Comparison of seismic activity for Llaima and Villarrica volcanoes prior to and after the Maule 2010 earthquake

Abstract

Llaima and Villarrica are two of the most active volcanoes in the Chilean Southern Volcanic Zone and presently show contrasting types of activity. Llaima is a closed vent edifice with fumarolic activity, while Villarrica has an open vent with a lava lake, continuous degassing and tremor activity. This study is focused on characterizing the relationships between volcanic and seismic activity in the months before and after the 2010 M8.8 Maule earthquake, which was located in NNW direction from the volcanoes. Time series for tremors, long-period and volcano-tectonic events were obtained from the catalogue of the Volcanic Observatory of the Southern Andes (OVDAS) and from the SFB 574 temporary volcanic network. An increase in the amount of tremor activity, long-period events and degassing rates was observed at Villarrica weeks before the mainshock and continued at a high level also after it. This increase in activity is interpreted to be caused by enhanced magma influx at depth and may be unrelated to the Maule event. In Llaima, an increase in the volcano-tectonic activity was observed directly after the earthquake. The simultaneous post-earthquake activity at both volcanoes is consistent with a structural adjustment response. Since this enhanced activity lasted for more than a year, we suggest that it is related to a medium-term change in the static stress. Thus, the Maule earthquake may have affected both volcanoes, but did not trigger eruptions, from which we assume that none of the volcanoes were in a critical state.

Appendices

Appendix 1: SFB catalogue construction

Catalogue construction

To retrieve a seismic activity catalogue of the volcanoes Villarrica and Llaima, the following three steps were performed:

• filtering in three bands,

• automatic triggering and

• event detection.

Filtering

The data were filtered in the following frequency bands:

(1) between 0.25 and 0.5 Hz, (2) between 0.8 and 2.4 Hz and (3) between 6 and 12 Hz.

While volcanic events such as LP, TR and some HB events have their main energy in the second band, local, regional and teleseismic events would also appear in the other frequency bands (see Fig. 10).

Fig. 10

Left spectrogram for a volcanic event in Villarrica volcano. Horizontal axis marks seconds from 09:00:00 (GMT) of 17 January 2011. Centre spectrogram for a volcanic event in Llaima volcano. Horizontal axis marks seconds from 09:00:00 (GMT) of 17 January 2011. Right spectrogram for a tectonic regional event, for comparison of frequency band content. Horizontal axis marks seconds from 05:00:00 (GMT) of 17 January 2011

Automatic triggering

The aim of an automatic triggering is to identify a sudden increase in the recorded seismic energy. The most common approach is the calculation of the ratio STA/LTA (short-term average/long-term average) of the seismic energy. A sudden increase in the seismic activity would affect the STA, increasing the ratio. When the ratio exceeds a threshold, the trigger is active and the time is stored. Once the event is finished or levels of energy are below the threshold, the ratio decreases again to normal values. The parameters of the triggering (Table 1) were manually adjusted by visual inspection.

Table 1 Triggering parameters for the three bands studied

Event detection

A seismic event should be triggered on all seismic stations, while a random increase in seismic energy on one station should not be seen at the others. This random increases in energy could be generated by animal farming, human activity and/or nearby traffic. To distinguish between an event and a random increase, the times of the trigger are compared on several stations. Here, only the short-period stations were used because of consistency. If there are triggers on 2 or 3 stations within a time window, it is assumed to be a real seismic event. The time window had a length of 15 s due to travel time differences and time shift of the trigger onset caused by higher noise.

Appendix 2: removal of aftershock sequence from time series

The removal of the aftershock sequence from time series of seismic events was performed by fitting a combination of a linear trend and modified Omori’s law to a relevant part of the time series. The modified Omori’s law describes the rate of occurrence of aftershocks t days after the main shock (e.g. Lomnitz (1994)) and can be approximated for relatively short times by (Utsu 1969, 1970):

$$ \lambda (t) \approx \frac{\alpha }{{(t + \beta )^{p} }} $$

(1)

where α varies between 1 and 2 (Lomnitz 1994) and α and β are fitting parameters. To maintain the overall trend in the time series, a combination of modified Omori’s law and a linear trend model was used to fit the data:

$$ \lambda (t) = \left\{ {\begin{array}{l} {c_{0} + c_{1} t,\quad t < 0} \\ {\frac{\alpha }{{(t + \beta )^{p} }} + c_{0} + c_{1} t,\quad t \ge 0} \\ \end{array} } \right. $$

(2)

where t = 0 is the day of the main shock, that is, 27 February 2010. The fitting parameters α, β, p, c ₀ and c ₁ (see values in Table 2) were determined by minimizing the difference between the observed time series and eq. (2) in a least-squares sense. The parameter p was constrained to be between 1 and 2 following Lomnitz (1994). After that, the Omori part (Eq. 1) was calculated separately and subtracted from the time series.

Table 2 Values of fitting parameters for the band 0.8–2.4 Hz, for Llaima and Villarrica volcanoes

Appendix 3: activity in other volcanoes

Here, we describe the activity presented by other volcanoes prior to or after the earthquake, according to the reports from OVDAS.

Lonquimay and Planchón Peteroa

The station at Lonquimay volcano, deployed in May 2010, recorded since August 2010 VT activity situated between 8 and 16 km away from the volcano, possibly related to the Liquiñe-Ofqui Fault. On Planchón Peteroa volcano, a station was installed after a white fumarole was observed during August 2010. In September, a continuous tremor was accompanying the fumarole, which presented an intermittent change from white to grey (ashes). The tremor lasted until February 2011 and was combined with VT and LP events. On 18 April 2011, the volcano started its eruptive process, throwing ashes in the near radius (≤10 km), as a continuation of its previous process started in September 2010.

Mocho-Choshuenco, Calbuco, Chaitén and Osorno

Mocho-Choshuenco volcano presented some activity (approx. 20 VT events) in November 2009. After the earthquake, in March 2010, an increase in activity (463 events between VT and LP) was observed, but still remained under normal levels. Most of the attention was on the Chaitén volcano that had began its eruptive cycle in May 2008 and continued until March 2010 (Siebert and Simkin 2002). Osorno volcano showed LP activity in May 2009 which decreased in January 2010 and some VT events starting in July 2009. Activity was stated as “normal” in May 2010. Calbuco presented some activity in 2009, but related to Chaitén, and most of the time the stations were malfunctioning.

Cordón Caulle

Cordón Caulle volcano, which erupted at the beginning of June 2011, was presenting a normal to low VT activity in June–July 2009 and two VT events/day on average, with magnitudes 0.5 ≤ M ≤ 3.2. From November 2009 to January 2010, the data states as “not reliable” due to technical problems with the station (less than 6 events in a complete month). In August and November 2010, a significant increase in activity was observed (125 VT/month and 122 VT/month, respectively). The number of events per month raised to ca. 40 VT/month and very few LP events (less than three) on the following months were reported. On 26 April 2011, a column of gas started to flow out of the volcano accompanied by a swarm of Hybrid and LP events. On 2 June 2011, the OVDAS declares the eruption after a total of 25 events/hour.