Search

The compaction of firn (the transition of snow to ice) is a process in glaciology that is difficult to constrain without the use of intrusive geophysical techniques. Full Waveform Inversion (FWI) is a seismic imaging technique used to obtain high-resolution images of the subsurface. This research introduces FWI to glaciological modelling and shows that FWI can improve upon the Herglotz-Wiechert travel time tomography method that's use is currently widespread. Using synthetic firn velocity models, acoustic FWI was shown to recover the compressional (P) wave velocity gradient of the firn column and could correct a velocity model that was a poor representation of the subsurface structure (up to a 5% velocity error). Additionally, FWI was able to detect ice layers of a minimum thickness of 10 m (half the minimum wavelength) within the firn column and provide constraints on their depth providing the ice slab did not occur in the shallow surface (20 m or less), where the steepest velocity gradient occurs. FWI is applied to two datasets. Firstly, legacy data acquired by the British Antarctic Survey on Pine Island Glacier were problematic, owing to an inconsistent, high-frequency source waveform that made FWI vulnerable to cycle skipping. Consequently, no improvement was observed to the seismic structure. Secondly, a dataset acquired specifically for this research was collected on Norway's Hardangerjøkulen Ice Cap. With this dataset a combination of Travel Time and Least-Squares FWI can be successful due to the use of a repeatable seismic source with lower frequency content (10 Hz), A VSP survey used to provide near-surface velocity constraints and near offset small receiver spacing (1 m). FWI provided a 6% change to the starting velocity model and improved the data match by 47%. The results showed that acoustic FWI must be considered from the seismic acquisition stage for the technique to be successful in characterising firn velocity structure. In this case, FWI can improve the model of the subsurface velocity structure from the HW technique and increase the match of the modelled seismic data.

In this thesis I investigate the tectonic structure and seismicity of West Antarctica using recently deployed seismic networks. To study the crustal thickness of West Antarctica's constituent tectonic blocks, I model crustal shear-wave velocity structure using a joint inversion of receiver functions and Rayleigh wave phase velocity dispersion. I model crustal thickness from 30-38 km in the Antarctic Peninsula, and 38-40 km in the Haag-Ellsworth Whitmore block. Within the West Antarctic Rift System (WARS), I model a crustal thickness range of 18-28 km, and show that the thinnest crust is in the vicinity of the Byrd Subglacial Basin and Bentley Subglacial Trench. I find that the thin WARS crust extends towards the Pine Island Rift, suggesting that the northern boundary of the WARS lies in this region, ~200 km north of its previously accepted position. I additionally forward model high frequency receiver functions to assess if any thick, low velocity subglacial sediment accumulations are present, finding a 0.1-0.8 km thick layer at 10 stations. Such subglacial sediment could provide a source region for the soft basal till that acts to accelerate ice flow. To investigate the active processes occurring in West Antarctica I study local seismicity recorded from 2015-2018. I locate 86 events during this period, of which I categorise 32 as `likely tectonic', 17 as `likely ice-quakes' and 37 as `possible ice-quakes'. The distribution of events correlates well with modelled regions of high uplift resulting from Glacial Isostatic Adjustment. I also investigate styles and patterns of cryoseismicity produced by Pine Island Glacier, recorded at nearby stations from 2016-2018. High frequency (1-5 Hz) tremor produced by iceberg calving was detected at stations >250 km away from the glacier's calving front, whilst smaller repetitive seismicity produced by the destruction of the glacier's ice melange is also located.

Volcanoes and glaciers can both pose a significant threat to life and property. Volcanoes can erupt suddenly, without warning, causing injury, death and damage to property. Glaciers generally present a hazard over longer timescales, melting or sliding into the oceans and contributing to sea-level rise. The movement of melt at volcanoes, and ice at and near the Earth's poles, can be investigated using microseismicity, emitted when these fluids and bodies release kinetic energy as they move. I use this microseismicity to study: melt moving from the deep crust and feeding Bardarbunga volcano, Iceland, before and after an eruption; icequakes at the bed of glaciers and ice sheets to study and constrain the physics of glacial sliding; and surface icequakes caused by crevassing, to see whether or not the crevasses observed are induced by hydrofracture. I use a combination of seismic observations and simple physical models to investigate these fundamental geophysical processes. Understanding the magmatic plumbing of a volcano is important for attempting to improve eruption forecasting. I analyse microseismicity before and after the Bardarbunga volcanic eruption, the largest eruption in Iceland in 230 years, to study possible pathways of melt from the deep crust to the shallow melt storage region. The seismicity and earthquake source mechanisms suggest that melt travels along a pathway that is approximately vertical, and laterally offset from the main shallow melt storage region by 12 km. However, it is also likely that an aseismic melt feed exists directly under Bardarbunga that we do not observe. These observations imply that volcanoes can be fed from depth, with lateral offsets of 10s kms, and that it is not adequate to monitor such volcanoes using seismicity alone. One critical process for constraining sea-level rise projections is glacial sliding. I describe how to detect and locate icequakes that originate at the bed of glaciers, which can be used to study glacial sliding. I analyse icequake source mechanisms for several glacial settings, comprising a range of spatial scales, in an attempt to unify the theory of stick-slip icequake failure, what it can tell us about glacial slip, and how such icequakes can be used to provide the first remotely measured values of bed shear modulus. The method used to remotely measure bed shear modulus will help constrain ice dynamics models and inform future passive cryoseismology studies of the Earth's ice sheets. Another poorly understood process is surface crevassing. Again, I analyse the source mechanisms of surface crevassing icequakes and show that they are tensile cracks, opening in the shallow subsurface. I present a novel method of obtaining depth estimates for these crevasse icequakes. Deriving crevasse depth is important since crevasse depth is usually limited by the stress distribution within the ice column, unless they are filled with water that can allow deeper propagation via hydrofracture. This mechanism is a possible pathway of meltwater propagating from surface to bed, lubricating the bed and exacerbating the movement of ice into the ocean. The same mechanism is also important for understanding how ice shelves, such as the Larson B ice shelf, break up, releasing onshore ice to flow unperturbed into the ocean. Such observations of crevasse fracture are therefore of great relevance for understanding key instabilities that could contribute to future sea-level changes. My results help inform our understanding of the aforementioned processes, which are critical for forecasting volcanic eruptions better, and informing and constraining ice dynamics models used for sea-level rise projections.