smoothing functions (splines and limited-order spherical harmonics and polynomials). By adapting statistical methods developed for the directional analysis of extreme ocean wave heights, we show how |dB/dt| GP distributions and RLs are generated for data sets sectored by compass direction (or equivalently by season, or MLT sector) and then combined and compared with GP distributions (and RLs) that ignore directionality or seasonal/diurnal variation.