Time Series Analysis of Continuous GPS Measurements Prior to a Volcanic Eruption: An Example of the Whakaari Volcano in New Zealand

• Activity Number: 322 - Analyses in Ecology, Epidemiology, and Environmental Policy
• Type: Contributed
• Date/Time: Wednesday, August 11, 2021 : 3:30 PM to 5:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): KATERINA TSAKIRI* and ANTONIOS MARSELLOS
• Companies: Rider University and Hofstra University
• Keywords: time series decomposition; moving correlation; KZ filter; GPS data

Abstract

Active volcanic areas are prone to eruptions and visiting under safe conditions has always been a challenge. GPS time series data for the period 2012-2021 in the volcano Whakaari/White Island in New Zealand have been analyzed for the determination of signals before the volcano eruption. For the analysis, the time series decomposition has been applied in the continuous recorded GPS location (east, north, vertical) data from the RGWI station using the Kolmogorov-Zurbenko (KZ) filter. The KZ filter removed the high frequency on the data and long-term patterns can be revealed in the analysis. Using the KZ filter and moving correlations on the raw data, we can identify possible signals and related periodicities of volcanic activity before an imminent volcanic eruption. The correlation between different vectors (north, east, and vertical) increases prior to and maximizes during the volcanic eruption. After removing of exogenic effects that influence the GPS signals, we determine the tectonic or/and volcanic changes before a volcanic eruption occurs. This methodology may facilitate a more accurate risk evaluation and can be applied in other locations, as well.