Time dependent deformation of Kilauea Volcano, Hawaii

In 1997 the continuous Global Positioning System (GPS) network was completed on Kilauea, providing the first network of daily position measurements during eruptions and earthquakes on Kilauea. Kilauea has been studied for many decades with continuous seismic and tilt instruments. Other geodetic data (e.g., campaign GPS, leveling, electronic distance measurements) are also available although they contain only sparse data. Data analysis methods used here include inverting multiple data sets for optimal source parameters and the spatio-temporal distribution of magma volume and fault slip, and combining GPS and seismic observations to understand flank tectonics. The field area for this study, Kilauea Volcano, was chosen because of its frequent activity and potential hazards.;The 1997 East Rift Zone eruption (Episode 54) was the first major event to occur after the completion of the continuous GPS network. The event lasted 2 days, but transient deformation continued for six months. This long-duration transient allowed the first spatio-temporal study of transient dike deformation on Kilauea from daily GPS positions.;Slow-slip events were discovered on Kilauea during which the southern flank of the volcano would accelerate seaward for approximately 2 days. The discovery was made possible because of the continuously operating GPS network. These slip events were also observed to correlate with small swarms of microearthquakes found to follow temporal pattern consistent with them being co- and aftershocks of the slow-slip event (Segall, 2006). Half-space models of geodetic data favor a shallow fault plane (∼ 5 km), which is much too shallow to have increased the Coulomb stress at the depths of the co- and aftershocks. However, optimizations for the slow-slip source parameters including a layered elastic structure and a topographic correction favor deeper models within the range of the co- and aftershocks. Additionally, the spatial distribution of seaward fault slip, fixed to a decollement structure 8 km under the south flank, and the locations of the microearthquakes suggest that both occur on the same structure.;In 2007, Episode 56 of the Pu'u 'O'o-Kupianaha eruption occurred. This episode was exciting both because it was the largest intrusion in the last decade, and because it occurred concurrently with a flank slow-slip event. The intrusion started on Father's day (June 17th), 2007 with increased seismicity and abrupt tilts at the summit and rift zones. Quasi-static models of the total deformation determined from GPS, tilt, and InSAR indicate that the intrusion occurred on two en echelon dike segments in the upper East Rift Zone along with deformation consistent with slow-slip in the same areas of previous events. The ∼ 2 m maximum opening occurred on the eastern segment near Makaopui crater. Unlike previous intrusions in 1997, 1999, and 2000, the dike model was not sufficient to explain deformation on the western flank. Additionally, a coastal tiltmeter installed in anticipation of a slow-slip event recorded tilts consistent with those observed during the 2005 slow-slip event. These observations led to the conclusion that a concurrent slow-slip event occurred. Geodetic models indicate a similar amount of decollement slip occurred as in previous slow-slip events.;Sub-daily GPS positions were used to study the spatio-temporal distribution of the dike intrusion. The time-dependent intrusion model shows that the intrusion began on the western en echelon segment before jumping to the eastern segment, which accumulated the majority of the 2 m of opening. Sub-daily GPS positions limit the number of stations available since there are very few continuous stations north of the East Rift Zone, where coverage is critical for separating the intrusion from the slow-slip. However, an ENVISAT interferogram at 08:22 on June 18, 2007 provides additional spatial coverage of deformation up to that point. Combining this image with the GPS and tilt data up to that point, we perform a quasi-static inversion for the intrusion source. The residual deformation indicates that slow-slip had not significantly progressed by the ENVISAT image. The slow-slip event occurred therefore at least 20 hours after the initiation of the dike intrusion. (Abstract shortened by UMI.)...