Seismic Refraction Data Analysis Using Machine Learning and Numerical Modeling for Characterization of Dam Construction Sites

Seismic refraction is a cost-effective tool to reveal subsurface P-wave velocity. Inversion of travel times for estimating a realistic velocity model is a significant step in the processing of seismic refraction data. The results of the seismic data inversion are stochastic and, thus, using prior information or complementary geophysical data can have a significant role in estimating the structural properties based on observed data. Nevertheless, sufficient prior information or auxiliary data are not available in many geophysical sites. In such situations, developing advanced computational modeling is a vital step in providing primary information and improving the results. To this aim, a new inversion framework through hybrid committee artificial neural networks (CANN) and the flower pollination (FP) optimization algorithm is introduced for inversion of refracted seismic travel times. Synthetic models generated by a forward modeling approach are used to train the machine learning model. Then, model parameters, such as the number of layers, thicknesses, and P-wave velocities, are predicted using a committee machine constructed based on several neural networks, which is achieved by averaging and stack generalization methods where the latter method provides a better result. Then, the CANN results are used in the FP inversion algorithm to estimate the final model as it provides essential prior information on the number of layers and model parameters, which can be used in the FP searching algorithm. The proposed inversion procedure is tested on different synthetic datasets and applied at a dam site to determine the number of layers and their thicknesses. Our findings indicate a successful performance on both synthetic and real data for automatic inversion of seismic refraction data.

Estimating bend-faulting and mantle hydration at the Marianas trench from seismic anisotropy

<p><span>Bend faults formed in oceanic lithosphere approaching deep ocean trenches promote water circulation and the formation of hydrous minerals. As the plate subducts, these minerals can dehydrate into the mantle wedge, generating the melts that feed arc volcanoes, or subduct fully into the deeper mantle. Balancing the global water budget requires an estimate of the amount of water recycled to the mantle by subduction, but current estimates for water fluxes at subduction zones span several orders of magnitude, mainly because of large uncertainties in the amount of water carried in the lithospheric mantle of the incoming plate. </span></p><p><span>We use active source seismic refraction data collected on the incoming plate at the Marianas trench to measure azimuthal seismic anisotropy in the uppermost mantle, and assess the degree of faulting and associated serpentinization of the uppermost mantle based on spatial variations in the observed anisotropy. We find that the fast direction of anisotropy varies with distance from the trench, rotating from APM-parallel at the eastern side of the study area to approximately fault-parallel near the trench. The fast direction orientations suggest that a coherent set of bend-faults are beginning to form at least 200 km out from the trench, although the extrinsic anisotropy signal from the faults does not substantially overprint the signal from preexisting mineral fabrics until the plate is ~100 km from the trench. The average (isotropic) mantle velocity decreases slightly as the plate nears the trench. Preliminary interpretation suggests that the observed spatial variations in anisotropy can be explained by serpentinization localized along pervasive, trench-parallel faults or joints.</span></p>

Assessing the rotation and segmentation of the Porcupine Bank, Irish Atlantic margin, during oblique rifting using deformable plate reconstruction

<p>With an increasing number of global and regional plate reconstruction models established in recent years, the motion of the Porcupine Bank, Irish Atlantic continental margin, underlain by orogeny-related pre-rift crustal basement terranes, have been investigated and restored as well.  However, these reconstructed models of the Porcupine Bank margin mainly depend on potential field data analysis and lack seismic constraints, failing to reveal the role of inherited crustal sutures during rifting and associated crustal deformation over geological time. In this study, five deformable models with distinct structural inheritance trends are established in GPlates by adjusting a previously published regional restoration model for the North Atlantic realm. For each model, driving factors (e.g., such as whether the Orphan Knoll is included, the altered rotational poles of the Flemish Cap, and the motion of the eastern border of the Porcupine Basin) are also taken into consideration. Crustal thicknesses from gravity inversion and seismic refraction data modelling are compared against those from these deformable plate reconstruction models to identify the most geologically reasonable one. The resulting preferred model has the Porcupine Bank subdivided into four blocks with each experiencing polyphase rotations and shearing prior to final continental breakup, implying strong inheritance and segmentation of the Porcupine Bank and the Porcupine Basin. The derived reconstructed paleo-positions over time of the Flemish Cap and the Porcupine Bank within the deforming topological network reveal new and evolving conjugate relationships during rifting, which are assessed using regional seismic transects from both margins. Finally, extensional obliquity between both margins is quantitatively restored, showing time-variant orientations due to the rotation and shearing of associated continental blocks, which contributes to unraveling the spatial and temporal evolution of southern North Atlantic rifting during the Mesozoic, prior to the initiation of seafloor spreading.</p>

THE LITHOSTRATIGRAPHY OF THE NEAR-SURFACE IN PART OF SEDIMENTARY KOLMANI FIELD IN NORTHERN BENUE TROUGH, NIGERIA, USING SOIL CORE AND SEISMIC REFRACTION DATA

Soil samples from 31 shallow boreholes were acquired at depths 0m, 1m, 2m, 3m, 4m, 5m, 7m, 10m, 15m, 20m, 25m, 30m, 35m, 40m, 45m, 50m, 55m, and 60m in Pingida (Kolmani Field) in Ako LGA, Gombe State, Nigeria. Using the same boreholes, seismic refraction data was also acquired. The aim of the survey was to delineate the near-surface lithology and velocity layering. The boreholes were drilled using rotary drilling rig and the core samples acquired and described using Wentworth Scale. Seismic refraction data acquired using a single trace Stratavisor NZXP portable digital recorder. The recording spread consisted of a single SM4- 10Hz geophone positioned at depths where the soil samples were taken. A hammer was used as the energy source and placed 3m away from the hole to obtain the first breaks. The refraction data was interpreted using UDISYS Version 1.0.0.0 software. The soil layers in the Kolmani Field have three distinct layers specified as follows, namely, top weathered and sub-consolidated layers made up of intercalation of sandstone, gravel ash clay and muddy coal shale. The lithologic strata do not correlate throughout the field resulting from the highly variable elevation which ranged from 317m and 524m with average of 389.16m. The top weathered layer of laterite intercalated with cobblestones with compressional wave velocity ranging from 342 ms-1 to 517 ms-1 with an average of 405.03 ms-1. Beneath the weathered layer is the sub-consolidated Clay layer intercalated with silt and laterite of compressional wave velocity ranging from 440 ms-1 to 1854 ms-1 of average of 826 ms-1. The underlying consolidated layer is the shale and coal layer having compressional wave velocity ranging from 1518 ms-1 to 4201 ms-1 with an average of 2162.65 ms-1. The dominant lithologic sequences encountered are laterite, clay, silt, sand, gravel, coal and shale. The results of this work can be used for static corrections in seismic reflection processing, planning and assessing risk for engineering structures, and for groundwater exploration. The laterite, clay, silt, sand, gravel, coal and shale can be utilized in agriculture, construction, process industries, and environmental remediation.