Automatic detection of ionospheric Alfvén Resonances using signal and image processing techniques

Beggan, Ciaran D.; Kelly, Gemma S.. 2014 Automatic detection of ionospheric Alfvén Resonances using signal and image processing techniques. [Poster] In: Hot Spring MIST, Bath, UK, 23-25 April 2014. (Unpublished)

Before downloading, please read NORA policies.

Preview

Text
MIST_Beggan_IAR_AutomaticDetection_poster.pdf
Download (6MB) | Preview

Abstract/Summary

Ionospheric Alfvén Resonances (IAR) are a type of ELF wave phenomenon in the 0.5-10 Hz frequency range and are detectable in magnetic field variations as measured by induction coil magnetometers. They are believed to be excited by electric fields from lightning strikes leaking from the troposphere into the ionosphere, causing the oscillation of magnetic field lines through the upper atmosphere. In September 2012, the British Geological Survey installed two orthogonal coils in Eskdalemuir in the Scottish Borders. The coils reliably record variations of the magnetic field in the frequency band from 0.1 - 100 Hz in the north-south and east-west components. IAR appear in the data almost on a daily basis. Investigation of the IAR events in the induction coil data show that they appear as a series of fringes in spectrograms (i.e. plots of power at time versus frequency) between 0.5 and 7 Hz, occasionally extending beyond the first Schumann resonance to 10 Hz. They typically appear after the onset of local night time, expanding in frequency width until midnight and fading a few hours before local dawn. There are usually between five and fifteen fringes. Their appearance varies with season – occurring most often in northern hemisphere winter. Given the complexity of the analysis and the volume of data, a new protocol for detecting and classifying key parameters was developed using image and signal processing techniques. In the first instance, individual spectra for time-series of 100 seconds are analysed for peak power in the frequencies between 0.5 and 10 Hz. Hence the frequency for each IAR is determined (i.e. the signal processing part). This generates an image of ‘spot’ frequencies for each IAR for each day. Next, a series of image processing filters are applied to the ‘spot’ images to connect the IAR occurrences together in a coherent manner. No manual intervention is required. The method picks out the IAR across an entire day of data, allowing statistics related to frequency (f), peak-to-peak frequency (∆f) and number of peaks etc. to be automatically determined from the entire dataset. We show the details and examples of the new method and the results from the first eighteen months of data.

Item Type:	Publication - Conference Item (Poster)
NORA Subject Terms:	Space Sciences Atmospheric Sciences
Date made live:	23 May 2014 12:25 +0 (UTC)
URI:	https://nora.nerc.ac.uk/id/eprint/507309

Item Type:

Publication - Conference Item (Poster)

NORA Subject Terms:

Space Sciences
Atmospheric Sciences

Date made live:

23 May 2014 12:25 +0 (UTC)

URI:

https://nora.nerc.ac.uk/id/eprint/507309