Automating long-term glacier dynamics monitoring using single-station seismological observations and fuzzy logic classification: a case study from Spitsbergen

• Published in: Journal of Glaciology
• Main Authors: GAJEK, W., TROJANOWSKI, J., MALINOWSKI, M.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Cambridge University Press (CUP) 2017
• Subjects: Earth-Surface Processes; Arctic; Hornsund; Journal of Glaciology; Kongsfjord*; Kongsfjorden; Spitsbergen
• Online Access: http://dx.doi.org/10.1017/jog.2017.25; https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0022143017000259

ABSTRACT Retreating glaciers are a consequence of a warming climate. Thus, numerous monitoring campaigns are being carried out to increase understanding of this on-going process. One phenomenon related to dynamic glacial changes is glacier-induced seismicity; however, weak seismic events are difficult to record due to the sparse seismological network in arctic areas. We have developed an automatic procedure capable of detecting glacier-induced seismic events using records from a single permanent seismological station. To distinguish between glacial and non-glacial signals, we developed a fuzzy logic algorithm based on the signal frequency and energy flow analysis. We studied the long-term changes in glacier-induced seismicity in Hornsund (southern Spitsbergen) and in Kongsfjorden (western Spitsbergen). We found that the number of detected glacial-origin events in the Hornsund dataset over the years 2013-14 has doubled. In the Kongsfjorden dataset, we observed a steady increase in the number of glacier-induced events with each year. We also observed that the seasonal event distribution correlates best with 1 month lagged temperatures, and that extreme rain events can intensify seismic emissions. Our study demonstrates the possibility of using long-term seismological observations from a single permanent station to automatically monitor the dynamic activity of nearby glaciers and retrieve its characteristic features.