Yes / The value of artificial intelligence and machine learning applications for use in heritage research is increasingly appreciated. In specific areas, notably remote sensing, datasets have increased in extent and resolution to the point that manual interpretation is problematic and the availability of skilled interpreters to undertake such work is limited. Interpretation of the geophysical datasets associated with prehistoric submerged landscapes is particularly challenging. Following the Last Glacial Maximum, sea levels rose by 120 m globally, and vast, habitable landscapes were lost to the sea. These landscapes were inaccessible until extensive remote sensing datasets were provided by the offshore energy sector. In this paper, we provide the results of a research programme centred on AI applications using data from the southern North Sea. Here, an area of c. 188,000 km2 of habitable terrestrial land was inundated between c. 20,000 BP and 7000 BP, along with the cultural heritage it contained. As part of this project, machine learning tools were applied to detect and interpret features with potential archaeological significance from shallow seismic data. The output provides a proof-of-concept model demonstrating verifiable results and the potential for a further, more complex, leveraging of AI interpretation for the study of submarine palaeolandscapes.

Using gravity, magnetic, bathymetric and seismic refraction data, I have constructed a geophysical cross-section of the central part of the northern Gulf of California. The section exhibits a crustal thickness of 18 km and features an anomalous block of high density lower basement (3.15 g/cm³) which probably resulted from rifting processes during the opening of the Gulf. The magnetization of the upper basement ranges from 0.0005 to 0.0030 emu/cm³. Three different layers of sediments are modeled, ranging from unconsolidated (1.85 g/cm³) to compacted (2.50 g/cm³). I present a deconvolution method for automated interpretation of magnetic profiles based on Werner's (1953) simplified thin-dike assumption, leading to the linearization of complex nonlinear magnetic problems. The method is expanded by the fact that the horizontal gradient of the total field caused by the edge of a thick interface body is equivalent to the total field of a thin dike. Statistical decision making and a seven point operator are used to insure good approximations of susceptibility, dip, depth, and horizontal location of the source. After using synthetic models to test the inversion method, I applied it to the Northern Gulf of California using data collected in 1984 by the Continental Margins Study Group at Oregon State University. Fault traces, computed by the deconvolution, are plotted on a map. The faulting pattern obtained is in good agreement with that proposed by other workers using other methods. The depths to the top of the faults range from 4 to 5 km in the eastern part of the Gulf, where they may be interpreted as the top of the structural basement. Deeper estimates are obtained for the western part of the Gulf. / Graduation date: 1990

Graduation date: 1988

Hydrocarbon deposits in the form of petroleum, natural gas, and natural gas hydrates occur offshore worldwide. Electromagnetic methods that measure the electrical resistivity of sediments can be used to map, assess, and monitor these resistive targets. In particular, quantitative assessment of hydrate content in marine deposits, which form within the upper few hundred meters of seafloor, is greatly facilitated by complementing conventional seismic methods with EM data. The North-East Pacific Time-series Undersea Network Experiment (NEPTUNE) is an underwater marine observatory that provides power and network connection to a host of instruments installed on the seafloor on the Cascadia Margin offshore Vancouver Island. The observatory’s aim is to provide a platform for very long-term studies in which access to data is available on a continuous basis. For this thesis project, a transient dipole-dipole time-domain EM system was constructed and deployed on the NEPTUNE network with the goal of long-term monitoring of a well-studied hydrate deposit in the area. The instrument includes a source transmitter of electrical current and individual receivers to measure small electric field variations. The instrument is powered by the NEPTUNE observatory and data can be collected remotely by connecting to the instrument through the web. Data collected so far from the instrument are consistent with a resistive structure. The best fitting model from 1D inversion is a 36 ± 3 m thick layer of 5.3 ± 0.3 Ωm resistivity, overlaying a less resistive 0.7 ± 0.1 Ωm halfspace. Average hydrate concentration, deduced with the aid of ODP-889 well-log derived Archie’s parameters, is approximately 72% of pore space in the resistive layer, consistent with the very high concentration of gas hydrates (~80%) recovered from seafloor cores. The weekly collection of data from the instrument shows that the resistive structure has changed little since monitoring began in October of 2010.

Hydrocarbon deposits in the form of petroleum, natural gas, and natural gas hydrates occur offshore worldwide. Electromagnetic methods that measure the electrical resistivity of sediments can be used to map, assess, and monitor these resistive targets. In particular, quantitative assessment of hydrate content in marine deposits, which form within the upper few hundred meters of seafloor, is greatly facilitated by complementing conventional seismic methods with EM data. The North-East Pacific Time-series Undersea Network Experiment (NEPTUNE) is an underwater marine observatory that provides power and network connection to a host of instruments installed on the seafloor on the Cascadia Margin offshore Vancouver Island. The observatory’s aim is to provide a platform for very long-term studies in which access to data is available on a continuous basis. For this thesis project, a transient dipole-dipole time-domain EM system was constructed and deployed on the NEPTUNE network with the goal of long-term monitoring of a well-studied hydrate deposit in the area. The instrument includes a source transmitter of electrical current and individual receivers to measure small electric field variations. The instrument is powered by the NEPTUNE observatory and data can be collected remotely by connecting to the instrument through the web. Data collected so far from the instrument are consistent with a resistive structure. The best fitting model from 1D inversion is a 36 ± 3 m thick layer of 5.3 ± 0.3 Ωm resistivity, overlaying a less resistive 0.7 ± 0.1 Ωm halfspace. Average hydrate concentration, deduced with the aid of ODP-889 well-log derived Archie’s parameters, is approximately 72% of pore space in the resistive layer, consistent with the very high concentration of gas hydrates (~80%) recovered from seafloor cores. The weekly collection of data from the instrument shows that the resistive structure has changed little since monitoring began in October of 2010.

This study represents the first detailed geological, geophysical and geochronological investigation of the continental shelf surrounding the Aliwal Shoal, ~5 km offshore of Scottburgh, in southern KwaZulu-Natal. Mapping of the seafloor geology using geophysics and direct observations from SCUBA diving transects were integrated with the seismic stratigraphy and constrained by new geochronological data. Four seismic stratigraphic units (A to D) were identified and interpreted with the subsequent sequence stratigraphic model consisting of four incompletely preserved stratigraphic sequences separated by three sequence boundaries (SB1 - SB3) comprising complex reworked subaerial unconformity surfaces. Sequence 1 is the deepest, subdivided by a basin-wide marine flooding surface (MFS1) into a lower Campanian (and possible Santonian) TST and an upper Maastrichtian combined regressive systems tract comprising HST/FRST deposits. SB1 follows Sequence 1 and spans most of the Tertiary representing multiple erosional events. Shelf sedimentation resumed during the Late Pliocene to early Pleistocene with deposition of Sequence 2, the shelf-edge wedge, which again was followed by erosion and non-deposition during the high frequency and amplitude Early to Middle Pleistocene sea-level fluctuations resulting in the formation of SB2. Sequence 3 consists of coast-parallel, carbonate cemented aeolianite palaeo-shoreline ridges of various ages overlying Sequence 1 and 2. Sequence 4 unconformably overlies all the earlier sequences and comprises a lower TST component displaying characteristic retrogradational stacking patterns and an upper local HST clinoform component showing progradation and downlapping. Inner and middle shelf TST units constrained between Sequence 3 ridges form thick sediment deposits showing a progression from lagoonal and lower fluvial-estuarine deposits, overlain by foreshore and shoreface sands, documenting the changing depositional environments in response to a sea-level transgression. Laterally, in the absence if Sequence 3 ridges, TST sediments comprise only a thin transgressive sand sheet. The upper HST component comprises a prograding shore-attached subaqueous-delta clinoform sediment deposit, the Mkomazi Subaqueous-Delta Clinoform (MSDC) which evolved in four stages. An initialization and progradation stage (Stage 1) (9.5 to 8.4 ka cal. B.P.) was interrupted by retrogradation (Stage 2) and backstepping of the system due to rapid sea-level rise between 8.4 to 8.2 ka cal B.P. Stage 2 backstepping of the clinoform controlled the subsequent overlying topset morphologies resulting in later stages inheriting a stepped appearance upon which shoreface-connected ridges (SCR’s) are developed. Stages 3 (8.2 to 7.5 ka cal. B.P.) and 4 (7.5 to 0 ka cal. B.P) show a change from ‘proximal’ topset aggradation to ‘distal’ foreset progradational downlap, linked to a change in the dominant sedimentary transport mechanism from aggradational alongshore to progradational cross-shore related to variations in accommodation space and the rate of sediment supply. Morphologically the MSDC is characteristic of sediment input onto a high energy storm-dominated continental shelf where oceanographic processes are responsible for its northward directed asymmetry in plan-view, for the lack of a well defined bottomset and for the re-organisation of its topset into very large SCR’s. The SCR’s are 1 - 6 m in height, spaced 500 to >1350 m apart and vary from 3 km to >8 km in length, attached on their shoreward portions to the shoreface between depths of -10 m to -15 m (average at -13 m) and traceable to depths exceeding -50 m, although the majority occur on the inner shelf between -20 m to -30 m. Several individual crests can be identified forming a giant shoreface-connected sand ridge field with a sigmoidal pattern in plan-view postulated to be a surficial expression of the subjacent retrogradational phase (MSDC Stage 2). SCR’s development occurred in two stages. Stage 1 involved deposition of sediment on the shoreface and ridge initiation during the MSDC Stage 2 retrogradational event. Sediment was reworked during sea-level rise generating clinoforms with proximal along-shore aggradation and distal across-shore progradation. This occurred during the last post-glacial sea-level rise from ca. 8.4 ka cal. B.P. SCR Stage 2 represents modern maintenance of the SCR system which is continually modified and maintained by shelf processes and consists of two physical states. State 1 considers SCR maintenance during fair-weather conditions when transverse ridge migration is dominant and driven by the north-easterly flowing counter current shelf circulation. State 2 considers SCR development during storm conditions when longitudinal ridge growth is suggested to occur as a result of storm return flows. Following the storm, the regional coast-parallel current system is restored and the fair-weather state then moulds the SCRs into a transverse bedform. Deposition on the MSDC is ongoing on a continental shelf that is still in a transgressive regime. The exposed seafloor geology comprises late Pleistocene to Holocene aeolianite and beachrock lithologies, deposited as coastal barrier and transgressive shoreface depositional systems. Extensive seafloor sampling was combined with a multi-method geochronological programme, involving the U-series, C14 and optically stimulated luminescence (OSL) to constrain the evolution of the aeolianite and beachrock complex. The Aliwal Shoal Sequence 3 ridge comprises three distinct aeolianite units (A1 to A3) which represent different types of dune morphologies deposited during the climatic and associated sea-level fluctuations of MIS 5. Units A1 and A2 deposited during the MIS 6/5e (~134 to ~127 ka cal. B.P.) transgression represent contemporaneous evolution of a coastal barrier system which consisted of two different dune forms associated with a back-barrier estuarine or lagoonal system. Unit A1 most likely originated as a longitudinal coastal dune whilst Unit A2 comprised a compound parabolic dune system that migrated into the back-barrier area across an estuary mouth/tidal inlet of the back-barrier system. The coastal barrier-dune configuration established by Unit A1 and A2 was most likely re-established during similar subsequent MIS 5 sea-level stands which during MIS 5c/b resulting in the formation of the back-barrier dune system of Unit A3. Palaeoclimatic inferences from Units A1 and A2 aeolianite wind vectors indicate a change from cooler post-glacial climates (lower Unit A1) to warmer interglacial-like conditions more similar to the present (upper Unit A1 and Unit A2). Unit A3 palaeowind vector data show variability interpreted to be related to global MIS 5c climatic instability and fluctuations. For Units A1, A2 and A3 pervasive early meteoric low-magnesium calcite (LMC) cementation followed shortly after deposition protecting the dune cores from erosion during subsequent sea-level fluctuations. Sea-spray induced vadose cementation in Units A1 and A2 may have been a key factor in stabilising dune sediment before later phreatic meteoric cementation. The final preserved Late Pleistocene depositional event in the study area was that of the storm deposit of beachrock Unit B5. Induration followed shortly after deposition by marine vadose high-magnesium calcite (HMC) cementation. Following deposition and lithification, Units A1, A2, A3 and B5 underwent a period of cement erosion associated with decementation and increased porosity due to either 1) groundwater table fluctuations related to the high frequency MIS 5 sea-level fluctuations and/or 2) carbonate solution due to complete subaerial exposure related to the overall MIS 4 - 2 sea-level depression towards the LGM lowstand. In addition to the decementation and porosity development Unit B5 also experienced inversion of the original unstable HMC cement to LMC. During MIS 4 to 2 the Aliwal shelf comprised an interfluve area which was characterised by subaerial exposure, fluvial incision of coast-parallel tributary river systems and general sediment starvation. Beachrock Units B1 to B4 were deposited in the intertidal to back-beach environments and subsequently rapidly cemented by marine phreatic carbonate cements comprising either aragonite or HMC. Unit B1 was most likely deposited at 10.8 ka cal. B.P., B2 at 10.2 ka cal. B.P, B3 at 9.8 ka cal. B.P and B4 <9.8 ka cal. B.P. thereby indicating sequential formation during the meltwater pulse 1b (MWP-1b) interval of the last deglacial sea-level rise. Unit B3 marks the change from a log-spiral bay coastal configuration established by Units B1 and B2 to a linear coastline orientation controlled by the trend of the pre-existing aeolianite units. This change in the morphology of the coastline is also documented by the shape of the underlying transgressive ravinement surface (reflector TRS, Sequence 4) which again was controlled by the subjacent sedimentary basin fill architecture and subsequent transgressive shoreline trajectory (Sequence 4). Sea-level rose at an average rate of 67 cm/100 years from B1 to B2 and 86 cm/100 years from B2 to B2 indicating an acceleration in the rate of sea-level rise supporting enhanced rates of sea-level rise during the MWP-1b interval which also seemed to have altered the coastal configuration and resulted in the closure of the southern outlet of the back-barrier estuarine system. Two cycles of initial aragonite followed by later HMC cement are tentatively linked to two marine flooding events related to different pulses of enhanced rates of sea-level rise during MWP-1b which are considered responsible for significant changes in the marine carbon reservoir ages. Comparisons of the U-series, C14 and optically stimulated luminescence (OSL) methods have shown OSL to be the most reliable method applied to dating submerged aeolianites and beachrocks. OSL not only provides the depositional age of the sediment but also does not suffer from open system behaviour, such as marine reservoir changes and contamination. Acoustic classification of the unconsolidated sediment samples resulted in the demarcation of 3 major acoustic facies, C to E, interpreted with sample analyses as quartzose shelf sand (C), reef-associated bioclastic-rich sand (D) and an unconsolidated lag and debris deposit (E). Grain size distribution patterns of the unconsolidated seafloor sediments indicate that the SCR system delivers fine and medium sand to the inner and middle shelf and imparts a general N-S trending pattern to the gravel and sand fractions. In addition grain size distributions support selective erosion of the seaward flank of the Sandridge with the remobilised sediment deposited in the Basin as low amplitude bedforms over the Facies E lag and debris pavement. The mud fraction is interpreted to be deposited by gravity settling from buoyant mud-rich plumes generated by river discharge. Integration of acoustic mapping, field observations and sample analyses indicate that the present distribution of the unconsolidated sediment is the result of a highly variable distribution of modern and palimpsest sediments which are continually redistributed and reworked by a complex pattern of bottom currents generated by the interaction of opposing oceanographic and swell driven circulation patterns. / Thesis (Ph.D.)-University of KwaZulu-Natal, Westville, 2012.

Marine geological surveys in Asköfjärden in the southern Stockholm Archipelago hasrevealed step like features in the sediments on the Baltic sea floor. The aim of this project is toanalyse the steps and possible formation processes from the survey data. The data used aretaken from the acoustic instruments multibeam echosunder and Chirp sonar sediment profiler.The multibeam reveal the seafloor topography and can detect water column features. TheChirp sonar produce sub-bottom profiles showing the sediments down to some tens of metersunderneath the sea floor. The multibeam data show multiple crescent shaped steps as well aselliptically shaped “pockmarks”. Water column data show what could be interpreted as seepsfrom the sea floor under these features. The sub bottom profile show deformation and fluidsignatures under the steps. Gas signatures and what could be other fluids are present. Theinterpretation concludes that the formation could be from either gas or groundwater seeps. Nodefinitive conclusions can be made from the data available for the present study, andapetrophysical or geochemical study of the study area cold provide further understanding ofthe formation of the steps.

Arc-continent collisions are believed to be an important mechanism for the growth of continents. Taiwan is one of the modern day examples of this process, and as such, it is an ideal natural laboratories to investigate the uncertain behavior of continental crust during collision. The obliquity of collision between the northern South China Sea (SCS) rifted margin and Luzon arc in the Manila trench subduction zone allows for glimpses into different temporal stages of collision at different spatial locations, from the mature mountain-belt in central-northern Taiwan to the 'pre-collision' rifted margin and subduction zone south of Taiwan. Recently acquired seismic reflection and wide-angle seismic refraction data document the crustal-scale structure of the mountain belt through these different stages. These data reveal a wide rifted margin near Taiwan with half-graben rift basins along the continental shelf and a broad distal margin consisting of highly-extended continental crust modified by post-rift magmatism. Magmatic features in the distal margin include sills in the post-rift sediments, intruded crust, and a high-velocity lower crustal layer that likely represents mafic magmatism. Post-rift magmatism may have been induced by thermal erosion of lithospheric mantle following breakup and the onset of seafloor spreading. Geophysical profiles across the early-stage collision offshore southern Taiwan show evidence the thin crust of the distal margin is subducting at the Manila trench and structurally underplating the growing orogenic wedge ahead of the encroaching continental shelf. Subduction of the distal margin may induce a pre-collision flexural response along the continental shelf as suggested by a recently active major rift fault and a geodynamic model of collision. The weak rift faults may be inverted during the subsequent collision with the continental shelf. These findings support a multi-phase collision model where the early growth of the mountain belt is driven in part by underplating of the accretionary prism by crustal blocks from the distal margin. The wedge is subsequently uplift and deformed during a collision with the continental shelf that involves both thin-skinned and thick-skinned structural styles. This model highlights the importance of rifting styles on mountain-building. / text

Cette thèse est une contribution à la caractérisation des fonds sous-marins par des techniques acoustiques. On s'intéresse aux fonds sédimentaires, principalement sédiments sableux. Les fonds de sables sont en effet fréquemment rencontrés par petits fonds sous nos latitudes. Les procédés existants de caractérisation acoustique des fonds visent le plus souvent à qualifier la géométrie du sol ou du sous-sol: morphologie du fond, typologie des faciès sédimentaires, identification du toit rocheux ... Toutefois, les détails du sous-sol marin (stratification et composition des sables) nous échappent le plus souvent et on a alors recours à des sondages in-situ ponctuels, coûteux et souvent difficiles à réaliser. Afin de résoudre ce problème, nous avons développé SCAMPI (Système de Caractérisation Acoustique Marine Propagation Interface). C'est un dispositif de caractérisation géoacoustique breveté qui vise justement à réduire notre myopie chronique dans les premiers mètres des sous-sols sableux immergés en calculant des profils verticaux des vitesses du son. Le système développé est typique d'un processus d'inversion basé sur des mesures distantes et indirectes (on ne touche pas le sol). Disposer de profils de vitesses pour caractériser le sous-sol est une étape nécessaire, mais insuffisante pour les applications visées. Pour ces dernières, il s’agit notamment de déterminer si le sable est fin ou grossier, s'il est homogène ou hétérogène, s'il contient des coquilles, s'il est compacté ou pas.Pour répondre à de telles questions, il est nécessaire d'utiliser des relations entre les vitesses du son et les propriétés des matériaux granulaires. Ces relations géoacoustiques sont quasi-inexistantes pour les sables marins, surtout pour les sables grossiers. Pour constituer de nouvelles relations géoacoustiques, il est proposé d’établir des bases de données à partir de mesures in-situ des vitesses acoustiques et des analyses des échantillons de sédiments prélevés au même endroit. Pour cela, un prototype de célérimètre a été développé, INSEA (INvestigation of SEdiment by means of Acoustic), qui permet de mesurer les vitesses et l'atténuation du son dans des sédiments, y compris dans des sables grossiers. / This thesis is a contribution to the seabed exploration by means of acoustical methods. We focus on sediment seabeds, especially on sand sediments because there are often encountered off European coasts. Existing acoustic methods for seabed characterization generally aim to qualify the sub-seafloor: sediment thickness or bedrock cap detection.Nevertheless, accurate sediment stratification or details of the involved sediment are generally out of reach. This is why SCAMPI (Sub-seafloor Characterization by Acoustic Measurements & Parameters Inversion) have been have designed. This patented device is a geoacoustical inversion method based on an underwater acoustic instrumentation towed in water column. It aims to identify and characterize sediment layers over a thickness of 5-10 meters below the seabed, quantifying major physical parameters as compressional speed. But vertical sound speed profiles of the sub-seabed is a necessary step but is insufficient to predict refined information about the sediment: is it coarse, homogeneous, does it contain inclusions ..?To give answers to these questions, geoacoustical relations linking acoustic parameters to sedimentological parameters are required. But these relations are sparse, particular for coarse sands.A velocimeter prototype INSEA (INvestigation of SEdiment by means of Acoustic) have been designed, to measure in situ acoustical parameters of the first centimeters of the seafloor, even in coarse sands. This work is the preliminary work leading to a new project which consist in building specific data bases to elaborate these geoacoustical relations and theoretical modeling in granular wet media suited to marine geophysics applications.

The use of iceberg-keel ploughmarks as proxy indicators of past and present iceberg morphology, keel depth and drift direction has seldom been approached in the southern hemisphere. Using high-resolution multi-beam swath bathymetry of the mid-shelf Pine Island Trough and outermost Weddell Sea shelf regions of Antarctica, detailed analysis of >13,000 iceberg-keel ploughmarks was undertaken. By considering the draft of icebergs calved from Antarctica today, calculated from detailed satellite altimetric datasets by this work, almost all observed ploughmarks were interpreted to be relict features. In Pine Island Trough, ploughmark planform parameters and cross-sections imply calving of a large number of non-tabular icebergs with v-shaped keels from the palaeo-Pine Island-Thwaites ice stream. Geological evidence of ploughmark form and modern water depth distribution indicates calving-margin thicknesses (949 m) and subaerial ice cliff elevations (102 m) equivalent to the theoretical threshold predicted to trigger ice-cliff structural collapse and calving by marine ice-cliff instability (MICI) processes. Thus, ploughmarks provide the first observational evidence of rapid retreat of the palaeo-Pine Island-Thwaites ice stream, driven by MICI processes commencing ~12.3 cal ka BP. On the Weddell Sea shelf, ploughmark morphologies imply considerable variation in palaeo-iceberg shape and size, most likely reflecting calving from multiple source margins. In turn, an absence of grounded ice on the Weddell Sea shelf and a palaeo-oceanographic regime comparable to today are implied at the time of formation. Analysis of a 3D seismic cube of the Sea Lion Field area of the North Falkland Basin reveals iceberg-keel ploughmarks incised into the modern- and palaeo-seafloor, formed by icebergs of varying shape and size that most-likely calved from the Antarctic Ice Sheet during three past glacial periods (estimated ages ~18 - 26.5 ka BP, ~246 ka BP, ~9.8 Ma BP). Despite illustrating the possibility of iceberg drift into the North Falkland Basin today, the relict ploughmark age implies little risk to any seafloor structures in the area, which might be required for hydrocarbon production. By these analyses, the significance of iceberg-keel ploughmarks as indicators of palaeo-glaciology and palaeo-oceanography at the time of formation is emphasised.