Scientific Monitoring of the Soufriere Hills Volcano at the Montserrat Volcano Observatory, 2000 - 2003

Dramatic improvements were made in the scientific monitoring programmes between 2000 and 2003 by Ricky Herd, Marie Edmonds and Glenn Thompson, which improved scientific understanding of the Soufriere Hills Volcano and brought acclaim to the Montserrat Volcano Observatory and the British Geological Survey.

MVO volcanologists Marie Edmonds and Ricky Herd working with external collaborators developed a revolutionary sulphur dioxide monitoring system between 2001 and 2002 called DOAS/ScanSpec, which provided a cheap, automated alternative to laborious manual collection of data with COSPEC. The result was about a 100 times improvement in the amount of data acquired, as well as freeing them up to perform other work. This breakthrough in the remote sensing of volcanic gases was a centrepiece at scientific conferences, the subject of an IMAX documentary, and brought much interest from other volcano observatories.

Ricky also developed a new photographic monitoring system. Previous attempts by many individuals to monitor the Soufriere Hills Volcano with video cameras had only worked temporarily, due to the enormous amount of power required, and the maintenance this entailed. Ricky saw that the solution to this was instead to use low power (still) digital cameras which nevertheless provided a much higher resolution than was possible with any conventional video camera. By automating this to take an image one a minute, and installing more than one camera station, time lapse video sequences could be made later showing the pattern of dome growth and collapse from different vantage points. Moreover, MVO finally had the ability to see the volcano in near-real-time, even though the observatory building did not itself provide a vantage point.

The seismic monitoring was in need of much work as a result of non-Y2K compliant operating systems and other software, reliance on specialist hardware and operating systems difficult to maintain locally, and lack of critical spares. While round-the-clock seismic monitoring underpinned safety, 50% of data were being lost, and acquisition systems had to be rebooted manually up to 25 times a day. The main data analysis computer had also been removed in early 1999, along with loss of all its software, during a period in which there was no Seismologist at MVO. So the effectiveness of seismic monitoring at MVO had significantly declined compared to 1996-1998 levels. Moreover, by the standards of US volcano observatories, the seismic monitoring at MVO was very outdated.

Glenn Thompson, who took over as MVO Seismologist in January 2000, brought a "can-do" approach and considerable experience of developing mission critical software at other volcano observatories. The most immediate problems were to put in place effective seismic data acquisition and alarm systems, as without these, MVO lacked the capability to alert the local and aviation authorities to frequent escalations in volcanic activity that could endanger lives. With a rapidly growing dome, this was even more critical. Datastreams from MVO's two seismic networks were merged for the first time, considerably improving the accuracy of event classification and location, and increasing efficiency. A vast array of data analysis tools were written in Matlab to modernise MVO's seismic monitoring capability and bring it inline with that at other leading observatories. Internal webpages were developed to host a wide range of near-real-time data which made it possible to make a detailed, yet rapid, assessment of all seismic parameters in response to volcanic alarms. Several safety nets were built into the seismic monitoring infrastructure including standardisation of operating systems and computer hardware for ease of maintenance, individual UPS for each computer to reduce brown-outs, reconfiguration of critical software to autoreboot in the event of a brown-out, mirroring of all mission critical computers to provide failover, on-the-shelf ready-to-go backups for each mission critical computer, and a diagnostic alarm system which monitored all potential points of failure and paged the Seismologist if any concerns were detected and put detailed information on internal webpages (the diagnostic alarm system also had a mirror). It was through measures such as these that an unprecedented level of seismic monitoring was achieved at MVO from 2001 to 2003, and this more than any other activity at MVO did the most to ensure safety.

Massive efforts were also made to secure seismic data and make it accessible to researchers. Prior to 2000, data archival had been a slightly haphazard affair. Numerous types of media had been used including Tracker tapes, CD-R, Zip100 diskettes and DDS-1 DAT tapes. It was the storage of the most valuable dataset - seismic data from the digital seismic network - that was particularly concerning. The archive software included no checks, and as a result a lot of data had been lost. So the first step was to put in place robust, foolproof data archival software. Next, there was only around 1GB of online storage capacity to store seismic data. New hard drives were acquired increasing overall capacity to around 1TB, and the major exercise of extracting old data from hundreds of tapes and diskettes began. Online databases were then built, and new copies of the data made to high capacity DDS-3 DAT tapes. By establishing online databases, first of all it meant new data processing could easily be applied to old data. This allowed new derived datasets to be compiled for research purposes. Moreover, it also meant that MVO was able to provide larger volumes of data to university researchers, and do so in a dramatically expediated manner.

In January 2003, MVO completed its move from the north of the island where it had been evacuated in 1997, to a purpose-built observatory above the main town of Salem, with excellent views of the volcano. Remarkably considering the complex data telemetry comprising dozens of field stations of different types and repeaters, this move was achieved without the loss of any data, at tense time when a new evacuation had come into effect, and all eyes were on the volcano.

The considerable advances in monitoring at MVO from 2000-2003 were the direct result of the British Geological Survey being able to appoint enthusiastic knowledgeable volcanologists and volcano-seismologists, who were primarily motivated by developing the monitoring infrastructure at MVO and training local staff, rather than pursuing their personal research interests. It was this drive that enabled to accomplish so much, despite the inadequacy of technical support, and the abusive and threatening behaviour that often went unchallenged at MVO.