MVO/VSG - Open Scientific Meeting
27 November 1996
GPS Measurements at the Soufriere Hills Volcano, Montserrat,
April-November 1996
John B. Shepherd
John B. Shepherd
Environmental Science Division, University of Lancaster, Lancaster LA1 4YQ
All of the following have made significant contributions to this programme of field measurement but have not necessarily been consulted about this contribution and may not agree with some of the conclusions.
Mark Davies, Joe Devine, Laurance Donnelly, Neil Dyer, Chloe Harford, Ricky Herd, Paul Jackson, Mike James, J-C Komorowski, Anne-Marie LeJeune, Ritchie Robertson, George Skerritt, Mark Stasiuk, Jane Toothill, Rob Watts.
We have made a comprehensive study of the deformation field associated with the current eruption of the Soufriere Hills Volcano using GPS. The study is complementary to a study made using conventional surveying techniques. Measurements cover the period April to November 1996. During this period over 100 benchmarks have been occupied. Over thirty baseline lengths have been measured regularly throughout this period at intervals of one to two weeks, except for a period in May-June when the equipment was returned to the manufacturers for repairs.
The receivers are Leica SR399 dual frequency GPS sensors with CR344 controllers. Data are processed immediately after collection using the Leica SKI software system and the broadcast satellite ephemeris. The method of measurement is the Rapid Static method whereby estimates of baseline lengths and the geodetic (WGS84) coordinates of the stations can be made based on 5-15 minute observation periods. Data are re-processed using the precise ephemeris as soon as this becomes available.
For convenience of field operation the principal stations have been grouped into three sub- networks each of which can be occupied completely in about 4-6 hours. One network covers the northern and eastern part of the volcano, a second covers the western flanks of the volcano and a third covers the whole volcano, linking the two smaller networks. There is a significant gap in the region to the southeast of the volcano which is almost totally inaccessible. The nearest stations to the currently-active dome are at distances of about 1.5 km. Benchmarks are either pre-existing survey points installed by the Department of Overseas Surveys 25-30 years ago or 2.5 cm diameter stainless steel pins. Wherever possible the new benchmarks have been cemented in to holes drilled 10-20 cm into solid rock. A few benchmarks, particularly those used for single occupations, are in less stable positions.
Line lengths in the smaller networks are in the range 1.5-2.5 km and in the larger network 5-7.5 km. Most lines pass close to the active dome and some pass directly across it. Stations not included, or only occasionally included, in the three main networks have been occupied for specific purposes including providing accurate locations for gravity and microgravity stations, points from which photographic, Vector binocular and theodolite measurements are made and base stations for helicopter/binocular surveys of the growing dome. These applications will be described separately. The manufacturers specifications for this system indicate that the accuracy which can be achieved using the Rapid Static technique is about ± (5mm + 1ppm baseline length). Previous measurements using a Total Station to a reflector within 500 metres of the dome (and therefore not routinely accessible for GPS measurements) had shown sustained displacements of order 1mm/day on baselines of length 1-2km throughout the current eruption, rising to up to 5 cm per day during periods of increased activity. The sense of motion indicated compression of the rocks to the east of the dome. Displacements of this order are well within the detection capability of the GPS system.
After an initial period of fairly serious teething problems with the instruments, the repeatability of baseline measurements appears to be rather better than this. The standard deviation of measurements of baseline length is only weakly dependent on the length of the baseline but much more strongly dependent on the elevation difference between the ends of the baseline. A preliminary relationship is:
where sigma is the standard deviation of a single measurement, D is baseline length and H is elevation difference. For the baselines measured regularly this relationship translates into standard deviations in the range 4-8 mm or strain levels of 1-2x10-6.
Within these limits we have detected no significant changes in any of the baseline lengths measured over the period from June to November when the system was performing best. This conclusion applies to all baselines measured in all three networks. In this context "significant" means that we have rarely observed differences between successive measurements, deviations from the long-term mean, or accumulated differences over the whole period which are greater than one standard deviation, and never observed differences greater than two standard deviations. All fits of linear trends to the data yield results which are not significantly different from zero.
These results indicate that, at a strain level of a few times 10-6, the strain field associated with this eruption is confined to distances less than 1.5 km from the dome. A more widespread displacement field, such as general uplift of the whole area which includes the networks, cannot be excluded but is not accessible to measurement by this technique.