Periodic Seismic Activity at the Soufriere Hills Volcano Rod Stewart MVO, Mongo Hill, Montserrat 13 September 1997 periodic.txt Introduction ------------ One of the most notable features of the seismic activity associated with the current eruption is that it often has a periodic nature. Regular cycles seem to occur in the seismic activity on four or five different time scales, from less than a minute to several decades. Observations of periodic seismicity are not unknown at volcanoes, although the mechanisms for them are not well understood. Tides, ocean and solid-earth, have often been proposed as a mechanism for near-diurnal periodicity, although at least one of these examples is erroneous (Stewart, in prep). Shorter-period variations are often seen as banded tremor, bands of continuous seismic tremor, sometimes harmonic, which may or may not occur at regular intervals. McNutt (references unknown) has compiled examples of banded tremor from active volcanoes. Most of it appears to be on the minute scale, but some of it is longer period, several hours. As far as I can remember, none of it has persisted for as long as tremor at Soufriere Hills volcano seems to have. The periodic activity which is of most interest in this eruption and the one which is examined in depth here, is periodicity on the time-scale of several hours, which has been observed more often than not since July 1996. It is thought that Montserrat is unique in having such a persistent periodicity to its activity. What makes us special is that periodicity is the defining thing. It occurs with many different types of earthquakes. This report documents this periodicity and attempts to analyse it in terms of volcanic activity. First give some examples. Much use of RSAM data, both from the SP and VME BB system. The latter has two advantages, sensitivity and non-non-linearity. But it does suffer from missing values. explain why. Some Comments on Seismic Noise and Tides ---------------------------------------- Since we are discussing variations of seismicity that occur regularly, we need to be cognisant of the possible effects of noise. Natural background noise could influence the interpretation in two ways. First, it will be directly measured in the RSAM data. Second, changes in the noise level will affect the detection threshold below which seismic signals cannot be seen. It would be thought that this is a very minor effect, but it is well known that the global earthquake catalogue is biased by this effect and there is one fairly well documented example of such effects at Rabaul volcano (Stewart, in prep). Normally, noise variations are diurnal. This includes cultural noise generated by human activity and natural noise generated mainly by the weather. The former is thought to dominate periodic phenomena and only seems to have a 24-hour component based around the daily grind. There are however some examples of natural diurnal noise - stations near pack ice can record freeze-thaw signals (AVO monthly report, date unknown). Stations in windy locations sometimes show a diurnal noise variation (Stewart, unpublished data from Ulawun and Rabaul volcanoes). I dont know any examples of non-diurnal natural noise variations. The effects of tides, ocean or solid-earth on noise are not known in this area. And it has been held to effect seismicity. However, we would expect such to be always 24 hours, or perhaps 12. Decade-scale periodicity (pre-eruption) ------------------------------------ Do not forget the apparent long-term periodicity. Prior to this eruption, there were four volcano-seismic crises, in 1897-98, 1933-37, 1966-67 and 1992-1995 (see Shepherd, Tomblin and Woo, 1971 and Ambeh and Lynch, 1996). The times between the starts of these crises is 36, 33 and 26 years. It is difficult to see such a sequence arising from pure chance, but no mechanism as yet has been proposed. Month-scale periodicity ----------------------- Barry's month-long cycles. Ten-hour-scale periodicity -------------------------- This is the most pervasive of the five types of periodicity and has been discussed in Stewart (1997). Here we enlarge in that discussion and include more results. Period of ten-hour-scale periodicity ------------------------------------ The periods of this periodicity are often routinely referred to as 12-hour, although a number of other periods have been noted. A more formal analysis of the different things is beyond the scope of this report, although we present here some thoughts on. The times and durations of the VT and hybrid earthquake swarms since the beginning of the eruption have been carefully recorded. Analysis of these gives some insight into the periodicity and is presented here for the 1997 data (after R. Luckett). 161 swarms were recorded, with inter-swarm times (from the start of one swarm to the start of the next) between 3.36 and 76.25 hours. The frequency-distribution of the inter-swarm interval is shown in Figure X, counted in one-hour bins. This appears to be a log-normal distribution. The interval between swarms has a modal value of 14.5 hours, a mean of 16.92 hours and a median of 14 hours. Tims things Hour-scale periodicity ---------------------- There is only one example of this, but it is so striking as not to be happenstance. This occurred on 3 September 1997. At the time, there was regular periodic low-level seismicity, with periods around 16 hours. Figure Xa shows the VME RSAM data for MB?? for the day. This shows well the cycle in the semi-continuous signals, lasting from 08:00 until 16:00 in the figure. Note that the time here is not correct, this is a consequenc of using BOB to process the data, the actual time was 50 minutes earlier, ie 07:10 to 15:10 UTC. Figure Xb shows the low-frequency part of the RSAM data, calculated by smoothing the data using a 100-point running mean filter. Figure Xc shows the ratio of the first two, that is the amplitude as a proportion of the local "background" signal level. This shows very clearly that the individual larger rockfalls that occured throughout this sequence arfe very evenly spaced, about 1 hour apart. There are seven consecutive such rockfalls, before a series of small rockfalls terminates the activity. Minute/second-scale periodicity ------------------------------- This subject has already been touched upon briefly in Stewart (1997), but will be enlarged upon here. Models ------ The development of a model to explain the cycles in the seismicity is beyond the scope of this report, probably beyond my comprehension. And it is difficult, nay foolish, to try and encompass all the different scales of periodicity. There are some things in the periodicity that might help constrain a mechanism. Its relationship to other things. And the changes in periodicity. Relationship to other things ---------------------------- Changes in periodicity ---------------------- Two types of changes in the periodicity have been observed. Smooth changes in the time between peaks. And abrupt changes in the periodicity. At other times, the period has been amazingly stable. Back in July/August ------------------- Extrusion of Santa Lobe, December 1996 -------------------------------------- 24 June 1997 ------------ Acknowledgements ---------------- All of the data presented here has been put together by that fine body of men and women, Team Seismic. I am indebted to them for this. Many of the ideas here are due to, or have been developed in conjunction with, various members of Team Seismic. By presenting them here, I am not claiming ownership and any proper publication should ensure proper attribution. I have not attempted that here for two reasons. First, I am presenting my thoughts on the data. Second, I just couldn't be arsed. References ---------- (Sorry but some of these are sketchy because I don't have them with me. The full references can probably be supplied on request.) Ambeh, W.B. and Lynch, L.L., 1996. Seismicity preceding the current eruption of the Soufriere Hills Volcano, Montserrat, West Indies, in Science, Hazards and Hazard Management - Volcanism in Montserrat (ed. R. Ahmed), Poster Session at The Second Caribbean Conference on Natural Hazards and Disasters, 9-12 October 1996, Kingston, Jamaica. Shepherd, J.B., Tomblin, J.F. and Woo, D.A., 1971. Volcano-seismic crisis in Montserrat, West Indies, 1966-67, Bull. Volcanol., 35, 143-163. Stewart, R.C., 1997. A report, unpublished report for Montserrat Volcano Observatory. Stewart, R.C., in prep. Tidal triggering of Rabaul earthquakes re-examined, submitted to Geophys. J., 1997.