Detecting Seismic Anomalies in Electromagnetic Data Observed by SWARM Satellites Using Big Data Analytics

Y Bi, Vyron Christodoulou, FG Wilkie, David H. Glass, Guoze Zhao, Bin Han

Research output: Contribution to conferencePosterpeer-review


This work will report the studies undertaken in the past years, which demonstrate the application of the developed anomaly detection algorithms to analyse seismic anomalies in the electromagnetic data observed by Swarm satellites. There are two issues in anomaly detection analytics, priori knowledge required in preprocessing and parameter setting for running the algorithms, and temporal relation is often lost when converting original time series data to another representation. This work addresses both issues by reducing the original data while retaining temporal information by using a Fuzzy System to quantitatively characterise and symbolically represent time series data and using a Bruteforce algorithm as a means for parameter configuration. We evaluate our algorithm over twelve benchmark datasets of different kinds of anomalies and compare three detection algorithms, namely CUSUM-EWMA, HOT-SAX and GrammarViz. Using the empirical benchmark results and the parameter setting, we apply our algorithm to electromagnetic data observed by the SWARM satellites to detect anomalies for three earthquakes.
Original languageEnglish
Publication statusPublished (in print/issue) - 27 Jun 2017
EventESA-NRSCC Dragon Symposium 2017 - Copenhagen, Denmark
Duration: 24 Jun 201730 Jun 2017


ConferenceESA-NRSCC Dragon Symposium 2017


Dive into the research topics of 'Detecting Seismic Anomalies in Electromagnetic Data Observed by SWARM Satellites Using Big Data Analytics'. Together they form a unique fingerprint.

Cite this