DC electrical resistivity surveying has shown much promise for investigating dikes and other earthen flood barriers. We are interested in the applicability of such data for aiding with maintenance and construction efforts in the Tantramar region of New Brunswick and Nova Scotia, Canada, where agricultural dikes form an important part of critical flood prevention infrastructure. Specifically, our goal is to develop efficient field survey and data processing protocols for detecting possible internal issues in the dikes ahead of further, more detailed geophysical surveying. The field survey protocol must be cost and time effective, given the large lengths of dikes that must be surveyed. The Tantramar dikes are expected to exhibit strong subsurface heterogeneity but accurately characterizing their internal structure may be challenging. Dikes have significant 3D geometry and traditional 2D DC surveying, and subsequent 2D inversion, fails to provide reliable and interpretable results. 3D surveying and inversion may be required but this represents significantly higher field costs. We performed a detailed synthetic inverse modelling study to help design our field surveying protocols. We used a representative model of a dike in the Tantramar region and we worked with the specifics of the surveying equipment available to us. We investigated and compared three possible data acquisition layouts proposed by other authors, we thoroughly compared the results of 2D versus 3D inversion on those layouts, and we performed a detailed investigation to assess best practices for 3D inversion mesh design. We are also incorporating joint interpretation with EM data, collected using mobile survey devices such as the Geonics EM31. Results from synthetic forward and inverse modelling are helping us develop future field data collection, processing and modelling protocols.

How to cite: Lelièvre, P., Vandenberg, E., Hebb, H., Butler, K., Lu, X., and Farquharson, C.: A protocol for assessing the effectiveness of electrical resistivity imaging for agricultural dike investigation, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8542, https://doi.org/10.5194/egusphere-egu23-8542, 2023.

Anthropogenic and natural hazards assessments need a good knowledge of the structures. A classical approach based on geological observations or soil mechanics investigations is often insufficient to characterize both the structures and the nature of subsurface materials. For several decades, near-surface geophysical methods have been integrated into a multidisciplinary strategy to improve the characterization of small-scale features of the subsurface. Electrical Resistivity Tomography (ERT) is a standard approach among these methods. This technique has several advantages, including easy deployment in the field and sensitivity to lithology, fluid contents, or chemistry. With this method, it is possible to detect and characterize the geometry of sliding surfaces on landslides and actives faults. It is also possible to set a permanent survey and obtain time-lapse images to describe temporal changes of resistivity within the subsurface and investigate dynamic processes, such as groundwater flows or soil moisture variations.
The ERT method consists of recording apparent resistivity data and inverting them to map the resistivity distribution at depth and to capture possible time changes. Many softwares, such as  Res2DInv, R2, or PyGimli, are now available to carry out the inversion. However, the quality assessment of the obtained models remains an open and challenging question. Indeed,  the robustness of the ERT results depends on factors such as the acquisition geometry, data error,  the resistivity contrast in the subsurface, the inversion procedure, and its parametrization. 
To overcome these limitations and allow a more accurate interpretation of the ERT models, we propose a new approach for assessing the reliability of ERT images. We propose a new algorithm called PySAM (Python Sensitivity Approach iMprovement) based on the open-source library PyGimli. This new tool provides relative and absolute error assessment on resistivity images from any ERT inversion software. We first illustrate the relevance of this new tool from synthetic tests associated with a well-contrained resisvity model. Next, we revisit the ERT image of the Topographic Frontal Thrust (TFT), a major active fault located in South Central Bhutan, and discuss its geometry which is a crucial parameter to discuss strain accommodation, and improve the seismic hazard assessment in Nepal, Bhutan, and northern India, one of the most densely populated regions.

How to cite: Gautier, M., Gautier, S., and Cattin, R.: A new approach to quantify the reliability of Electrical Resistivity Tomography (ERT) images, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12499, https://doi.org/10.5194/egusphere-egu23-12499, 2023.

The induced polarization (IP) method is an extension of the electrical resistivity method that allows the measurement of both the electrical conductive and capacitive properties of the subsurface; it is one of the main methods applied in landfills to characterize the geometry and composition of waste as well as the migration of leachate. Commonly, landfill IP investigations are based on measurements along several 2D lines. Considering the complexity of landfills, we investigate here the resolving capabilities of 2D parallel electrode lines with inline measurements, and 3D electrode configurations (grid array with electrodes set in a quadratic mesh and circular array with electrodes set in four concentric circles) through a numerical study and field measurements. The field surveys were conducted on two landfills with different waste composition, with measurements in the frequency range between 1 and 240 Hz to solve the frequency-dependence of the electrical properties. The results of both the numerical study and the field data show a lack of sensitivity in the case of the 2D configuration leading to the creation of artefacts in the conductivity magnitude and phase imaging result. An underestimation of IP values is also seen for these arrays; such effects are particularly critical in the case of heterogeneously distributed IP anomalies, which are typical in landfills. In contrast, the tested 3D configurations are able to resolve the geometry of the electrical units correctly and anomalies are more sharply defined compared to the results obtained by 2D configurations. Furthermore, our results show that the grid array with crossline measurements and multiple dipole orientations provides better results than the circular array, which lacks in the resolution in the central area. Additional investigations of the frequency-dependence of the field data demonstrate that for the different study areas only 3D configurations provide smooth spectra of the conductivity magnitude and phase, which is essential for an accurate estimation of relaxation (e.g., Cole Cole) parameters.

How to cite: Moser, C., Flores Orozco, A., and Binley, A.: Resolving capabilities of 3D electrode configurations for spectral induced polarization surveys, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2363, https://doi.org/10.5194/egusphere-egu23-2363, 2023.

EZPONDA is a FEDER funded project, which aims to study both the mechanichal and chemichal processes related to the erosion of the coastal area in the Basque country, France. In the Socoa flysh cliff, the presence of fractures roughly perpendicular to the shoreline control the nucleation and the growth of underground erosion cavities. The ‘Socoa Semaphore cavity’ is the most striking one, with a propagation of the void up to 30 m inland. Mapping sea cliff fracturation extent around this cavity is a critical aspect to anticipate possible future erosion processes.

Assuming the permeability of the fractured material is higher than the permeability of the nonfractured material, mapping water infiltration in the subsurface may be used as a proxy to map the fracturation extent. In this work, we propose to monitor the sea water infiltration during high tide using passive seismic listening and self potential electrical response to illuminate fractures in the surrounding of the ‘Socoa Semaphore cavity’.

72 vertical component autonomous 5 Hz seismic sensor were deployed at the surface over 5000 m2 with an average interstation distance of 10 m. Continuous records were collected between 19/09/2020 and 22/09/2020 (4 days) during a large tidal event to include 8 high tides with a coefficient higher than 100.  It should be noted that the first two days of measurements were carried out over the weekend. The self-potential signals were recorded at the ground surface using a set of 20 nonpolarizable Pb/Pbcl2 electrodes. Data were recording using a Campbell Scientific CR1000 datalogger, with multiplexer chips used to switch between the pole electrodes. Voltage were measured between the 19/09/2020 11am to 21/09/2020 16pm.

The seismic spectrograms show that between 5-20 Hz, anthropological activities such as trafic and harbour modulate the seismic energy for all sensors. In the 20-40 Hz frequency range, the sea height modulates the seismic energy for all sensors, with a seismic energy decreasing as a function to the distance to the coast.  For a large frequency range between 10-50 Hz, we observe that the relative change in median spectral amplitude during high tides with respect to the median amplitude during the full observation period exhibits highest value over a restricted area (400 m2) located east to the the ‘Socoa Semaphore cavity’, which extends far beyond the known void extent. We argue that this area with a singular geophysical signature may be related to the presence of fracturation. Self potential measurement shows a lower noise during the night (around 4 mV) compare to the day (about 10 mV). In addition the noise is higher on Monday (about 20 mV). Self potential measurement show periodic oscillations with a period of 6.4 hours approximately, corresponding to half the tidal cycle. Amplitude variations of self potential signal is more delicate to be interpreted and need further development. 

How to cite: Deparis, J., Gaudot, I., Bretaudeau, F., Baltassat, J.-M., and Garnier, C.: Mapping sea cliff fracturation using passive seismic and self potential responses : case study of the Socoa flysh cliff (Basque Country, France), EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5073, https://doi.org/10.5194/egusphere-egu23-5073, 2023.

The problem of landslides is one of the greatest challenges in geohazard research. Due to their unpredictability, and complicated genesis, their detailed and accurate observation is necessary. Despite many studies on the subject, a general scheme for their recognition has still not been developed. An additional, and important fact that has recently been observed is the impact of the current state of the climate, and the human response to it. 

In the presented research results, an example where anthropogenic factors can have a significant impact on the evolution of a creeping landslide is described. As a result of changes in precipitation over years, artificial snowmaking is necessary to extend and even maintain the ski season on ski slopes and results in the unique characteristics of those landslides. In this presentation we shows the results of 4 years of geophysical observations, integrating multiple methods from geophysical imaging and remote sensing to determine the characteristics of the landslide, its changes and potential danger. The methods used, such as passive seismological monitoring, seismic tomography, electrical resistivity tomography, reflection imaging, terrestrial laser scanning and electromagnetic slingram in a time-lapse scheme allowed us to obtain an image of a temporally and spatially variable structure with remarkable accuracy. Additionally, there were also made an AMT profile with deep recognition range. The results obtained and their joint interpretation can serve as a reference in the study of similar landslide cases, where anthropogenic and climatic factors can significantly impact the evolution of such phenomena.

This research was funded by the National Science Centre, Poland (NCN), grant number 2020/37/N/ST10/01486.

How to cite: Marciniak, A., Majdański, M., Kowalczyk, S., Cader, J., Nawrot, A., Owoc, B., Stan-Kłeczek, I., Górszczyk, A., Gajek, W., Oryński, S., and Czarny, R.: Integrated imaging of a landslide as a result of 4 years of observations  – A case study from Outhern Carpathians, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2585, https://doi.org/10.5194/egusphere-egu23-2585, 2023.

Experiments on a plot-scale of apparent electrical conductivity (ECa) and resistivity (ERT) variation with correlation of soil properties were studied for soil mapping at the experimental farm of Mohammed VI Polytechnic University of Benguerir (UM6P). The ElectroMagnetic Induction (EMI) technique was applied using a soil sensor EM38-MK2 which provides auxiliary ECa data sets with accuracy. The other method ERT was used to measure the electrical resistivity. The study was supported by soil sampling to ensure the reliability and potential of ECa measurements for soil mapping.  ECa readings in mS/ m ranged from 12 to 26 and 8 to 20 respectively in the vertical (ECa-V) and horizontal mod. ECa and ERT readings correlated best with soil properties such as texture (clay and sand), and upper soil chemical properties (OM, CEC, Ca2+, Fe2+and Mg2+). A modest correlation was found between ECa-V, clay and subsurface water content (r = 0.80), (r = 0.79). The linear relationship found between apparent electrical conductivity and soil clay content explained 80% of the measured variability. The results of the study raised the hope that soil mapping by ECa measurement can fairly represent the spatial variation of soil properties such as texture, chemical fertility and organic matter content. The use of spatial variability in EC as a co-variate in statistical analysis could be a complementary tool in the evaluation of experimental results.

How to cite: Aallem, F.-E., Charbaoui, A., Ahmed, L., Kchikach, A., Jaffal, M., and Jnaoui, Y.: Geophysical methods of soil fertility mapping for precision agriculture applications in semi-arid regions (Morocco), EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8469, https://doi.org/10.5194/egusphere-egu23-8469, 2023.

In geophysical data inversion, one way to decrease the non-uniqueness of the solutions is to incorporate structural constraints. Such structural constraints are typically derived from collocated geophysical data, which are more sensitive to subsurface structures and parameter contrasts than the to-be-inverted data. When using a smooth regularization operator, a straightforward approach is to reduce the local weight of the smoothness constraints in model regions where we expect an interface. However, when using such an inversion approach, the capability to reconstruct a sharp interface relies only on the structural a priori information; i.e., model areas where no structural a priori information is available are solely controlled by the standard smoothness constraints. Therefore, this approach is not optimal in practice, as the structural a priori information is often not complete.

In this study, we evaluate a structurally-constrained inversion approach based on the Minimum Gradient Support (MGS) regularization, which is capable to promote sharp interfaces also in areas where no structural a priori information is explicitly specified. We test and evaluate this regularization approach for the inversion of frequency-domain electromagnetic induction (FD-EMI) data, where we use a constant-offset 3D GPR data set to derive structural a priori information. Our field data set covers an area of about 120 m x 50 m and has been collected at a field site in Kremmen, Germany, to explore peat deposits. Our results demonstrate that the proposed structurally-constrained inversion approach helps in finding a reliable subsurface structures (e.g., peat thickness) as well as a reliable reconstruction of the subsurface electrical conductivity distribution within the peat formation (e.g., related to varying degrees of peat decomposition) and within the sandy substratum.

How to cite: Klose, T., Guillemoteau, J., Vignoli, G., Koyan, P., Walter, J., Herrmann, A., and Tronicke, J.: Structurally-constrained FD-EMI data inversion using a Minimum Gradient Support (MGS) regularization, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7067, https://doi.org/10.5194/egusphere-egu23-7067, 2023.

The Radio-Magnetotelluric (RMT) method is a geophysical near-surface imaging technique with a broad range of possible applications. In 2020, the GFZ Potsdam has acquired a newly developed horizontal magnetic dipole transmitter that allows the application of the RMT method even in regions with an insufficient coverage of radio transmitters which normally serve as source signal. First controlled-source RMT measurements were conducted at three different locations in Chile in 2020. Further measurements were recently conducted in Ireland.  As we are able to store the raw time series, we have full control over the subsequent data processing. The processing tools at GFZ include the modular processing suite EMERALD, which was originally designed for MT processing, but has recently been adapted to process RMT data. One main difference is that in RMT the transmitter data is considered as signal, while in natural source MT this would be regarded as electromagnetic noise that needs to be removed using automated robust statistical approaches. However, processing the entire time series in an automated manner has a large drawback: The different emitted frequencies are transmitted in a sweep implying that only a smaller fraction of the time series contains the required signal for a particular target frequency and leading to an unfavourable signal-to-noise ratio. Since it is technically impossible to have the same time base for the data logger and the transmitter with an accuracy of a few nanoseconds, an automated detection scheme is required to find time segments that contain the transmitter signal. Usually, several Gigabytes of raw time series are collected during field measurements, making manual editing and supervision of the time series virtually impossible. However, a careful selection of appropriate time segments is essential for the success of the data processing. To address the challenge, machine learning algorithms have a high potential to solve both problems. Initial experience was gained with a recurrent neural network approach in order to identify suitable time segments (Patzer & Weckmann, EMTF 2021 – conference contribution and personal communication). However, many questions remained open, e.g. if other machine learning algorithms can result in better performances, which machine learning algorithms are in principle suitable for the characteristics and properties of RMT time series and which parameters should be used as input variables (features) for the algorithms. A large number of machine learning algorithms exist, which can be divided into different groups according to their operating principle and their activity fields. We will test unsupervised methods, especially for clustering the data, to identify a set of suitable input variables. Subsequently, we will use these features to train supervised algorithms as logistic regression, support vector machine and different kinds of neural networks to find the best performing algorithm. We will mainly use the RMT data from Chile within the training process. Furthermore, we will test if the trained algorithm is applicable to other new data sets measured at different locations (e.g. Ireland) and/or with different equipment.

How to cite: Platz, A., Weckmann, U., and Patzer, C.: Smart data selection – Using machine learning for an automated controlled-source Radio-Magnetotelluric data processing, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-917, https://doi.org/10.5194/egusphere-egu23-917, 2023.

In (time-domain) Electromagnetic Induction (EMI) surveys, an image of the electrical conductivity of the subsurface is obtained non-invasively. An accurate interpretation of the data is computationally expensive as it requires a full (high fidelity) 3D simulation of the induced electric currents embedded within an iterative and ill-posed inverse problem. Therefore, this forward model is usually approximated with a 1D forward model (low fidelity model) which only considers horizontal layers and for which fast analytical forward models exist. Recent work [1] has shown that a multidimensional forward model can be relevant in time-domain Airborne EM inversion. To be more precise, we provided an appraisal tool for quasi/pseudo-2D inversion to indicate that fast forward 3D modelling for time-domain (Airborne) EM data is still worthwhile and, in fact, necessary, in some areas. Surrogate modelling and machine learning may replace 3D forward modelling on a mesh during a 3D inversion.

In this contribution, we first demonstrate the initial steps towards creating an efficient surrogate model for 3D modelling with only 5000 samples in the training dataset. Rather than predicting the high-fidelity or 3D data directly, we predict the relative error between the high and low fidelity data. The idea behind this approach is that predicting the difference with a relatively good low-fidelity model is easier and more robust than trying to find a surrogate for the full data set. The computation of low fidelity data via the 1D approximation is no longer a computational burden, yet it explains most of the variability in the observed data. The residual variability, originating from the non-1D nature of the subsurface, is predicted with a Gaussian process regression model. Combining the low-fidelity model with a trained correction term via Machine Learning saves significant computation times. We show encouraging results, currently limited to two layers, where the trained surrogate model proves to produce a significant ‘learning gain’ in 92,5% of the cases (see Figure 1), meaning that it can significantly reduce the residual multidimensional variability. The cases where the surrogate model makes the prediction of the high-fidelity data worse, occur at the limits of the training data space, indicating that those cases could be resolved by generating more training data in those areas.

Figure 1 – The learning gain on the test dataset by using the trained surrogate model

References

[1] Deleersnyder, W., Dudal, D., & Hermans, T. (2022). Novel Airborne EM Image Appraisal Tool for Imperfect Forward Modeling. Remote Sensing, 14(22), 5757. https://doi.org/10.3390/rs14225757

How to cite: Deleersnyder, W., Dudal, D., and Hermans, T.: Machine learning assisted fast forward 3D modelling for time-domain electromagnetic induction data – lessons from a simplified case, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12015, https://doi.org/10.5194/egusphere-egu23-12015, 2023.

Diffraction imaging has become an established tool in exploration seismology thanks to its potential to provide high-resolution information that is complementary to that contained in the corresponding reflected wavefield. In ground-penetrating radar (GPR) research, data processing schemes often neglect the diffracted wavefield, focusing instead on higher-amplitude reflected arrivals. However, these data typically contain a rich diffraction background due to the structural complexity of the near surface environment. Whereas the application of diffraction imaging to 2D GPR data has already been demonstrated, the potential of diffraction imaging for 3D GPR data is still underexplored.

Building on recent studies, we adapt a coherence-based diffraction imaging workflow, originally designed for seismic data, to common-offset GPR data. The first step of the proposed scheme is the separation of diffracted arrivals from the often predominant reflections, i.e. the faint diffracted portion of the data is separated and made accessible for dedicated processing. To this end, we approximate the reflected wavefield in a fully data-driven fashion by means of a coherent stacking scheme, and we subtract it from the data. The remaining diffracted wavefield can then be further enhanced through a second local coherent stacking procedure. Ultimately, wavefield focusing of the diffraction-only data yields an image of the distribution of subsurface scatterers.

The above-described analysis is applied to a range of 3D GPR data sets in an exploratory fashion. The localization of diffracting structures in these data sets provides valuable additional information about small-scale subsurface heterogeneities that can complement standard reflection analyses.

How to cite: Klahold, J., Schwarz, B., Bauer, A., and Irving, J.: Exploring the potential of 3D diffraction imaging for GPR data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12845, https://doi.org/10.5194/egusphere-egu23-12845, 2023.

Within the last decade, the diffracted wavefield has gained increasing importance for the processing of both seismic and ground-penetrating radar (GPR) measurements. In both communities, the separation of the diffracted wavefield remains a notorious challenge that has been approached with different deterministic methods, ranging from poststack wavefront attributes to plane-wave destruction and coherent wavefield separation. While each of these deterministic methods has characteristic advantages and drawbacks, all of them require the adaptation of processing parameters for each application, particularly when crossing scales from seismic to GPR measurements. In this study, we propose to train a convolutional autoencoder to separate the reflected and diffracted wavefields in a generalized fashion. For this purpose, we have generated highly variable synthetic seismic data that contain reflections, diffractions and noise using an algorithm that allows to compute each component individually, resulting in an automatized generation of data and labels. In order to account for the complexity of field data, we complemented the synthetic data with a large set of reference seismic and GPR field data results from coherent wavefield separation, a deterministic method, in which the reflected wavefield is modeled and adaptively subtracted from the input data. With this dataset we trained a supervised convolutional autoencoder and applied the trained neural network to seismic and GPR field measurements that were not part of the training data. The results show that the trained autoencoder is able to generalize and successfully separate the reflected and diffracted wavefields even for complex field data, resulting in an on-the-fly diffraction separation that requires no choice of parameters and is likewise applicable to both seismic and GPR data.

How to cite: Bauer, A., Schwarz, B., Walda, J., and Gajewski, D.: Deep learning diffraction separation for seismic and GPR data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9713, https://doi.org/10.5194/egusphere-egu23-9713, 2023.

Numerical modelling has been proved as an incomparable tool to understand the structure of the earth and the processes beneath the earth’s surface. Finite difference method (FDM) plays a dominant role among various numerical methods for the purpose of seismic modelling and exploration. FDM provides a comprehensible solution to the partial difference equations defining the propagations of seismic wave. These partial differential equations consist of derivatives in time and space domain. FDM can be applied by defining the elastic wave-field and model parameters at every position on a discrete mesh. Reverse-time migration (RTM) is based on exploding reflector model and it is better than other migration techniques for the interpretation of various seismic models. The present work shows the forward modelling and reverse time migration of point diffractor body placed in dipping layer of vertically transverse isotropic (VTI) medium. A 12^th order space and second order time differentiation RTM scheme have been used to interpret the location and extent of a point diffractor placed in dipping layer of VTI medium. The earth model under study is of the size 1400 m x 600 m. A dipping layer and a diffractor of size 18 m x 18 m has been placed in the VTI model. The FORTRAN code developed for FDM scheme of VTI model performs various requisite studies like stability criteria, numerical dispersion and the boundary conditions within the code. The output from the FDM code are the synthetic records at surface which after processing fed as an input in the FORTRAN code developed for RTM scheme. The position and extent of the diffractor placed in the dipping VTI medium layer has been detected properly using RTM scheme. Another FORTRAN code is developed in which forward and reverse wave propagation snapshots has been cross-correlated using various cross-correlation imaging conditions. A Laplace filter is then designed to efficiently resolve the position and extent of the diffractor in the dipping VTI medium layer.

How to cite: Sharma, S., Joshi, A., Singh, J., Pandey, M., Rastogi, R., and Srivastava, A.: Identification of Point Diffractor body placed in dipping Vertically Transverse Isotropic medium using Reverse Time Migration, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-313, https://doi.org/10.5194/egusphere-egu23-313, 2023.

The Horizontal-to-Vertical Spectral Ratio (HVSR) method and Multi-Channel Analysis of Surface Waves (MASW) method are commonly used as a joint fit technique to retrieve the 1-D shear wave velocity. The Kumaon Himalaya consists of major thrusts like MFT, MBT, SAT, NAT and MCT, from South to North) and other tectonic features. These geological structures are observed in the form of lineaments on the surface. In the present study, 2-D section of shallow shear wave velocity structure has been estimated along the transect crossing South Almora Thrust (SAT) in the Kumaon Himalaya to study the variation of shear wave velocity across the thrust. In the present work, the ambient noise survey and Multi-Channel Analysis of Surface Wave (MASW) survey has been conducted along the road profile crossing the South Almora Thrust (SAT) at equally spaced stations of 3 Km. The 1-D shear wave velocity has been used to prepare the 2-D section of shear wave velocity. The lineaments in this division have been identified by the variation in the two dimensional shear wave velocity section prepared from the so obtained 1-D shear wave velocity in this profile. The study shows that there is a good correlation between variation of shear wave velocity in the region and major tectonic features of the area. The geological sections in this area has been compared with the obtained 2D structure which give a fair amount of idea about dip of SAT in this area.

How to cite: Pandey, M., Joshi, A., Sharma, S., and Singh, J.: Shear wave velocity variation across the South Almora Thrust, Kumaon Himalaya using Joint Inversion of Horizontal-to-Vertical Spectral Ratio (HVSR) and Dispersion curve, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-879, https://doi.org/10.5194/egusphere-egu23-879, 2023.

The versatility, cost-efficiency and easy deployment of seismic sensor nodes facilitate geophysical monitoring in environments that were previously inaccessible for instrumentation, and among them landslides and unstable slopes, most of the time located in remote mountains. Using nodes allows for the setup of dense arrays with sensor inter-trace distances that become compatible with the geometries and dimensions of the geological structures to image. This becomes particularly true for landslides which have complex 3D architectures (hummocky bedrocks, layering, multi-dimensional fractures, diverse geotechnical material, deep and perched aquifers and water circulations) and are shallow processes with respect to the classical investigation depths and sensitivity of most geophysical survey techniques.

Here we develop a specific processing workflow to allow the computation of 3D shear-velocity models with Ambient-Noise-based tomography applied to dense arrays of seismic stations. The workflow is applied to a dataset acquired at the Viella shallow landslide (France) developed in altered schists and moraine deposits. We deployed 70 IGU-16HR-3C-5Hz SmartSolo sensors (EOST/PISE service) with inter station distances of 70 m for a period of 25 days.

The processing consists in several steps, all of them being tuned to the specific case of shallow depths of investigation. In areas where only few strong (M_L>4) earthquakes are triggered, with a low azimuthal distribution, surface-waves velocity fields are complex to estimate with earthquakes. Ambient noise cross-correlation tomography has the advantage of using the ambient noise to model the surface waves velocities by retrieving the interstation Green’s functions. The main hypothesis for retrieving the Green’s functions is a homogeneous noise-source distribution, which is never achieved in a natural environment. Therefore, data filtering and daily stacking are crucial to reduce the effect of non-uniform noise distributions and lead to consistent velocity models. Due to the noisy environment of Viella (torrential flows, farming activity, anthropogenic noise), several procedures were implemented to optimize the processing (reduction of the coherent noises in the processed data, use of a pseudo-topography to estimate as accurately as possible the inter-station distances and travel times). We then computed the dispersion curve diagrams for the surface waves on which we applied a strict selection to only keep the consistent part of the surface waves dispersion curves. The selection parameters were optimized for the Rayleigh and Love waves. Then, we inverted the inter-station travel times to compute group velocities maps at several frequencies. Finally, we proceed to a Markov-Chain-Monte-Carlo inversion of each of the dispersion curves extracted from the group velocity maps. We finally obtained a 3D shear velocity model, which is further combined with geological and borehole information in order to document the 3D structure.

The objectives are to present the processing workflow developed specifically for shallow imaging and the retrieval of 3D heterogeneities; effects of the processing parameters will be discussed on the Viella dataset. The approach developed for Viella is generic and has been further applied to other geological processes (permafrost at the Chauvet rock glacier, Marie-sur-Tinée mudslide), and the models will be discussed.

How to cite: Lajaunie, M., Rimpot, J., Zigone, D., Broucke, C., Malet, J.-P., Weiskopf, E., Hibert, C., Ducasse, J., and Bertrand, C.: Ambient noise shear-wave tomography for shallow landslide structural models retrieval from dense 3D seismological arrays., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11541, https://doi.org/10.5194/egusphere-egu23-11541, 2023.

Seismic inversion methods performed by a deep neural network trained in a supervised learning manner have shown successful inversion performance in synthetic data examples that target small areas. These deep-learning-based seismic inversions use time-domain wavefields as input data and subsurface velocity models as output data. Since the time-domain wavefields include both traveltimes and amplitudes of seismograms, the size of the input data is considerably large. Therefore, studies that apply deep-learning-based seismic inversions trained on large amounts of field-scale data have not yet been conducted. In this study, to apply the deep-learning-based seismic inversion technique to field-scale data, the velocity models are predicted using only traveltimes of seismic waves as the input data instead of the full time-domain wavefields. If the traveltime information is used as input data, the resolution of the inversion result is diminished, but the data size is significantly decreased, which can reduce GPU memory usage and speed up network training. We call this approach deep-learning traveltime tomography. The results obtained from this method can also be used as initial velocity models for full-waveform inversion. For network training, a large number of field-scale synthetic velocity models and corresponding first-arrival traveltimes with towed-streamer acquisition are created, and then the network is trained with the synthetic dataset. As a result of performing deep-learning traveltime tomography on an example of synthetic velocity models simulating the seafloor strata, inversion results similar to the labels were obtained. Therefore, it was confirmed that the deep-learning traveltime tomography method can immediately predict a field-scale velocity model, unlike the existing deep-learning-based seismic inversion.

How to cite: Jo, J. H. and Ha, W.: Sesimic Traveltime Tomography Using Deep Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1953, https://doi.org/10.5194/egusphere-egu23-1953, 2023.

Joint inversion of surface-waves and receiver functions has been widely used to image Earth structures to reduce the ambiguity of inversion results. We propose a deep learning method (DL) based on multi-label Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) with a spatial attention module, named SrfNet, for deriving the Vs models from Rayleigh-wave phase and group velocity dispersions and receiver functions (RFs). We use a spline-based parameterization to generate velocity models instead of directly using the existing models from real data to build the training dataset, which improves the generalization of the method. Unlike the traditional methods, which usually set a fixed Vp/Vs ratio, our new method takes advantage of the powerful data mining ability of CNN to simultaneously constrain the Vp model. A loss function is specially designed that focuses on key features of the model space (such as the Moho and the surface sedimentary layer). Tests using synthetic data demonstrate that our proposed method is accurate and fast. Application to southeast of Tibet shows a consistent result and comparable misfits to observation data with the previous study, indicating the proposed method is reliable and robust.

How to cite: Wang, F., Song, X., and Li, J.: Joint inversion of Surface-wave Dispersions and Receiver Functions based on Deep Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3223, https://doi.org/10.5194/egusphere-egu23-3223, 2023.

The use of ambient noise for passive seismic imaging has evolved into a cutting-edge, low-cost, and environmentally acceptable method of exploring the subsurface. This technique dispenses with active seismic sources, alternatively uses ambient seismic noise. Theoretical investigations have approved that an estimate of the empirical Green’s function between receivers could be obtained from the cross-correlation of ambient noise and/or dispersed coda waves. This Green’s function is mostly made up of fundamental Rayleigh waves, propagating between two receivers as if they would be caused at one of them. The applications of ambient noise surface wave tomography, from engineering and urban developments to regional and continental scales, have led to the mapping of the area's velocity model, which chiefly corresponds to the structural/geological units.

Because of numerous devastating catastrophes in recent years, several countries have made flood protection a priority. However, sea-dikes are considered remarkably heterogeneous and may fail due to their construction and/or reinforcing structures; they are potentially subject to stress by sea waves during the tidal cycle and seasonal heat variations, resulting in the water infiltration. Internal abnormalities cannot be recognised in the early stages of erosion, although visual assessments may often be relied on. In this study, we outline a passive seismic survey that was carried out to investigate technical and methodological aspects of passive seismic methods along with their application in a sea dike monitoring perspective. 

The SEEWALL project is a collaborative project, seeking to create innovative methodology to monitor the temporal evolution of sea dikes and detect early deterioration. We deployed 160 permanent 3-component MEMS accelerometers spaced 2 meters apart on top of a dike on the island of Noirmoutier (France), which was exhibiting moderate water infiltrations at its base. Despite the inhomogeneous distribution of the ambient noise sources, exploitable empirical Green's functions can be retrieved mostly from the cross-correlation of vertical component data. We estimate the surface wave phase velocity dispersion curves  using a time-frequency analysis; strictly speaking, after preconditioning the data, the cross-correlation is carried out in the frequency domain by carefully windowing data, from which each  empirical Green's functions is derived; their cross-correlations are stacked linearly by hours. The arrival times of the causal and anti-causal parts are often not fully symmetrical, indicating the diversity of major noise sources. The phase velocities measured on both positive and negative lag-times, as a function of the frequency, are computed using the phase-shift method. Interpretation of the phase velocity dispersion curves is challenging due to the geometry of the dike at the scale of the intended wavelength (a few tens to hundreds meters). But the pattern of the dispersion data appears to be relatively stable over time. It is also consistent with the dispersion curves we have obtained using active seismic hammer-shots, performed along the structure. For monitoring, we suggest using F-K spectra to highlight the variety of energy density over time, in order to advance a deeper understanding of data analysis; this enables us to discover any changes that might not be otherwise obvious.

How to cite: Kahrizi, A., Lehujeur, M., Abraham, O., Lescoat, A., Michel, L., Bardainne, T., Vivin, L., Boulay, C., Blanchais, J., Devie, T., Palma Lopes, S., Durand, O., and Gugole, G.: Ambient seismic noise processing to monitor sea dikes: the case of Noirmoutier, France, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4987, https://doi.org/10.5194/egusphere-egu23-4987, 2023.

Gheith Alfakri¹   Emmanouil Parastatidis¹  Stella Pytharouli¹

Abstract: Previous studies present evidence that microseismic monitoring could be a favourable potential technology for brownfield land site investigations, e.g. in the identification of buried objects in the shallow subsurface (< 3 m). More specifically, the presence of buried objects change the characteristics (amplitude and frequency) of a mechanical wave that propagates through a medium where this object lies. These changes have to-date only been observed at recordings from stations that are located directly above the buried object. To investigate whether a buried object can be ‘seen’ by more sensors located in the vicinity above the object, we carry out a series of numerical simulations. We examine the propagation of a sine wave emitted by a point source on the surface of a medium and study the frequency, amplitude and emitted energy from that sine wave and how these are affected by local changes in the mechanical properties of the model. For the duration of each simulation, we record the velocity history at a number of points on the free surface of the model. Numerical simulations are carried out in FLAC3D. First, we look on how the distance between the source and the monitoring points changes what we record. We examine two cases : In Case A, the monitoring stations and the buried object are at a distance less than 30 meters from the seismic wave source. In Case B, the monitoring stations and buried object are at a distance more than 30 meters from the seismic wave source. We apply spectral analysis to the resultant seismic velocity time histories as recorded at a number of monitoring stations at the free surface of the model. Our results for Case A show that an object can be detected at a monitoring station located directly above the object to a depth of 1-2 meters. Results for Case B show that an object can be detected at the monitoring station that is deployed directly above the object to a depth of up to 4-5 meters, and it can also be detected at neighbouring stations, at distances approximately equal to the depth of the object. In addition, we study factors having an impact on the amount of energy of the seismic wave emitted, i.e. depth of the object from the surface and its mechanical properties. Our analysis indicates that by increasing the depth of the object, the amount of reflected seismic energy decreases. The changes in the mechanical properties of the materials lead to a change in seismic wave propagation velocity and frequency. Results from our numerical simulations present evidence that microseismics can be used as a complementary, low-cost site investigation tool for applications where very shallow depths are of particular interest such as those at brownfield sites. This can have significant implications on the way site investigations on brownfield sites are carried out, with microseismics providing an alternative to sites where traditional non-intrusive methods such as GPR and/or resistivity tomography are limited due to ground properties.

How to cite: alfakri, G.: Microseismic monitoring of wave propagation through heterogeneous media: a tool for brownfield site investigation?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8304, https://doi.org/10.5194/egusphere-egu23-8304, 2023.