Search

Seismic surveys are widely used to characterise the properties of glaciers, their basal material and conditions, and ice dynamics. The emerging technology of Distributed Acoustic Sensing (DAS) uses fibre optic cables as seismic sensors, allowing observations to be made at higher spatial resolution than possible using traditional geophone deployments. Passive DAS surveys generate large data volumes from which the rate of occurrence and failure mechanism of ice quakes can be constrained, but such large datasets are computationally expensive and time consuming to analyse. Machine learning tools can provide an effective means of automatically identifying seismic events within the data set, avoiding a bottleneck in the data analysis process.Here, we present a novel approach to machine learning for a borehole-deployed DAS system on Store Glacier, West Greenland. Data were acquired in July 2019, as part of the RESPONDER project, using a Silixa iDAS interrogator and a BRUsens fibre optic cable installed in a 1043 m-deep borehole. The data set includes controlled-source vertical seismic profiles (VSPs) and a 3-day passive record of cryoseismicity. To identify seismic events in this record, we used a convolutional neural network (CNN). A CNN is a deep learning algorithm and a powerful classification tool, widely applied to the analysis of images and time series data, i.e. to recognise seismic phases for long-range earthquake detection.For the Store Glacier data set, a CNN was trained on hand-labelled, uniformly-sized time-windows of data, focusing initially on the high-signal-to-noise-ratio seismic arrivals in the VSPs. The trained CNN achieved an accuracy of 90% in recognising seismic energy in new windows. However, the computational time taken for training proved impractical. Training a CNN instead to identify events in the frequency-wavenumber (f-k) domain both reduced the size of each data sample by a factor of 340, yet still provided accurate classification. This decrease in input data volume yields a dramatic decrease in the time required for detection. The CNN required only 1.2 s, with an additional 5.6 s to implement the f-k transform, to process 30 s of data, compared with 129 s to process the same data in the time domain. This suggests that f-k approaches have potential for real-time DAS applications.Continuing analysis will assess the temporal distribution of passively recorded seismicity over the 3 days of data. Beyond this current phase of work, estimated source locations and focal mechanisms of detected events could be used to provide information on basal conditions, internal deformation and crevasse formation. These new seismic observations will help further constrain the ice dynamics and hydrological properties of Store Glacier that have been observed in previous studies of the area.The efficiency of training a CNN for event identification in the f-k domain allows detailed insight to be made into the origins and style of glacier seismicity, facilitating further development to passive DAS instrumentation and its applications.

Traditionally interpretations assume that as magma-rich rift settings mature, the magmatism accommodates greater amounts of extension at the expense of mechanical deformation. However, the importance of faulting in the final stages of magma-rich rifting remains poorly constrained, with the data (e.g. structural geological mapping, seismic reflection and borehole data) from rifts near to break-up a rarity. The Danakil Depression (Northern Afar), is undergoing the final stages of continental break-up, thus providing the ideal natural laboratory to conduct high resolution, quantitative analysis on the architecture, extension and subsidence facilitated by faulting in an active rift setting before seafloor spreading initiates. >500 rift axis faults were identified using remote sensing data (satellite imagery, DEMs), with quantitative analysis showing an increase in fault density, length and connectivity away from magmatic segments. Kinematic and earthquake focal mechanism data demonstrate a transition from transtensional opening in the northern and central sub-regions of the rift to oblique opening in the southern Giulietti Plain and Tat-Ali sub-regions of the Danakil Depression. The oblique opening is attributed to the along-axis step between the Erta-Ale and Harak sub-regions. Integration of seismic reflection and borehole data with the mapped faults shows that extension is primarily accommodated by magmatism within the rift center, with faulting more significant towards the ends of the rift. ~30% of crustal extension is accommodated by axial faulting in areas of low magmatism, highlighting the importance of faulting even in the final stages of magma-rich rifting. Comparing our findings with spreading ridge morphology and structure, which is relevant due to the rift maturity and extensive magmatism present, we conclude that the Danakil Depression is in a transitional stage between continental rifting and seafloor spreading. Spatial changes in the importance of faulting and magmatism in accommodating extension, alongside rift morphology, resemble the relationships observed along spreading ridges. From our observations we have shown that axial faulting still plays a vital role in the final stages of break-up despite the increased importance of magmatism.

When modelling ice sheet and glacier dynamics, a consideration of basal conditions is essential. Bed topography, hydrology and materials provide important controls on ice flow; however, the materials underlying large sections of the polar ice sheets are unknown. Seismic amplitude-versus-offset (AVO) analysis provides a means of inferring glacier bed properties, namely acoustic impedance and Poisson's ratio, by measuring bed reflectivity as a function of incidence angle.However, existing methods of applying AVO to glaciology only consider the compressional-wave component of the wavefield and solutions suffer from non-uniqueness. This can be addressed using multi-component seismic datasets, in which a strong converted-wave arrival (downgoing compressional-wave energy converted to shear-wave energy upon reflection at the glacier bed) is often present. We present a method of jointly inverting compressional (PP) and converted-wave (PS) seismic data to improve constraint of glacier bed properties.Using synthetic data, we demonstrate that for typical survey geometries, joint inversion of PP- and PS-wave AVO data delivers better-constrained bed acoustic impedance and Poisson’s ratio estimates compared with PP-only inversion. Furthermore, joint inversion can produce comparably constrained results to PP inversion when using input data with a smaller range of incidence angles/offsets (0-30 degree incidence for joint inversion, versus 0-60 degrees for PP- only). This could simplify future field acquisitions on very thick ice, where obtaining data at large incidence angles is difficult.Joint AVO inversion therefore has the potential to improve identification of glacier bed materials and simplify field acquisitions of glacial AVO data. We also present preliminary results from Korff Ice Rise, West Antarctica, where better constraints on bed conditions can help improve our knowledge of ice sheet history in the Weddell Sea sector. Routine measurements of this kind will help constrain ice-sheet model inputs and reduce uncertainty in predictions of sea-level rise.

Subglacial bedforms such as mega-scale glacial lineations and drumlins are commonly thought to form during active ice flow. They are often present in deglaciated areas with various elongation ratios, consisting of different materials , information which led to the development of different formation theories. However, these exposed examples were formed not only by subglacial processes during glaciation but also altered by processes during and after deglaciation. Here, we analyse in-situ properties and topography beneath Rutford Ice Steam, a fast flowing ice stream in West Antarctica to evaluate current theoretical models of bedform formation. We present a combination of seismic and radar data, including high-resolution 3D radar topography covering the upstream end of a bedform. Data acquisition and processing of the high-resolution 3D radar dataset result in a horizontal resolution of 24 m along- and across-track and a vertical resolution of 12 m. Using seismic acoustic impedance and calibrated radar reflectivity subglacial properties of the bedforms as well as the surrounding area are identified.A depression around the upstream end of a 360 m wide, 50 m high and more than 13 km long bedform was observed for the first time analysing the high-resolution 3D radar topography. The depression consists of a deepening up to 45 m deep and 360 m wide and is seen to extend around 10.5 km downstream. Radar reflectivity reveals that the material the depression is excavated into at least partly consists of low porosity material. Radar reflectivity and seismic acoustic impedance along the bedform imply a stiffer upstream end which softens along flow. The subglacial topography and properties give evidence that the bedform and the depression are formed by a combination of erosional and depositional processes. Both processes are likely interlinked, as implied by the comparable volume of the moat and the bedform at the upstream end of the bedform. Based on these observations we support or reject common bedform formation theories beneath Rutford Ice Stream.

Comprehensive descriptions of the seismic properties of glaciers and ice masses require that both compressional (P-) and shear (S-) wave components are considered. Among these properties is the seismic attenuation, expressed by the Quality Factor (Q). Q is valuable for two reasons: first, to correct measurements of seismic amplitude for wavelet propagation effects, as in reflection amplitude-versus-angle (AVA) studies. Second, Q is an indicator of ice properties such as temperature and impurity content, and laboratory/field studies of soils and geological materials suggests that the ratio of the compressional- and shear-wave quality factors, Qp/Qs, may indicate fluid saturation (particularly when considered jointly with the velocity ratio Vp/Vs). Thus, a measurement of Qp/Qs could usefully inform the hydrological structure of the firn and indicate variations in the density of the firn column. Despite its importance, few studies appear to have measured Qp in firn columns and none appear to have measured Qs in firn. Doing so for either compressional- or shear-wave arrivals is challenging, due to the ray paths followed by the diving wave first arrivals and their accurate representation in attenuation measurement methods. In preparation for an AVA study of bed properties at Korff Ice Rise, West Antarctica, we have used spectra of diving wave first arrivals and a modified spectral-ratio method to measure Qp and Qs as a function of depth in the firn column. Shot gathers with vertically oriented geophones at offsets of 2.5 - 1000m were used to measure Qp. For detecting the shear component, the geophones were oriented horizontally; in this configuration, diving and reflected shear phases were recorded with high signal-to-noise ratios. The variation of Q with depth is represented as discrete constant-Q layers with thicknesses between 6 and 27 m. Qp shows progressive increases in depth from 21 ± 3 in the uppermost 20 m (where Vp < 3000 m/s), to 246 ± 30 between 74 and 80 m depth (3750 m/s < Vp < 3770 m/s). Qs increases from 14 ± 4 in the uppermost 20m, to 80 ± 6 between 80 and 90m depth. The ratio Qp/Qs varies throughout the depths measured, from Qp/Qs ~ 1.5 at the surface, to Qp/Qs ~ 3 at 80 m. This is broadly consistent with previously quoted values, but the variation may imply that Qp/Qs is influenced by firn structure.Similar measurements at a variety of sites could help to inform a relationship between Qp, Qs and firn properties. In the immediate future, the measurement of Q in the firn will aid measurements of bed reflectivity, and help to determine the material properties of the ice-bed interface.

Peatlands have long been recognized as providing a wide range of ecosystem services valuable to humans. In recent decades their role in the global climate and particularly their importance in long-term carbon sequestration has come into focus. Peatlands and peat basins are an important carbon store globally, and are estimated to cover nearly 25% of the Scottish landscape: they constitute a significant carbon stock, but being able to accurately estimate the volume of peat stored in coastal basins, both locally and regionally, remains a time-consuming process. Traditional methods of investigating peat depth and volume involved the measurement of peat to depth of contact with a mineral horizon, such as sand. This process is conducted with a peat depth probe or corer, with the spatial density of measurements varying significantly with basin size. Volumetric assessments based on such measurements therefore require interpolation between control points, leading to unquantifiable errors particularly if the base of peat has significant and unrecorded topography. Geophysical methods, in particular the 3D application of ground-penetrating (GPR), offer a promising solution to improve the accuracy in basin volumetrics.In this paper, a 3D dataset of 100 MHz GPR data was acquired with a Mala Geosciences Rough Terrain system over a buried Holocene coastal environment near Arisaig, northwest Scotland. 3D surveying involves the acquisition of a suite of parallel GPR profiles, with a small profile separation to capture the full variability of subsurface structure. For this site, a profile was acquired every 0.5 m, over an area of 62 x 32 m. The site is also sampled by 39 boreholes, which record the base of peat between 1-3.2 m depth and indicate a peat volume of 3720 m3. By revealing the true topography of the base of the basin, the GPR data suggest that the borehole-derived volume is overestimated by almost 50%, and instead predict a basin volume of 2529 ± 200 m3. Of this, 2064 ± 200 m3 is classified as organic peat (81.6%) and the remaining 465 ± 200 m3 is marine clay (18.4%). The principal source of error in this estimate is in the constraint of the GPR velocity, required to convert the time-axis of the GPR dataset to depth. This was measured at 0.034 m/ns ± 8%.The acquisition of 3D GPR data is nonetheless time-consuming and requires precise positional control to locate the GPR antennas and avoid misinterpretation. Nonetheless, sufficient topographic information is captured even if the acquisition had recorded only every 5th GPR profile: for this downsampled dataset, the estimated basin volume is 2490 m3 ± 200 m3 (a difference of only 2.5% from the full 3D dataset). 3D survey methods, therefore, give confidence to a volumetric estimate, but the need for full-resolution 3D sampling can likely be relaxed. However, GPR surveys reveal subsurface variability that would be difficult to reconstruct from a sparse set of borehole observations. Nonetheless, some amount of borehole control is invaluable for validating the GPR data and providing ground-truth control of subsurface structure.

Glacier forelands contain valuable information on past glacier dynamics and associated climatic conditions, particularly at small mountain glaciers where responses to climate change are rapid. To maximize the potential of glacial landforms as palaeoclimate indicators, a thorough understanding of the controls on landform genesis and subsequent evolution is required. Traditionally, such landforms have been studied using glacial geological techniques such as sedimentary logging. While these provide valuable in situ information they have numerous limitations, namely poor availability and spatial extent of exposures. Near-surface geophysics provides an efficient and non-invasive means of studying subsurface conditions in numerous sedimentary settings, offering spatially extensive information on substrate material properties and architecture. However, the logistically challenging terrain, remote location and complex structure of proglacial environments has limited the development of geophysical techniques for studying the internal architecture of glacial landforms.Here, we explore the application of three geophysical methods to investigate proglacial substrates: ground penetrating radar (GPR), seismic refraction and multi-channel analysis of surface waves (MASW). Three sites with contrasting sediment properties were surveyed at the foreland of Midtdalsbreen glacier in southern Norway; (a) a 100 m2 area of glaciotectonised sandy sediments, (b) a ~2 m high lateral moraine ridge containing stratified silts, sands, and gravel and (c) a terminal moraine ridge with a peak crest height of ~5 m and an open blockwork of cobbles and boulders at its surface. At all sites, we deployed 25 MHz and 100 MHz GPR antennas and undertook seismic surveys with 50−75 m long geophone spreads and a sledge-hammer source to sample to target depths of around 10−15 m. Through comparing the results from sites (a) to (c), we assess the capabilities and limitations of each of the aforementioned techniques for proglacial substrate imaging and characterisation, we analyse how their performances vary across these settings and outline factors that contribute to a successful geophysical investigation. The ease of analysis and achievable investigation depths of the geophysical data and the applicability of seismic interpretation methods varied considerably depending on the surface terrain and structural complexity of the site. Our results show how the combination of GPR and seismic data can assist with the internal characterisation of glacial moraines when a relatively simple subsurface structure is present. However, basic seismic inversions likely lack the sophistication to resolve seismic structure in all but the simplest of layered models. We offer suggestions on how to optimise field time in more complex settings, where more sophisticated seismic inversion algorithms (e.g. tomography) or 3-D GPR surveys could be better-suited.Our experience should help advance the use of geophysics in proglacial studies. It should serve as a guide for future survey planning, and help avoid typical pitfalls such that field time can be optimised. It is hoped that geophysical survey methods will play an increasing role in the understanding of proglacial sedimentary landforms and their associated palaeoenvironments.

Distributed acoustic sensing (DAS) involves detecting seismic energy from the deformation of a length of optical fibre cable, offers considerable potential in the high-resolution monitoring of glacier systems. Subglacial conditions and sediment properties exert a strong control on the basal sliding rate of glaciers, but identifying the connectivity of drainage pathways and their hydraulic conductivity remains poorly understood. This is due in part to the limitations of instrumental methods to monitor these processes accurately, whether by locating cryoseismic emissions in passive seismic records or actively imaging the subglacial environment in seismic reflection surveys. Here, we explore the application of a borehole survey geometry for constraining the thickness and distribution of subglacial sediment deposits around a DAS installation on Greenland’s Store Glacier.Store Glacier is a fast-moving outlet of the Greenland Ice Sheet. The instrumented borehole is drilled near the centre of a drained supraglacial meltwater lake, 28 km upstream of the Store Glacier terminus, and within 100 m of an active moulin, representing a continuous supply of water to the glacier bed. The borehole, which terminates at the glacier bed at a depth of 1043 m depth, is instrumented throughout its length with Solifos BruSENS fibre-optic cable, and monitored with a Silixa iDASTM interrogator. A suite of ~30 vertical seismic profiles (VSPs) was recorded at various azimuths and offsets (up to 500 m) from the borehole, using a 7 kg sledgehammer source. Initial analyses of VSP data implied a 20 [+17, -2] m thickness of sediment immediately beneath the borehole. These analyses are refined by considering the full suite of VSP data, to map spatial variations in the thickness of subglacial sediment layers. This is undertaken using an iterative ray-tracing scheme, which seeks to minimise the differences in the arrival-time of direct seismic energy and subglacial reflections received at various depths in the borehole. Englacial compressional (P-) wave velocities are measured from cross-correlating direct arrivals (= 3700 ± 75 m/s in the upper 800 m of the glacier, 4000 ± 75 m/s between 880-950 m, 3730 ± 75 m/s through basal ice). For the subglacial sediment, we use a P-wave velocity of 1839 m/s, consistent with a value constrained in nearby surface seismic reflection data. To improve the definition of subglacial reflections and the constraint of their arrival times, data are first enhanced using frequency-wavenumber filtering.Our approach suggests that sediment thickness is ~30 m directly beneath the borehole, potentially thinning by 10 m approximately 75 m further south. In reality, the seismic velocity through the sediment layer is unconstrained, but travel-time variations are themselves indicative of changes in either P-wave velocity and/or sediment thickness. Our work further highlights the interpretative potential of borehole DAS approaches, in support of conventional surface-based seismic analysis.

Seismic surveys are widely used to study the properties of glaciers, basal material and conditions, ice temperature and crystal orientation fabric. The emerging technology of Distributed Acoustic Sensing (DAS) uses fibre optic cables as pseudo-seismic receivers,reconstructing seismic measurements at a higher spatial and temporal resolution than is possible using traditional geophone deployments. DAS generates large volumes of data, especially in passive mode, which can be costly in time and cumbersome to analyse. Machine learning tools provide an effective means of automatically identifying events within these records, avoiding a bottleneck in the data analysis process. Here we present initial trials of machine learning for a borehole-deployed DAS system on Store Glacier, West Greenland. Data were acquired in July 2019, using a Silixa iDAS interrogator and a BRUsens fibre optic cable installed in a 1043 m-deep borehole. The interrogator sampled at 4000 Hz, recording both controlled-source Vertical Seismic Profiles (VSPs), made with hammer-and-plate source, and a 3-day passive record of cryoseismicity.We used a Convolutional Neural Network (CNN) to identify seismic events within the seismic record. A CNN is a deep learning algorithm that uses a series of convolutional filters to extract features from a 2-dimensional matrix of values. These features are then used to train a modelthat can recognise objects or patterns within the dataset. CNNs are a powerful classification tool, widely applied to the analysis of both images and time series data. Previous research has demonstrated the ability of CNNs to recognise seismic phases in time series data for long-rangeearthquake detection, even when the phases are masked by a low signal-to-noise ratio. For the Store Glacier data, initial results were obtained using a CNN trained on hand-labelled, uniformly-sized windows. At present, these windows have been targeted around high signal-to-noise ratio seismic events in the controlled-source VSPs only. Once trained, the CNN achieved accuracy of 90% in recognising whether new windows contained coherent seismicenergy.The next phase of analysis will be to assess the performance of the CNN when trained and tested on large passive DAS datasets. The method will then be used for the identification and flagging of seismic events within the passive record for interpretation and event location. The identified signals will be used to provide information on the glacier’s seismic velocity structure, ice temperature and ice crystal orientation fabric and anisotropy. Basal reflections were identified and will be used to provide information on subglacial material properties and conditions of Store Glacier. The efficiency of the CNN allows detailed insight to be made into the origins and style of glacier seismicity, facilitating further advantages of passive DAS instrumentation.

Full Waveform Inversion (FWI) is a well-established seismic imaging technique used in the exploration industry to acquire high resolution, high precision velocity models of the subsurface from seismic data. Although FWI is computationally expensive and requires customized data acquisition, the technique has the potential to improve subsurface glaciological imaging.Firn is formed as an intermediate material (of density ~400 – 810 kg m-3) as snow is compressed into ice (~810 – 917 kg m-3). Variations in surface conditions and periods of surface melting commonly lead to the presence of discrete layers and lenses of refrozen (‘infiltration’) ice within the firn column; layers that can be from millimetres to several tens of metres thick. Therefore, firn characteristics provide a tool for reconstructing climate conditions relating to the amount of snow accumulation, melt, temperature conditions and subsequent snow preservation. Given the complexity of these relationships, it has not been possible to develop a theoretical model that predicts accurately variations in firn properties or density with depth. Consequently, seismic techniques, which are logistically less demanding than extracting firn cores, are typically used to reconstruct these physical properties of the firn column.Firn seismic velocity is often derived from seismic data using the Herglotz-Wiechert (HW) inversion. A velocity trend would be expected to increase from ~400 m s-1 in snow through to ~3,800 m s-1 in ice. Thus, the presence of infiltration ice within the firn column results in anomalously high velocity intervals at shallow depths. HW inversion can be limited by the accuracy of first-break picking (specifically in the near offset, where a small error in the travel time pick gives the greatest variability to the HW velocity output), and it cannot recover the velocity inversion below a refrozen ice layer without elastodynamic redatumming. Importantly, FWI has the capacity to mitigate issues such as these, and thereby potentially offers a new standard for glaciological seismic modelling.Using seismic datasets obtained from Pine Island Glacier, Antarctica, and synthetic data that simulate firn columns that include substantial thicknesses of infiltration ice (‘ice slabs’, up to 100 m thick and from 5-80 m deep), we show how FWI improves on current seismic techniques in terms of identifying the velocity variations associated with both included ice layers and the firn underlying them. We present a best practice methodology for the use of FWI with glaciological data, including (i) the extraction of a source wavelet from the data for the use with modelling, (ii) the steps needed to ensure a consistent waveform, (iii) the appropriate offset-to-depth ratio, and (iv) the requirement of a constraint for the uppermost part of the velocity model. Finally, we evaluate the robustness of the FWI approach by comparing it with well-established HW methods for building velocity models.

There are numerous theoretical descriptions of the subglacial conditions (water availability, subglacial geology, flow dynamics) required for the formation of subglacial lineations, such as mega-scale glacial lineations and drumlins, that are known to be indicative of fast ice flow. Traditionally, mapping in de-glaciated areas, both onshore and offshore, has been undertaken using bathymetric maps, satellite data and field observations; here, lineations currently beneath the Rutford Ice Stream (West Antarctica) have been mapped using ground-penetrating radar (GPR) and seismic methods.The Rutford Ice Stream is more than 2 km thick, of which 1.4 km are located below sea level. The ice surface speed at the grounding line is >1 m per day, and satellite observations indicate a stable ice flow over the past 30 years. The ice-bed interface is assumed to be at the pressure-melting point, while the bed can be divided into a region of soft, deforming sediment, and one of stiff, non-deforming, sediment. Long, elongated lineations, up to ~14 km, up to 150 m high, and 50-500 m wide, are found aligned in the ice-flow direction in the area of the soft sediment, within which the deposition of a drumlin was observed over a period of To study the detailed architecture of the lineations, 3D grids of GPR data were acquired during the Antarctic Summer Season 2017/18, enabling 3D-processing and imaging of lineations. Using this unique dataset, in conjunction with previous publications plus data from the paleo record, we hope to better understand the possible mechanisms of formation of subglacial lineations as well as subglacial conditions at the Rutford Ice Stream.

Warming of the polar ice sheets causes changes in the hydrological regime of surface layers of firn and ice. Surface meltwater may undergo perennial storage of liquid water above the firn-ice transition, which could slow sea level rise or cause sudden release events, when storage capacity is reached. Firn aquifers have been commonly observed within the lower percolation zone of the southeastern Greenland ice sheet during the past decade, and more recently, across some Antarctic ice shelves. Knowledge of the geographic extent and fractional liquid water content (and storage) of such aquifers will enable a better understanding of their effects on the sub- and en-glacial hydrologic system and is crucial for accurate predictions of the contribution of meltwater discharge to global sea level rise.Quantitative geophysical analysis from surface observations can be used to infer hydrological properties of the firn and ice without time intensive direct drilling, providing an efficient spatial distribution of properties along with an estimate of their uncertainty. Furthermore, by combining multiple types of geophysical observations, joint inversions allow ambiguities of one methodology to be mitigated by resolution in the other.Here, we demonstrate that this joint approach is a powerful complement to the conventional geophysical analysis of firn aquifers, by combining seismic, ground penetrating radar and borehole data to characterise aquifer properties, using the ‘MuLTI’ algorithm. In particular, we incorporate seismic shear wave velocities (Vs), derived from surface (Rayleigh) waves offering a promising means of distinguishing zones containing liquid water, into independent compressional wave velocity, density, and radar soundings of the water table. We find Vs decreases from 1600 m/s in the unsaturated firn above the water table at around 15 m depth, to 800 m/s through saturated ‘clean’ firn aquifer at around 25 m depth. However, at lower elevations, Vs increases to 1250 m/s through thicker, older firn aquifer where there are many ice lenses, which are interpreted to correspond with episodes of refreezing aquifer water as the system has evolved through time. With access to multiple seismic wave velocities (compressional and shear) through the aquifer, a more accurate estimate of liquid water content can be derived. Thus, the application of the MuLTI algorithm to this pressing new problem can deliver an accurate assessment of firn aquifer properties, and provide clear uncertainty limits which will be valuable for predictive modelling.

Material anisotropy within a glacier both influences and is influenced by its internal flow regime. Anisotropy can be measured from surface seismic recordings, using either active sources or natural seismic emissions. In the past decade, Distributed Acoustic Sensing (DAS) has emerged as a new, and potentially transformative, seismic acquisition technology, involving determining seismic responses from the deformation of optical fibres. Although DAS has shown great potential within engineering and resources sectors, it has not yet been widely deployed in studies of glaciers and ice masses.Here, we present results from a glaciological deployment of a DAS system. In July 2019, a Solifos BRUsens fibre optic cable was installed in a 1050 m borehole drilled on Store Glacier in West Greenland. Vertical seismic profiles (VSPs) were recorded using a Silixa iDAS interrogation unit, with seismic energy generated with a 7 kg sledgehammer striking a polyethene (UHMWPE) impact plate. A three-day sequence of zero-offset VSPs (with the source located ~1 m from the borehole top) were recorded to monitor the freezing of the cable, combined with offset-VSPs in along- and cross-flow directions, and radially at 300 m offset.P-wave energy (frequency ~200 Hz) is detectable through the whole ice thickness, sampled at 1 m depth increments. The zero-offset reflectivity of the glacier bed is low, but reflections are detected from the apparent base of a subglacial sediment layer. S-wave energy is also detectable in the offset VSP records. The zero-offset VSPs show a mean vertical P-wave velocity of 3800 ± 140 m/s for the upper 800 m of the glacier, rising to 4080 ± 140 m/s between 900-950 m. In the deepest 50 m, velocity reduces to 3890 ± 80 m/s. This variation in vertical velocity is consistent with the development of an anisotropic ice fabric in the lowermost 10% of the glacier. The full dataset also contains natural seismic emissions, highlighting the potential of DAS as both an active and passive seismic monitoring tool.DAS offers transformative potential for understanding the seismic properties of glaciers and ice sheets. The simplicity of the typical VSP geometry makes the interpretation of seismic travel-times less vulnerable to approximations, and thus the derivation of seismic properties more robust, than in conventional surface seismic surveys. As an addition, DAS facilitates VSP recording with unprecedented vertical and temporal resolution. Furthermore, the sensitivity of the optical-fibre to both P- and S-wave particle motion means that a comprehensive suite of acoustic and elastic properties can be inferred.

The transformation of snow into ice is a fundamental process in glaciology. The yearly accumulation of fresh snowfall increases the overburden pressure, changing the snow’s properties such that it transitions into firn and pure glacier ice thereafter. Additionally, periods of melt and variations in subsurface and surface conditions can lead to the presence of ice layers and firn aquifers within the firn column. Therefore, firn characteristics provide a tool for evaluating past and present climate conditions relating to the amount of snow accumulation, melt, temperature conditions and the subsequent preservation of the snow. Due to the importance of relationships between firn and other glaciological processes (e.g., settling, sublimation, recrystallization and other deformation processes) it has not been possible to develop a theoretically-based model which accurately predicts firn properties with depth. Therefore, methods of measuring firn are either intrusive or rely on (potentially unreliable) empirical conversions. Full Waveform Inversion (FWI) may offer a new standard for glaciological seismic modelling, mitigating issues within current seismic modelling techniques and paving the way for the recovery of elastic properties, including density. Constraining firn properties also leads to improved corrections for deeper seismic responses, e.g. glacier bed reflectivity. Using seismic datasets obtained from Norway’s Hardangerjøkulen Ice Cap (60.47°N, 7.49°W) along with varying synthetic firn column scenarios (introducing the presence of ice lenses and firn aquifers), we show how FWI can mitigate the dependence on intrusive techniques and empirical relationships. Furthermore, we compare the robustness of the FWI approaches versus traditional glaciological approaches to velocity model building (Herglotz-Wiechert inversion).

During the last two decades, ground-penetrating radar (GPR) methods, have grown in popularity for acquiring high-resolution images of the stratigraphy, internal structure and wider context of geomorphology, as well as the reconstruction and evolution of buried landscapes. GPR offers centimetre-scale resolution of the subsurface, allowing 3D visualization of abrupt changes in palaeo-environments. Although often complemented by core data, GPR interpretations can also be extended beyond regions of ground-truth control. However, for all these advantages, GPR data interpretation can be non-intuitive and ambiguous, with the technique seldom giving images that immediately resemble the expected subsurface geometry. Interpretation can be made yet more onerous when handling the large 3D data volumes that are commonly available with modern GPR technology.In this paper, we outline the development of a semi-automated GPR feature-extraction tool, based on the image processing techniques ‘Edge Detection’ and ‘Thresholding’. Developed initially for medical image analysis, we investigate them as a means of assisting the analysis of GPR data for subsurface geomorphic features. Given that GPR reflectivity can be related to changes in lithology and/or pore fluids, the structure and extent of subsurface depositional environments can be efficiently estimated using these algorithms. When benchmarked against representative core control, the 3D architecture of the palaeo-landscape can be reconstructed from the GPR dataset.We present a 500 MHz GPR dataset collected over a buried Holocene coastal dune system in Llanbedr, Gwynedd, North Wales, which has since been reclaimed for use as an airfield. Core data, with maximum depth 2 m, suggest rapid vertical changes from sand to silty-organic units, and GPR profiles suggest that similar lateral complexity is likely across the dataset. By applying thresholding methods to top-down depth slices, the environment is effectively characterised. Furthermore, automatic extraction of the local reflection power with depth yields a strong correlation with the vertical variation of organic content. Similar analyses away from core control could, therefore, deliver a powerful proxy for parameters derived from invasive core logging.

Following a southward progression of ice-shelf disintegration along the Antarctic Peninsula, Larsen C Ice Shelf is the focus of ongoing investigation regarding its future stability. The ice shelf is known to be experience surface melt, and commonly features surface meltwater ponds. Here, we use a flowline model and a firn density model to date and interpret observations of melt-affected ice layers found within five 90 m boreholes distributed across the ice shelf. We find that units of ice within the boreholes, which have densities exceeding those expected under normal compaction metamorphism, correspond to two climatic warm periods within the last 300 years on the Antarctic Peninsula. The more recent warm period, from the 1960s onwards, has generated distinct sections of dense ice in two boreholes in Cabinet Inlet, close to the Antarctic Peninsula mountains – a region currently affected by föhn winds. Previous work has classified these layers as refrozen pond ice, requiring large quantities of mobile liquid water to form. Our flowline model shows that, whilst preconditioning of the ice began in the late 1960s, it was probably not until the early 1990s that twentieth-century ponding began. The earlier warm period occurred during the 18th century and resulted in two additional sections of anomalously dense ice deep within the boreholes. The first, in one of the Cabinet Inlet boreholes, consists of ice characteristic of refrozen ponds and must have formed in an area currently featuring ponding. The second, in a mid-shelf borehole, formed at the same time in an area which now experiences significant annual melt. Further south on the shelf, the boreholes sample ice that is of an equivalent age but which does not exhibit the same degree of melt influence. This west–east and north–south gradient in past melt distribution resembles current spatial patterns of surface melt intensity. Using flowlines to trace the advection and submergence of continental ice identified in boreholes, we demonstrate that, even by the time the ice reaches the calving front, only the upper 40 to 50 % of the shelf is composed of meteoric ice accumulated on the shelf. This vertical composition implies that basal crevasses must be confined within continental and/or basally accreted ice, and therefore will be unaffected by current climate-induced firn compaction.

In recent decades, rapid ice-shelf disintegration along the Antarctic Peninsula has had a global impact through enhancing outlet glacier flow, and hence sea level rise, and the freshening of Antarctic Bottom Water. Ice shelf thinning due to basal melting results from the circulation of relatively warm water in the underlying ocean cavity. However, the effect of sub-shelf circulation on future ice-shelf stability cannot be predicted accurately with computer simulations if the geometry of the ice-shelf cavity is unknown. To address this deficit for Larsen C Ice Shelf, west Antarctica, we integrate new water-column thickness measurements with existing observations. We present these new data here along with an updated bathymetry grid of the ocean cavity. Key findings include relatively deep seabed to the south-east of the Kenyon Peninsula, along the grounding line and around the key ice shelf pinning point of Bawden Ice Rise. In addition, we can confirm that the cavity’s southern trough stretches from Mobiloil Inlet to the open ocean. These areas of deep seabed will influence ocean circulation and tidal mixing, and will therefore affect the basal-melt distribution. These results will help constrain models of ice-shelf cavity circulation with the aim of improving our understanding of sub-shelf processes and their potential influence on ice shelf stability. The data set comprises all point measurements of seabed depth and a gridded data product, derived using additional measurements of both offshore seabed depth and the thickness of grounded ice. We present all new depth measurements here as well as a compilation of previously published measurements used in the gridding process. The gridded data product is included in the supplementary material. The underlying seismic data sets which were used to determine bed depth and ice thickness are available at https://doi.org/10.5285/315740B1-A7B9-4CF0-9521-86F046E33E9A (Brisbourne et al., 2019), https://doi.org/10.5285/5D63777D-B375-4791-918F-9A5527093298 (Booth, 2019), https://doi.org/10.5285/FFF8AFEE-4978-495E-9210-120872983A8D (Kulessa and Bevan, 2019) and https://doi.org/10.5285/147BAF64-B9AF-4A97-8091-26AEC0D3C0BB (Booth et al., 2019).

This study suggests that cold-ice processes may be more widespread even within temperate glacial systems, than previously assumed. We present the first direct observations of cold-ice at the snout of the temperate glacier Midtdalsbreen an outlet of the Hardangerjøkulen icefield (Norway) from 43 line-kilometres of ground penetrating radar data. Results show a 40 m-wide cold-ice zone within the majority of the glacier snout, where ice thickness is

Seismic amplitude-versus-angle (AVA) methods are a powerful means of quantifying the physical properties of subglacial material, but serious interpretative errors can arise when AVA is measured over a thinly-layered substrate. A substrate layer with a thickness less than 1/4 of the seismic wavelength, λ, is considered "thin", and reflections from its bounding interfaces superpose and appear in seismic data as a single reflection event. AVA interpretation of subglacial till can be vulnerable to such thin-layer effects, since a lodged (non-deforming) till can be overlain by a thin (metre-scale) cap of dilatant (deforming) till. We assess the potential for misinterpretation by simulating seismic data for a stratified subglacial till unit, with an upper dilatant layer between 0.1–5.0 m thick (λ / 120 to > λ / 4, with λ = 12 m). For dilatant layers less than λ / 6 thick, conventional AVA analysis yields acoustic impedance and Poisson's ratio that indicate contradictory water saturation. A thin-layer interpretation strategy is proposed, that accurately characterises the model properties of the till unit. The method is applied to example seismic AVA data from Russell Glacier, West Greenland, in which characteristics of thin-layer responses are evident. A subglacial till deposit is interpreted, having lodged till (acoustic impedance = 4.26±0.59 × 106 kg m−2 s−1) underlying a water-saturated dilatant till layer (thickness < 2 m, Poisson's ratio ~ 0.5). Since thin-layer considerations offer a greater degree of complexity in an AVA interpretation, and potentially avoid misinterpretations, they are a valuable aspect of quantitative seismic analysis, particularly for characterising till units.