A Geophysical Study of Active Volcanic Regions and Subduction Zones
The solid Earth is a dynamic system and as such is constantly changing. Some of these changes involve the large energetic release of volcanic eruptions, megathrust earthquakes or the moving of large ice sheets at the poles towards the oceans. Each of these phenomena have either a direct or an indirect impact for human populations and understanding them might be the only solution to be more resilient to their effects. Particularly, I am interested in understanding how volcanoes work and the effect that they might have in the places that surround them. Are volcanic regions intrinsically different from other places on Earth? Do they produce more earthquakes or less? What exactly happens inside, and nearby, of a volcano between and during major explosive eruptions? And do they influence the thermal structure of the surrounding crust? My doctoral dissertation has focused on addressing these types of questions by utilizing geophysical methods that include modern techniques applied to old data and the application of classic methods to modern data.Chapter 1 explores the hypothesis that volcanic regions might be seismically different from non-volcanic regions by studying aftershock productivity, the ability for any given earthquake in a region to trigger another earthquake, in Japan. Focusing our study area on Japan provides a great advantage for studying the statistical relationship of earthquakes and volcanoes, since the country has more than 120 active volcanoes and it is densely instrumented with seismometers. Our study showed that, in general, earthquakes in volcanic regions have the same chance of triggering another earthquake than earthquakes in non-volcanic regions, suggesting that aftershock productivity is not controlled by heat-flux in the crust but rather by other mechanisms such as the thickness of brittle crust. Chapter 2 addresses the question of what happens inside a volcano while it is undergoing a prolonged explosive eruption sequence. The case study is Okmok Volcano in the Aleutian Islands in Alaska. In 2008 Okmok broke its historical pattern of effusive activity during the last 100 years by undergoing a Volcanic Explosivity Index (VEI) 4 hydromagmatic eruption. Co-eruptive seismicity is difficult to measure but essential to interpreting an eruption. It is potentially the highest resolution technique to chronicle the failure of the volcanic edifice and the evolution of volcanic vents. Standard earthquake detection methods often fail during this time as the eruption itself produces seismic waves that obscure the earthquake signals. We address this problem by applying template matching combined with machine-learning and fingerprint-based techniques (EQTransformer and FAST) to expand the existing seismic catalog of the Alaska Volcano Observatory (AVO) by finding seismic signals during the continuous eruptive sequence. We find that: (1) the most significant co-eruptive seismic sequences occurred nearly half-way through the eruption and included a burst of long-period (LP) earthquakes directly under the eruptive vent followed by ash-rich plumes. Subsequent LP episodes occurred in bursts that migrated and were again followed by plumes. The LP earthquake rate and the plume rates anti-correlate. (2) A distinct population of volcano-tectonic (VT) earthquakes maintain a steady rate over a large region on the southern side of the caldera. These co-eruptive VTs have a larger b-value than before or after the eruption, which agrees with an open (extraction)/close(injection) system. (3) The cessation of the eruption is marked by a sudden burst of LP earthquakes which is accompanied and followed by a steady stream of small VT earthquakes that occur south (and therefore distinct) from the previous concentration of seismicity at the active vents. (4) The opening sequence extends to ~15 km depth and is marked by a migration, within minutes of the onset of the eruption, towards the previously identified center of deformation, which is located under the center of the caldera and is distinct from the eruptive vent. (5) The center of the caldera lacks any type of seismicity throughout the eruption. These previously inaccessible observations demonstrate that the co-eruptive LP and VTs show distinct behavior, likely signifying distinct controlling processes. LPs are strongly localized in space and time and directly precede, but are not concurrent with, mass ejection in contrast with the steady, widely distributed VTs. This distinction is consistent with LPs being the signature of an eruptive process that occurs when the pressurization state is high, i.e., volcanic system partially closed, whereas VTs are a secondary effect that does not track the deflation of the magma reservoir and occur at a low stress-state, as indicated by the b-value, during the eruption which may be indicative of an at least partly open volcanic system. Chapter 3 focuses on studying the effects of active volcanism on the thermal structure of a polar region, where such effects can have implications for how fast ice-sheets and glaciers flow into the Southern Oceans. In this study, we used the data of an airborne transient electromagnetic survey conducted in the McMurdo Sound, Antarctica, to investigate spatial variations of geothermal heat-flux (GHF) on a regional scale. We successfully increased the number of geothermal heat-flux constraints in the region by a factor of 5 and our results show that GHF variations are weakly controlled by the tectonic structure in the region but strongly controlled by the active volcanic province. We found that the only area clearly exhibiting high GHF (>100 mW/m2) is Ross Island, which hosts one of the most active mafic volcanoes in the world, Mount Erebus. The GHF measurements from the western portion of the Terror Rift, the coastal ranges of the McMurdo Dry Valleys, show the lowest GHF values (e.g., 60-70 mW/m2), but they also reveal local variations within the study area possibly related to cooling magma bodies from >~5Ma, while the rift-shoulder uplift to the west is typified by somewhat higher values (80-90 mW/m2). The regional compilation of GHF constraints combined with our new data further reinforces the conjecture that areas of high GHF in Antarctica will generally be associated with Cenozoic volcanic and magmatic activity while tectonic history has comparatively smaller impact on GHF distribution. These lessons from this best studied part of Antarctica may aid further improvements of a continent-wide GHF dataset which in turn can help improve future ice-sheet models and the projections of Antarctica’s contribution to near-future global sea level rise. Finally, chapter 4 discusses the occurrence of large, repeated earthquakes along the southeastern Mexican subduction zone and the relationship of the microseismic foreshocks and slow slip phenomena during the nucleation process of one of these earthquakes. After a complex sequence of tectonic events that started with the large Mw8.2 intraplate Tehuantepec earthquake, in February 2018, a Mw7.2 earthquake ruptured under Pinotepa Nacional, Oaxaca (hereby referred to as the Pinotepa earthquake). As it will be shown, this earthquake was preceded by an increase in Coulomb Failure Stress change as well as an increase in the interface coupling influenced by the growth of a slow-slip event (SSE) caused by the large Tehuantepec earthquake. At the same time, a cascade of microseismicity showed an increasing rate in the hypocentral region, leading up to the rupture of the Pinotepa earthquake. These observations suggest that the nucleation process of the Pinotepa earthquake was driven by a long-range change in the stress regime due to the SSE that was developing downdip but could not penetrate the hypocentral region. Furthermore, the rupture area of the Mw7.2 Pinotepa earthquake superimposes the area covered by the aftershocks of the Ms7.1 1968 earthquake. Multiple teleseismic records available for both earthquakes provide an unprecedented dataset and the possibility to test the hypothesis that they ruptured the same distinct asperity along the megathrust boundary. The comparison of these records reveals that both earthquakes produced nearly identical seismic waveforms, indicating that, indeed, they ruptured the same frictional patch, and it validates the asperity model originally proposed by Lay and Kanamori (1981).