Natural marine sediments are heterogeneous with respect to sediment‐physical properties, and have a wide range in composition and structures. For many years, sediment‐physical characterization has relied primarily on laboratory experiments. However, the investigation of small‐(grain‐)scale sedimentary structures, which appear to control many sediment (re‐)depositional and emplacement mechanisms, requires new analytical methods. Here, we test high‐resolution X‐ray synchrotron micro‐tomography (μCT) to qualitatively and quantitatively investigate structural differences, in 3D, between two lithological end‐member types of marine sediments: a coarse‐grained, sandy sediment and a fine‐grained, silty‐clay sediment. Our results show clear compositional and structural differences between the two end‐members, as well as between samples taken from the same lithological unit. These differences can be attributed partly to different sediment types, that is, coarse‐versus fine‐grained sediments, but also reveal a dependency on the sedimentation regime. We find that pore space distribution is highly spatially variable, even down to a sub‐millimeter scale. Such high variability in porosity would be missed by standard geotechnical experiments, which only provide information averaged over far larger sediment samples. The identification of small‐(grain‐)scale changes in pore space, however, directly impacts sediment properties such as permeability, which in turn is crucial for the understanding of geological processes such as fluid flow and storage capacity of sediments and assessing hazards such as the preconditioning of submerged slopes to collapse. Our results therefore demonstrate the potential of μCT to investigate the internal structure of natural sediments, obtaining information that is not resolved or lost in data acquired through other analytical methods.

Summary This abstract presents an integrated database created for the TNW wind farm and evaluates the importance of various features for CPT prediction. The database includes data from the geological structural model, seismic attributes, and geotechnical CPTs. We train different regression and classification models on the database to evaluate the models' predictive power and discuss the importance of features. We use leave-group-out cross-validation to assess the best estimate CPT predictions and the predictive intervals. The evaluation shows that the different models based on random forest, gradient boosting, or artificial neural networks all have similar predictive accuracy. It also shows that the most essential features of the different models were the same. Based on that, we see that the most critical feature is the soil unit defined in the geological structural model. Our models are less accurate when a soil unit's main grain sizes vary. It is, therefore, essential for the predictive models we have worked with to get soil units correct. To do that, integrated interpretation of the geology, geophysical and geotechnical data is essential. We also see that seismic attributes strongly linked to geotechnical soil properties seem more critical than pure geophysical attributes.

The exploration and development of offshore renewable energy sources necessitates the handling of vast amounts of multi-disciplinary data across extensive areas. An efficient method of linking and integrating these data into a coherent ground model has become a theme of great interest. Such a ground model should assist developers in determining subsequent site investigations, optimizing field layout and foundation design, identifying geohazards and implementing suitable risk mitigation strategies. However, current practices often result in a qualitative integration, where geological formations are identified without assigning geotechnical properties to them, rendering the ground model unsuitable for site development. Moreover, the complexity of geological histories, particularly in relation to glacial and interglacial cycles and sea-level fluctuations, poses significant challenges. Despite the scarcity of geotechnical data, efficient integration of multi-disciplinary data is essential for creating reliable ground models that capture uncertainty accurately. Geophysical data provides information on the 3D geological and structural framework, while unit properties can be derived from in situ tests and laboratory data consistently. Our goal is to develop a fully integrated ground model, applicable to geotechnical applications, such as the Ten Noorden van de Waddeneilanden Wind Farm Zone (TNW), located off the Dutch coast. We propose a site-specific, data-driven integrated ground model that facilitates further development of the TNW. This quantitative ground model merges all available geotechnical and geological Ground Investigation data with geophysical data into a comprehensive, consistent 3D ground model, highlighting stratigraphic and spatial variations in ground conditions across the site.

Geotechnical and geophysical observations are the pillars upon which marine site survey ground models are built. Ideally, such models would accurately capture the distribution of physical properties within the subsurface such that the design and installation phases for offshore engineering projects can be completed safely and efficiently. To this end, there has been significant interest in leveraging quantitative information from both data types, allowing for a more effective integration. Critical in this process is capturing the evolution of prediction uncertainties, both in the final model and at intermediate stages. We present a fully integrated, quantitative ground model for the Ten noorden van de Waddeneilanden Wind Farm Zone in the southern North Sea. Ultra-high frequency seismic, CPT and borehole data were integrated using a stochastic workflow to capture and propagate uncertainties originating from each input and processing step. The final ground model includes synthetic CPT predictions across the wind farm area (including uncertainties), from which engineering design parameters were estimated.

The rapid growth in demand for offshore renewables has driven significant interest in linking near-surface geophysical and geotechnical data to develop integrated ground models. Using a machine learning-based workflow to link parameters we have predicted synthetic cone penetrometer (CPT) responses from geophysical data. We have sought to determine how many physical CPTs are required to train the machine learning algorithm to predict the magnitude and uncertainty of a synthetic CPT appropriately. This is a complex, multivariate analysis influenced by variable ground conditions and the types of infrastructure being installed. We provide insight into this project-critical issue, illustrating the influence of the depositional environment, loading history for a particular soil unit, and the foundation concept design on the quantity of CPT information required to obtain an adequate prediction of the ground conditions from geophysical inputs. While the results are not definitive, they provide valuable indicators for optimization of survey plans.

The distribution of physical properties within the shallow subsurface has a major influence on the design and installation phases of offshore engineering applications. Factors such as over-pressured weak layers, hard layers, and the soil strength profile, all constrain the type and size of foundations that are appropriate for any given site. A quantitative, integrated ground model is therefore vital for accurate and reliable characterization of the soils and underlying processes to minimize the economic and environmental risks. We present a quantitative analysis of a variable, locally high amplitude, seismostratigraphic boundary from the Hollandse Kust West designated development zone using ultra-high-frequency multichannel seismic reflection data. Combining seismic Q-factor and post-stack acoustic impedance inversions, the high amplitude sections of this stratigraphic boundary are quantitatively interpreted as over-pressure generated by trapped pore fluids. The resultant reduced loading capacity of these over-pressured soils compared to hydrostatic soils are discussed in a geological and foundation design context.

Summary The in-depth integration of sparse 1D geotechnical data with 2D UHR seismic reflection in a consistent geological framework forms the back-bone of the data-driven ground model approach. In order to predict CPT or geotechnical parameters (and their uncertainties) across the entire development area, one typically relies on geostatistical methods, like 3D kriging, Considering that the 2D line spacing is often larger than key geological phenomena, this interpolation will lead to uncertainty. In this paper, we investigate the effect of line spacing and geological complexity on the model prediction, using the TNW site (offshore the Netherlands) as a case study. We focus on an area with ultra-high-resolution 3D data, and decimate the volumes of 4 sub-sets with distinct geological features and complexity in order to assess the uncertainty on the interpolation, using both geostatistical and machine learning methods.

Shear stiffness is critical in assessing the stress–strain response of geotechnical infrastructure, and is a complex, nonlinear parameter. Existing methods characterise stiffness degradation as a function of strain and require either bespoke laboratory element tests, or adoption of a curve fitting approach, based on an existing data set of laboratory element tests. If practitioners lack the required soil classification parameters, they are unable to use these curve fitting functions. Within this study, we examine the ability and versatility of an artificial neural network (ANN), in this case a feedforward multilayer perceptron, to predict strain-based stiffness degradation on the data set of element test results and soil classification data that underpins current curve fitting functions. It is shown that the ANN gives similar or better results to the existing curve fitting method when the same parameters are used, but also that the ANN approach enables curves to be recovered with ‘any’ subset of the considered soil classification parameters, providing practitioners with a great versatility to derive a stiffness degradation curve. A user-friendly and freely available graphical calculation app that implements the proposed methodology is also presented.

A key component of any marine site investigation project is the development of a geohazard register that identifies all the geohazards that potentially could be present within (or immediately adjacent to) a site. Part of the remit for such a register is to also provide an, often heuristic, assessment of the hazard risk. For many marine infrastructure projects, the list of geohazards is incredibly extensive, including factors that influence all elements for infrastructure design; from foundations to cables or pipelines. However, the majority of the geohazards identified often pose only a low-level hazard that can be relatively easily mitigated through appropriate decision making during operational planning and/or design. Key to making these decisions is the availability of information that allows the hazard to be quantified. In this paper, I look at ways in which the hazard can be better quantified by leverage further information from the (often) early phase geophysical data. Better quantification of these hazards earlier in the project cycle should facilitate more streamlined operations with less risk of over-design and (hopefully) fewer surprises during installation.

Summary As part of the energy transition, there is a significant growth in offshore wind energy developments on a global scale. These projects require the integration of large volumes of geotechnical and geophysical data, which need to be put in a consistent geological context to understand the geological complexity of the site, and its implications for engineering design. In this paper, we present the state-of-the-art development of a data-driven, integrated ground model for the TNW (Ten noorden van de Waddeneilanden) offshore wind site, the Netherlands. The model starts from establishing a detailed seismostratigraphic or structural model, and culminates in a predictive 3D ground model, which contains parameters – as well as their uncertainty – relevant for geotechnical engineering and design applications (e.g., foundation design, geohazard assessment, layout design, etc.). The models combine seismic inversion and machine learning techniques to optimize the workflow.

The standard workflow for marine magnetic surveys relies heavily on the assumption that the analytic signal amplitude and position of maxima provide reliable information about the size and location of the potential UneXploded Ordnance (UXO). Here we present studies using synthetic data that demonstrate how the aspect ratio (length/width), orientation and geographic location of the UXO can significantly affect the amplitude and shape of the analytic signal and cause the peak to migrate away from the center of the target. The implications of this for UXO site surveys are two-fold. Firstly, the dependence of the signal amplitude on the target orientation and geographic location mean that potentially important UXO could be disregarded if a simple amplitude thresholding is used to isolate potential targets. Secondly, the complex relationship between the target shape/orientation and location of the signal peak can cause positional errors of up to several meters, which are critical for subsequent mitigation procedures.

Summary Ultra High Frequency-Multi Channel Seismic (UHF-MSC) data is increasingly being implemented in industry and academia to analyse and quantify shallow subsurface (c. <200m) conditions. However, its use is reliant on the production of consistent and reliable imaging or qualitative interpretations and the use of bespoke quantitative methodologies. The imaging of UHF-MCS data can be cast through the use of either post-stack, pre-stack, time-domain and depth-domain. Traditionally post-stack time migration has been preferred processing methodology due to its low computational costs and relative insensitivity to velocity errors. While pre-stack time migration improves on signal to noise ratios and produces higher fidelity images, it is pre-stack depth imaging (PSDM) which allows for the highest fidelity images, specifically where lateral velocity variations occur and the geology of the near-surface is complex. However, the increased imaging potential of pre-stack depth migration is countered by a greatly increase computational and user cost. Comparison of these migration strategies gives the opportunity to analyse the image improvements and the robustness of the velocity models used. The initial results show that the improvement in imaging and production of high-fidelity velocity models through the use of PSDM outweigh the increased computational and user cost during the UHF-MCS processing workflow.

Time-lapse (4D) seismic imaging is now widely used as a tool to map and interpret changes in deep reservoirs as well as investigate dynamic, shallow hydrological processes in the near surface. However, there are very few examples of time-lapse analysis using ultra-high-frequency (UHF; kHz range) marine seismic reflection data. Exacting requirements for navigation can be prohibitive for acquiring coherent, true-3D volumes. Variable environmental noise can also lead to poor amplitude repeatability and make it difficult to identify differences that are related to real physical changes. Overcoming these challenges opens up a range of potential applications for monitoring the subsurface at decimetric resolution, including geohazards, geologic structures, as well as the bed-level and subsurface response to anthropogenic activities. Navigation postprocessing was incorporated to improve the acquisition and processing workflow for the 3D Chirp subbottom profiler and provide stable, centimeter-level absolute positioning, resulting in well-matched 3D data and mitigating 4D noise for data stacked into [Formula: see text] common-midpoint bins. Within an example 4D data set acquired on the south coast of the UK, interpretable differences are recorded within a shallow gas blanket. Reflections from the top and bottom of a gas pocket are imaged at low tide, whereas at high tide only the upper reflection is imaged. This case study demonstrates the viability of time-lapse UHF 3D seismic reflection for quantitative mapping of decimeter-scale changes within the shallow marine subsurface.

Pore pressures higher than hydrostatic correspond to localized reductions of the level of shear stress required to induce lateral mass movement in a slope, and therefore play a key role in preconditioning submarine landsliding. In this paper, we investigate whether multi-channel seismic reflection data can be used to infer potentially destabilizing pore-pressure levels at a resolution and sensitivity useful for in-situ slope stability characterization. We simulate the continuous deposition of sediment on consolidating slopes in two scenarios, with combinations of sedimentation rate and permeability distribution leading to disequilibrium compaction. Ultra-high-frequency (UHF; 0.2–2.5 kHz) seismic reflection data are computed for each model and a stochastic full waveform inversion (FWI) method is used to retrieve the sub-seabed properties from the computed seismograms. These are then interpreted as time–depth variations in the effective stress (σʹ) regime, and therefore local overpressure ratio and factor of safety, using a combination of p-wave velocity to σʹ transforms. The results demonstrate that multi-channel UHF seismic data can provide valuable constraints on the distribution of physical properties in the top 50 m below seabed at a sub-metric scale, and with a sensitivity useful to infer destabilizing excess pore pressure levels. Thematic collection: This article is part of the Measurement and monitoring collection available at: https://www.lyellcollection.org/cc/measurement-and-monitoring

Rivers (on land) and turbidity currents (in the ocean) are the most important sediment transport processes on Earth. Yet how rivers generate turbidity currents as they enter the coastal ocean remains poorly understood. The current paradigm, based on laboratory experiments, is that turbidity currents are triggered when river plumes exceed a threshold sediment concentration of ~1 kg/m3. Here we present direct observations of an exceptionally dilute river plume, with sediment concentrations 1 order of magnitude below this threshold (0.07 kg/m3), which generated a fast (1.5 m/s), erosive, short-lived (6 min) turbidity current. However, no turbidity current occurred during subsequent river plumes. We infer that turbidity currents are generated when fine sediment, accumulating in a tidal turbidity maximum, is released during spring tide. This means that very dilute river plumes can generate turbidity currents more frequently and in a wider range of locations than previously thought.

Annually laminated sediments (varves) provide excellent temporal resolution to study rapid environmental change, but are rare in the early part of the Last Termination (similar to 19-similar to 11.7 ka BP). We present a new >400 varve year (vyr) varve sequence in two floating parts from Windermere, a lake at the southern margin of the mountains of northwest England. This sequence records the final retreat of the Windermere glacier at the southern edge of the Lake District Ice Cap during the transition from Heinrich Stadial 1 (similar to 18-similar to 14.7 ka BP) into the Lateglacial Interstadial (similar to 14.7-similar to 12.9 ka BP). Laminated sediments from four lake cores from Windermere's northern and southern basins were investigated and shown to be varved. These sequences are integrated with seismic reflection evidence to reconstruct south-to-north deglaciation. Seismic and sedimentological evidence is consistent with gradual stepped ice retreat along the entire southern basin and into the northern basin between 255 and 700 vyr prior to the appearance of significant biota in the sediment that heralded the Lateglacial Interstadial, and had retreated past a recessional moraine (RM8) in the northern basin by 121 vyr prior to the interstadial. The Lateglacial interstadial age of this biota-bearing unit was confirmed by C-14-dating, including one date from the northernmost core of similar to 13.5 cal ka BP. A change in mineralogy in all four cores as the glacier retreated north of the Dent Group (the northernmost source of calcareous bedrock) and a decrease in coarse grains in the varves shows that the ice had retreated along the entire North Basin at similar to 70 vyr prior to the Lateglacial Interstadial. The estimated retreat rate is 70-114 m yr(-1) although buried De Geer moraines, if annual, may indicate retreat of 120 m yr(-1) with a >= 3 year stillstand at a recessional moraine halfway along the basin. The glacier then retreated north of the lake basin, becoming land-terminating and retreating at 92.5-49 m yr(-1). The northernmost core has a varve sequence ending at least 111 vyr after the other core chronologies, due to the increased proximity to remnant ice in the catchment uplands into the early Lateglacial Interstadial. We show that almost all of the glacier retreat in the Windermere catchment occurred before the abrupt warming at the onset of the Lateglacial Interstadial, in keeping with similar findings from around the Irish Sea Basin, and suggesting a similar retreat timescale for other radial valley glaciers of the Lake District Ice Cap. The seismic and core evidence also show the potential for a much longer varve chronology extending at least 400 and potentially over 1000 vyr further back into Heinrich Stadial 1 (18-14.7 ka BP), suggesting that glacier retreat in the Windermere valley initiated at least before 15.5 ka BP and perhaps 16 Ka BP. (C) 2019 The Authors. Published by Elsevier Ltd.