Can repeating glacial seismic events be used to monitor stress changes within the underlying volcano? -Case study from the glacier overlain Katla volcano, Iceland

The glacier overlain Katla volcano in South Iceland, is one of the most active and hazardous volcano in Europe. Katla eruptions result in hazardous glacial floods and intense tephra fall. On average there are eruptions every 50 years but the volcano is long overdue and we are now witnessing the longest quiescence period in 1000 years or since the settlement. Because of the hazard the volcano poses, it is under constant surveillance and gets a good share of the seismic stations from the national seismic network. Every year the seismic network records thousands of seismic events at Katla with magnitudes seldom exceeding M3. The bulk of the seismicity is however not due to volcano tectonics but seems to be caused mainly by shallow processes involving glacial deformation. Katla's ice filled caldera forms a glacier plateau of several hundred meters thick ice. The 9x14 km oval caldera is surrounded by higher rims where the glacier in some places gently and in others abruptly falls off tens and up to hundred meters to the surrounding lowland. The glacier surface is marked with dozen depressions or cauldrons which manifest geothermal activity below, probably coinciding with circular faults around the caldera. Our current understanding is that there are several glacial processes which cause seismicity; these include dry calving, where steep valley glaciers fall off cliffs and movements of glacier ice as the cauldrons deform due to hydraulic changes and geothermal activity at the glacier/bedrock boundary. These glacial events share a common feature of containing low frequency (2-4 hz) and long coda. Because of their shallow origin, surface waves are prominent. In our analysis we use waveforms from all of Katla's seismic events between years 2003-2013, with the criteria M>1 and minimum 4 p-wave picks. We correlate the waveforms of these events with each other and group them into families of highly similar events. Looking at the occurrence of these families we find that individual families are usually clustered in time over several months, and sometimes families may reappear even up to several years later. Using families including many events and covering long periods (10-20 months) we compare the coda (the tail) of individual events within a family. This is repeated for all the surrounding stations. The analysis, coda wave interferometry (cwi) is a correlation method that builds on the fact that changes in stress in the edifice lead to changes in seismic velocities. The coda waves are highly sensitive to small stress changes. By using a repeating source, implying we have the same source mechanism and the same path, we can track temporal stress changes in the medium between the source and the receiver. Preliminary results from Katla suggest that by using the repeating glacial events and the coda wave interferometry technique we observe annual seismic velocity changes around the volcano of ca. 0.7%. We find that seismic velocities increase from January through July and decrease in August to December. These changes can be explained by pore-water pressure changes and/or loading and de-loading of the overlain glacier. We do not find immediate precursors for an impending eruption at Katla; however we now have a better understanding of its background seismicity.