Constraints and conundrums resulting from ground-deformation measurements made during the 2004-2005 dome-building eruption of Mount St. Helens, Washington

Abstract

A prolonged period of dome growth at Mount St. Helens starting in September-October 2004 provides an opportunity to study how the volcano deforms before, during, and after an eruption by using modern instruments and techniques, such as global positioning system (GPS) receivers and interferometric synthetic aperture radar (InSAR), together with more traditional ones, including tiltmeters, triangulation, photogrammetry, and time-lapse photography. No precursory ground deformation was detected by campaign GPS measurements made in 2000 and 2003, nor by a continuous GPS station (JRO1) operating ~9 km to the north-northwest of the vent area since May 1997. However, JRO1 abruptly began moving downward and southward, toward a source centered about 8 km beneath the volcano, concurrently with the start of a shallow earthquake swarm on September 23, 2004. The JRO1 velocity slowed from ~0.5 millimeters per day (mm/d) in late September–early October 2004 until spring 2005. Thereafter, it was essentially constant at ~0.04 mm/d through December 2005. In similar fashion, the growth rate of the welt on the south crater floor slowed from 8.9 m 3 /s during October 4–11 to 6.4 m 3 /s during October 11-13, 2004; this trend continued after emergence of the first lava spine on October 11. The volumetric extrusion rate decreased from 5.9 m 3 /s during October 13-November 4, 2004, to 2.5 m 3 /s during December 11, 2004-January 3, 2005, and for the remainder of 2005, it was in the range 2.0-0.7 m 3 /s. Fifteen continuous GPS stations, installed soon after the eruption began, showed radially inward and downward ground motions through December 2005. Likewise, InSAR observations spanning the first year of the eruption indicate broad subsidence centered near the vent. Model-derived estimates of source-volume decrease from September 23, 2004, to October 31, 2006, are 16-24×10 6 m 3 , substantially less than the volume erupted during the same period (87×10 6 m 3 through October 21, 2006). The discrepancy can be explained by a combination of magma expansion and recharge in the source region. Lack of precursory deformation at JRO1 suggests that the conduit is poorly coupled to the rest of the edifice, so the rising magma column was able to push ahead older conduit material rather than intruding it. Constraints on conduit length and radius require that reservoir magma (as opposed to conduit-filling magma) reached the surface early during the eruption, probably soon after CO 2 emission rates peaked in early October 2004. If rapid emergence of spine 3 (the first whaleback-shaped extrusion) in late October 2004 marked the arrival of reservoir magma, then the volume of conduit material flushed from the system was about 20×10 6 m 3 --the volume of surface deformation plus spines on November 4, 2004. The corresponding radius for a cylinder extending from the surface to depth d = 5 km is 35.7 m, or 28.2 m for d = 8 km. The average ascent rate through the conduit, assuming reservoir magma began its rise on September 23, 2004, was 120 m/d for d = 5 km, or 190 m/d for d = 8 km. Observed lineal extrusion rates were 2-10 m/d, so the conduit must widen considerably near the surface. Equating magma flux through the conduit to that at the surface, we obtain a vent radius of 125 m and an extrusion rate of 5.7 m 3 /s--both values representative of the early part of the eruption. Lack of precursory inflation suggests that the volcano was poised to erupt magma already stored in a crustal reservoir when JRO1 was installed in 1997. Trilateration and campaign GPS data indicate surface dilatation, presumably caused by reservoir expansion between 1982 and 1991, but no measurable deformation between 1991 and 2003. We conclude that all three of the traditionally reliable eruption precursors (seismicity, ground deformation, and volcanic gas emission) failed to provide warning that an eruption was imminent until a few days before a visible welt appeared at the surface--a situation reminiscent of the 1980 north-flank bulge at Mount St. Helens.