Predict part 2: Building a Geo-Database and a 3D subsurface model for the historical center of Rome

Immersive visualization of 3D subsurface ground model developed from sparse boreholes using virtual reality (VR)

With the rapid development of computing and digital technologies recently, three-dimensional (3D) subsurface models for accurate site characterization have received increasing attention, for example, with various data-driven methods developed for 3D subsurface modeling. This leads to a need for validating the 3D modeling results obtained from each method and comparing the performance of different methods in a fair and consistent manner. To address this need, a benchmarking study, which is often used in machine learning (ML), is presented in this study to compare the performance of different 3D subsurface modeling methods in four aspects, including accuracy, uncertainty, robustness, and computational efficiency. A suite of performance metrics is proposed for the four aspects above. Multiple sets of real cone penetration test (CPT) data are compiled in the benchmarking study for quantifying performance of 3D modeling methods using sparse measurements as input, a typical scenario in geotechnical practice. The benchmarking study is illustrated using an in-house software package called Analytics of Sparse Spatial Data based on Bayesian compressive sampling/sensing (ASSD-BCS), which can directly generate high-resolution 3D random field samples (RFSs) from sparse measurements. The evaluation results show that ASSD-BCS provides accurate estimates with quantified uncertainty from sparse measurements. In addition, ASSD-BCS exhibits remarkably high computational efficiency and performs robustly under different benchmarking cases.

Understanding the reservoir conditions through 3D subsurface modeling is the key to optimize the exploration stage in geothermal field. A calibrated reservoir model based on updated data can be very important for this process. The main challenge of reservoir characterization in a geothermal field is the lack of subsurface data, therefore surface data are useful for reservoir modeling. This study utilized Sorik Marapi geothermal field data as a reference for reservoir modeling. This field is one of the geothermal fields in Indonesia that has been recently drilled, with results indicating the existence of a high temperature-neutral acidity resource. Initial reservoir model has been built from the previous study to create conceptual 3D subsurface model which includes structural, lithology, resistivity, and temperature distribution from surface exploration data, including surface mapping, remote sensing image interpretation, the magnetotelluric method, and subsurface data from six wells data. The objective of this paper is to calibrate the initial reservoir model with information from an additional ten new wells data to improve delineation for updated reservoir area in the field. Software that allowed multidisciplinary data integration from surface to subsurface information was used for the calibration of the initial 3D model. The workflow to calibrate the model started with data loading and quality control, preparing the old 3D model and comparing it to new well data, analyzing the comparison, and updating the 3D model. Finally, the new delineation of reservoir zone can be determined. The result of this study is an updated 3D subsurface static model defining the vertical and lateral reservoir boundaries, as well as the prime resource areas, which would be the basis for designing future well targets, and parameters for a dynamic reservoir model. The same model can be expanded to construct the numerical model to match the natural state condition of the field and make forecasts of the future reservoir behavior at different operating conditions. The main properties of the updated 3D model are lithology and temperature, which are important in geothermal reservoir delineation.

We present a new method to incorporate paleo-surface temperature effects in steady state 3D conductive temperature models. The workflow approximates the transient effects and incorporates these into steady state models, using appropriate source and sink terms for radiogenic heat production. This allows for rapid models, which can be easily used in ensemble approaches for data assimilation of high-resolution temperature models for geothermal resource assessment. The workflow is demonstrated for the Netherlands which is a sedimentary basin with a wealth of deep (temperature) data from groundwater and oil and gas wells and past studies on the 3D temperature distribution. 3D subsurface temperature models of the Netherlands ranging up to 5 km depth systematically overpredict temperatures at shallow (<1500 m) depth. Analysis of both shallow (<600 m, >200,000 measurements) and deep temperature measurements (1–6 km, >1500 measurements), clearly demonstrates a shallow thermal gradient in over 200 locations of ~20 °C km −1 for the top 400 m underlain with a deep geothermal gradient of ~31 °C km −1 for the 2–4 km interval. Improvements in 3D subsurface modelling regarding the shallow part are accomplished by adding a paleo-surface temperature correction related to glaciation effects of the Weichselian glacial period. This paleo-surface temperature correction proved to be the missing link between two distinctive geothermal gradients observed and is consistent with earlier findings for limited datasets. The consistent overprediction of modelled temperatures in 74% of locations for the top 2 km which are regularly distributed over the Netherlands demonstrates that the influence of paleo-surface temperatures is rather uniform over large areas and not significantly overprinted by other effects such as groundwater flow. The updated model, marked by up to 10 degrees cooling compared to models ignoring the paleo-surface temperature effects, has major implications for assessing geothermal resource potential up to 2 km depth. • Novel approach for modelling paleo surface temperature effects in thermal models for sedimentary basins. • High resolution integrated deep and shallow database for validation of thermal effects of paleo surface temperatures. • Improved thermal model shows excellent fit with observations.

Biwako-seigan fault zone is one of the most active faults in Japan, located around western shore of Lake Biwa. According to The Headquarters for Earthquake Research Promotion (2004), its probability of earthquake within 30 years is from 0.09% to 9%. To simulate strong motion, caused by the rupture of Biwako-seigan fault zone, we have constructed 3D subsurface structural model around Lake Biwa. The 3D subsurface structural model is constructed using the data of geophysical and geological surveys in and around Lake Biwa. We have modified the model by comparing the observed H/V spectral ratio between earthquakes and theoretical one. Using the modified 3D subsurface structural model, we simulated ground motions for moderate earthquakes. As the results, simulated waveforms of ground velocity agree well with observed ones. By the geological method and adjusting theoretical H/V spectral ratio to observed one, it is possible to re-construct 3D subsurface structural model which is applicable to strong motion simulation, even though there are few data of geophysical surveys.

Although geomorphic evidence and shallow geometry of active faults are significant for the understanding and assessing of fault activity and seismic hazards, it is challenging to acquire high-resolution topographic data and shallow geometry of the Yushu fault by conventional methods. Here, we present a case study to reconstruct the detailed surficial and subsurface geometry of the Yushu fault using terrestrial laser scanning (TLS), multi-frequency ground penetrating radar (GPR) and trenching. TLS was suitable for measuring the high-resolution three-dimensional (3D) topographic data of the fault. GPR surveys with different frequency antennas (25 MHz, 100 MHz, 250 MHz and 500 MHz) were conducted to image the shallow geometry of active faults at different depths and spatial resolutions. The typical groove landscape, parallel to surface traces of the fault, was clearly observed on the TLS-derived data. A ~40 m width narrow fault system and three faults were identified on the different frequency GPR profiles. Furthermore, faults F1 and F2 were supposed to be boundary faults but were sinistral-lateral strike-slip faults with a normal component, while fault F3 was inferred as the secondary fault. The western trench section, despite the limited investigation depth (~2 m), was well consistent with the 500 MHz GPR result, especially in the location of fault F2. Finally, a 3D surficial and subsurface model was established from the TLS-derived data and GPR data offering multi-sensor and multi-view spatial data to characterize and understand the fault’s kinematics and characteristics. In addition, the shallow geometry of the fault on the GPR results would be better interpreted with the help of the corresponding surficial data. The study results demonstrate that a combination of TLS, multi-frequency GPRs and trenching can be successfully used for reconstructing a detailed surficial and subsurface geometry of the Yushu fault. It will play an increasing role in comprehensive understanding and assessing fault behavior and seismic hazards, especially on the Tibetan Plateau and the adjacent area.

Started as a project at a geoscientific hackathon in 2018 and released to the public in 2020, GST[AR] is an app-based attempt of utilizing Augmented Reality (AR) to bring 3D geological data to almost everyone with a smartphone or tablet. In this way, multiple european-based geological surveys already offer some of their models to experts and interested users alike today. 3D subsurface models especially are great for public engagement and education as they are easier to grasp and fun to interact with. GST[AR] joined the growing list of tools that allow users to visualize geological data without the need for expensive and proprietary software, but chose to do it with the rather novel technology of AR. AR holds great potential since it is a fun and intuitive way to interact with 3D data and is readily available on most portable devices. Enabling&amp;#160;users to directly manipulate a 3D scene is essential for user engagement, but also for gaining a deeper understanding of the spatial relations and dimensions. Simpler methods like creating an animation, or still image, of a 3D model fall short in that regard because they lack the interactive component. Compared to AR, Virtual Reality is allowing for a much higher level of immersion, but it also comes at the cost of a more convoluted and expensive setup and user isolation. GST[AR] offers the user a list of multiple data sources, providing additional background information for some of them. After a model and a set of features have been selected, the downloaded 3D model can be placed in AR. The app then gives the user the&amp;#160;means to scale or rotate the model, and even to look "below the surface" by highlighting individual parts. It is also possible to share a session with multiple peer devices to view the same model in the same physical space and spark a conversation. In these sessions every user is able to manipulate the model or place down markers to make sure that all peers know what specific part is being discussed at the moment. While a connection to a running instance of GST Web&amp;#160;to download subsurface models was required in the past, a tool has been developed that allows everyone to convert input data into an app friendly state and host it on their own machine. At the point of writing this abstract, this is limited to GoCad and Wavefront (OBJ) input files, but the plan is to expand that list in the future. A new way of opening models within the app by means of simply scanning a QR code aims to make it easier and faster to engage with potential users. In this presentation we will look into&amp;#160;the capabilities of the app, ways for everyone to utilize their own models, and the potential this holds for conferences and education.

All terrestrial environmental processes involve the soil including hydrological, geological, meteorological, ecological and anthropological factors. The stress on natural soils increases significantly with agricultural and urban land use as a consequence of the world‘s growing population. This leads among others to deforestation and soil sealing and consequently to the decrease of natural habitats and resources. Thus, protection and preservation of soil and the maintaining and avoiding of negative affects from land-use change marks is a challenge for the policy, agricultural economics, and the geosciences. A perspective for maintaining and avoiding extensive and destructive land use is given by the optimization of current land use. Therefore reliable information of soil and subsoil properties are needed. However, direct analysis of crucial soil properties, e. g. grain size or soil moisture is still time consuming and costly, and provides only single point information. But in particular soil data for medium and large-scale areas are needed for the assessment of soil development and future-oriented planning. Proximal soil sensing techniques (PSS) offers an opportunity for obtaining data from medium and large–scale areas time and cost efficient. However, all PSS methods response only indirectly to the relevant soil properties and could be affected by several soil properties. Hence, a recent challenge is the improvement of PSS data evaluation and interpretation. The presented PhD thesis addresses the improvement of data evaluation and interpretation of the PSS methods electromagnetic induction (EMI) and gamma spectrometry (GS) at three different test sites and three different problems. For each problem an individual adjusted approach was develop, applied and critical discussed. In Part I the study consider the moisture distribution at a land slide affected hill slope in Austria by means of EMI. The presented study monitored the temporal, spatial and vertical behavior of soil-moisture distribution at a previously identified dynamic slope area over a period of nine month. By the assumption of relative temporal stability of soil properties, seasonal changes in measured electric conductivity (EC) should originate from soil moisture content. This study also faces the challenge of shifts in absolute EC values resulting from different calibration situation or different EMI devices and provides an opportunity for comparability of different EC data. By this approach an delineation of soil moisture pattern was successful derived. Part I also explores the visualization of temporal changes in three-dimensional subsurface data. Part II focuses the problem of synthesis and simplification of multilayered input data of a floodplain in Central Germany towards a clear delineation of surface. This study tackle the problem by a K-means cluster algorithm in order to generate a 2dimensional map from the test site that includes the main characteristics from divergent input data. However do the generated cluster partitions really reflect the main characteristics of soil properties? Hence, this study also addressed on the reliability of such cluster maps by validation of independent soil properties such as grain-size, thickness of soil layers, and the color. The results show that not all partitions can be confirmed by independent soil samples; one of three clusters significantly differs from the others, the other two clusters could not confirmed by the considered parameters. Part III investigates a floodplain of a low-mountain river in Switzerland in order to detect the ancient active stream channels (AAC). This part is subdivided into two approaches; first a 3D subsurface model as a result from the iterative inversion of predicted EMI values was generated. Thereby various electric conductivity maps (EC) were generated by forward modeling and compared with the corresponding measured data. The study use the best fitted input data for the generation of the 3D model. In a second approach a K-means cluster map for the floodplain surface was generated that combines the main characteristics from multilayered subsurface data by synthesis and simplification analogue to Part II. The obtained cluster characterizes different soil conditions, which are indicative for the delineation of AAC. Although developed under specific site conditions all demonstrated approaches offers portability and should be applied in other applications.

Subsurface characterization and clear distinction between layer boundaries in geological profiles are essential for successful completion of engineering projects. Due to limitations in accessing the vast and diverse subsurface information, efficient management of geotechnical data is of special importance. This study presents the methodology of building a digitally formatted and integrated spatial database using geotechnical data and geographic information system (GIS). The development of comprehensive geotechnical (geo)-database involves (1) Collection of borehole data from various reputed sources, (2) Validation of data in terms of accuracy and redundancy and (3) Standardizing and organizing the geotechnical information for incorporating into the database. The database is then integrated with GIS to provide the advantage of visualizing, analysing and interpreting the geotechnical information spatially. In addition, stratigraphic data stored within the spatial database is utilized for constructing three-dimensional (3D) subsurface models. A simplified approach is developed for 3D subsurface modelling with a combination of the Relational database management systems (Microsoft Access), Arc-GIS and Spatial interpolation techniques. The proposed methodology is illustrated using nearly 175 borehole data collected from various projects aimed at providing an exhaustive database and an integrated 3D environment for the Chennai City of south India for better interpretation and decision-making in regard to sustainable urban development. The study also demonstrates the application of the developed subsurface model in evaluating the effectiveness of multichannel analysis of surface wave (MASW) tests to estimate the depth to bedrock, which will be of immense use in foundation studies and in ground response analysis.

Building a geodatabase for mapping hydrogeological features and 3D modeling of groundwater systems: Application to the Saguenay–Lac-St.-Jean region, Canada

Three-dimensional (3D) subsurface modelling and visualization of seven regional unconformity-bounded stratigraphic units, (synthems) of the Messinian-Quaternary succession of the Po Plain Basin and Savigliano-Alessandria Basin are here described. 3D model is developed using the EnterVol Ctech® software environment, grounding on information stored in GeoPiemonte Map database. This modelling allows to visualise the geological data and to establish topological relationships between the analysed objects, coupling the data processing capabilities of GIS with 3D modeling and sharing the result on 3D WebGIS service of ARPA Piemonte Geoportal.

Project Hydropower System Senj 2 (HPS Kosinj / HPP Senj 2) is planned to utilize the remaining hydropower potential of the Lika and Gacka watershed by upgrading the existing Senj hydropower system (HPS). As part of the HPP Senj 2, the construction of the Gusić polje 2 Compensation Basin is planned. It is formed by construction and reconstruction of about 3664,06 m long lateral embankments, active storage of 2.87 million m³ used for daily inflow regulation. The water side of the embankments and basin are waterproofed using the same techniques and the same materials – geomembrane. For this type of technical solution, it is important to calculate settlements of basin and embankments foundation soil. Over the years, geotechnical investigations, geophysics and exploratory boreholes, were carried out in the area. Spatial distribution and stratums thicknesses on karst landscapes is very difficult to estimate, and there is not one estimation technique that can resolve that task. In this paper, we examined application of machine learning to jointly interpret geophysical and borehole datasets for modelling subsurface spatial model of investigated area. Borehole data are used as hard and geophysical as soft data. Geophysical data are interpolated and borehole data associated these interpolated values. Furthermore we present machine learning models for classification using interpolated geophysical data, with basic model metrics. Classification models are applied to generate a 3D subsurface model of geotechnical stratums. Using 3Dsubsurface model, paired with mechanical properties, we can generate settlement map for the case of full reservoir. As well, the result is corresponding spatial uncertanty model, that highlights highest uncertanty areas that can be subject of further geotechnical inveatigation works. The proposed method proved to be suitable to jointly interpret database from common geotechnical investigation works in civil engineering practice. It is also suitable for identifying areas that need further investigation.

Large volumes of glacial meltwater drain along the interface between the ice sheet and its bed, thereby influencing glacier dynamics. It is known from the geological record and modern glacial systems that channelized subglacial meltwater discharge generates high erosion rates, leading to the formation of overdeepenings and tunnel valleys, some over 500 m deep. It is therefore essential to constrain the depth of potential subglacial erosion under future ice sheets when searching for locations of high-level radioactive waste repository sites.The aim of this project is to quantify the meltwater-driven erosion under the past ice sheets in northern Germany and evaluate the erosion potential during future glaciations. To achieve this goal, we develop a next-generation dynamic numerical model simulating subglacial meltwater erosion on soft beds. In the first step, we built subsurface reservoir models at different scales and resolutions to examine the impact of model resolution on the subsequent erosion modelling. The 3D subsurface model approximately covers the area of the Northwest German Basin (40,000&amp;#160;km&amp;#178;), has a depth of 2000 m, and comprises lithostratigraphical units from the Permian Zechstein to the Pleistocene. The basin fill has a complex structure due to salt tectonics, and the main challenge was to generalize the complex lateral and vertical lithofacies/hydrofacies relationships.Two large-scale, low-resolution subsurface reservoir models were constructed. The first model does not include Quaternary deposits. This model was created to simulate the formation of Middle Pleistocene tunnel valleys and compare/validate the results with the Pleistocene record of the Northwest German Basin. The second large-scale subsurface model includes the Quaternary deposits and will be used to simulate subglacial erosion during future glaciations. A smaller high-resolution subsurface model, covering an area of about 2000&amp;#160;km&amp;#178;, will then be used to test the effects of model size and model resolution on the simulation of subglacial erosion.

Subsurface geological-geotechnical modelling to sustain underground civil planning

Unforeseen ground condition often leads to the occurrence of cost overrun and project delay in Malaysia’s architecture, engineering, and construction (AEC) industries. The implementation of Building Information Modelling (BIM) for subsurface was attempted to reduce uncertainties associated with the underground conditions. A pilot case study was performed to understand the 3D subsurface modelling workflow in BIM environment. A construction site in Ara Damansara, Petaling Jaya, Malaysia was selected for the pilot case study. The site investigation data of the study area were interpreted and input into AutoCAD Civil 3D with geotechnical module to perform the 3D subsurface modelling at various detail levels. The workflow of the modelling process in case study was documented for future references. Through this pilot study, it was found that the adoption of BIM for subsurface modelling could produce soil profiles for different engineering purposes such as visualisation of subsurface conditions during project initiation stage, preliminary conceptual design, and detailed analysis and design of geotechnical structures, particularly foundation. The BIM implementation for geotechnical data improves the management of site investigation data, enhances the visualisation of subsurface profile, and improves the efficiency in communicating the subsurface conditions to all stakeholders in a project team.