Abstract: In the period range from 1 to 600 seconds, geomagnetic pulsations are one major source of induction of
electric currents in the earth. Geomagnetic pulsations are, in particular, indicators of physical processes
in the magnetosphere or act as induction source for magnetotellurics.
Geomagnetic pulsations are sorted in two main classes.
Continuous pulsations (Pc) are quasi-periodic oscillations with amplitude about the nT. The amplitude has
diurnal, seasonal and even longer periods variations. Pc pulsations are separated in five main classes. For
example, Pc3 amplitude is maximized during daytime and minimal during night-time.
Irregular pulsations (Pi) are short signals containing few oscillations compared to Pc pulsations. They are
also subdivided into two main classes with different behaviours.
We have developed an automatic detection scheme of geomagnetic pulsations in 1Hz-magnetic observatory
data, based on continuous wavelet transform (CWT) and Fourier surrogates. CWT is used to detect, on a
time-frequency plane, where geomagnetic pulsations are characterized by wavelet coefficients with
significant amplitude. Two criteria are set: (i) a minimum amplitude of the cross-wavelet spectrogram,
and (ii) a minimum area size of thresholded coefficients. Threshold in amplitude is obtained using
Fourier surrogates (obtained through the computation of synthetic magnetic time series based on the
computation of an average hourly power spectrum over the analysed day). The population of surrogates
is then compared to the actual analysed time series to obtain significant wavelet coefficients. The area
criterion is computed from the size of the wavelet kernel above a predefined value. From the final set of
wavelet coefficients, we are also able to compute horizontal polarisations attributes.
We show that this approach allows to automatically detect and characterize geomagnetic pulsations activity
on geomagnetic observatory time series. The methodology has been implemented on a computing
cluster to allow for fast detection and characterization of geomagnetic pulsations. Results of this algorithm
have been compared to the literature about geomagnetic pulsations to assess for efficiency. We aim at
implementing this procedure to make available geomagnetic pulsations characteristics in an automatic
way and on a planetary scale.
Programme: 139
