Machine Learning Methods Applied to Riometer Data Classification
Date
2021-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Humankind’s growing reliance on advanced technologies reveals potential vulnerability to space weather events caused by the Sun’s effects on Earth. Space weather events may cause disruption and damage to necessary technologies. Machine learning methods are being developed to predict and classify space weather events so precautions may be taken to mitigate the impact. This thesis explores the application of machine learning methods to previously recorded riometer (ground-based receivers used in monitoring ionospheric behavior) data. The University of Calgary’s Auroral Imaging Group has been overseeing a network of riometers since 1989. Riometer data, though plentiful, is noisy and subject to cosmic and terrestrial interference. Utilizing riometer data currently requires manual assessment of data by experts which takes a significant amount of time. The purpose of this thesis is to design a method to automate space weather data classification. In this thesis, data is selected from one riometer site to facilitate the development of preprocessing methods and machine learning model design. First, methods were developed to manage raw data and produce a filtered signal. Next, the preprocessed data was explored to produce various features describing the behavior of riometer data. Lastly, a neural net was designed and trained to classify data behavior. The trained model succeeds in the binary classification of riometer data.
Description
Keywords
Machine Learning, Space Weather, Riometer, Signal Processing, Classification
Citation
Arnason, K. (2021). Machine learning methods applied to riometer data classification (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.