Gravity measurements contain important information about the subsurface at every spatial scale. Satellite missions provide highly accurate, global measurements, that allow building and testing 3D models of the Earth even for regions where other geophysical measurements are sparse. Depending on the envisaged model resolution and application, gravity data from regional or local surveys (e.g. ship and airborne) can be incorporated to provide higher resolution. However, solutions to the gravity inverse problem are non-unique and constraints by independent geological and geophysical observations need to be integrated to develop reliable subsurface density models. Such models, in turn, provide insights into subsurface variations in composition and thermo-mechanical state.In this session, we welcome contributions that integrate gravity and/or gradient data (at all scales) with other geophysical measurements and geological information to better understand the structure, properties and processes of the Earth’s subsurface. Hence, we would like to discuss diverse applications, from global scale, where mantle properties are assessed, down to the reservoir scale. We also welcome more technical contributions that address related topics like innovative joint inversion methods, model uncertainty estimation and determinations of parameter sensitivities, as well as temporal variations of the gravity field, and new interpretation software and techniques.
4:15pm - 4:45pm
Probabilistic Machine Learning for improved Decision-making with 3-D Geological Models
1RWTH Aachen University, Germany; 2Terranigma Solutions GmbH, Aachen, Germany; 3Staatstoezicht op de Mijnen, Den Haag, Netherlands
Geological models, as 3-D representations of subsurface structures, can be combined with gravity inversions to obtain geometric representations of geological objects with similar porperty distributions. These models are built on prior assumptions and imperfect information, and they often result from an integration of geological and geophysical data types with varying quality. These aspects result in uncertainties about the predicted subsurface structures and property distributions, which will affect the subsequent decision process.
We discuss approaches to evaluate uncertainties in geological models and to integrate geological and geophysical potential-field information in combined workflows. A first step is the consideration of uncertainties in prior model parameters on the basis of uncertainty propagation (forward uncertainty quantification). When applied to structural geological models with discrete classes, these methods result in a class probability for each point in space, often represented in tessellated grid cells.
A logical extension is the integration of geological forward operators into geophysical inverse frameworks, to enable a full flow of inference for a wider range of relevant parameters. We investigate here specifically the use of probabilistic machine learning tools in combination with geological and geophysical gravity and magnetic modeling. Challenges exist due to the hierarchical nature of the probabilistic models, but modern sampling strategies allow for efficient sampling in these complex settings. We showcase the application with examples combining geological modeling and geophysical potential field measurements in an integrated model for improved decision making.
4:45pm - 5:00pm
Improving gravity inversion by geostatistical simulation of constraining data - case study: southern Africa crustal thickness model
1Technische Universität Bergakademie Freiberg, Germany; 2University of Twente, Enschede, Netherlands
The inversion of gravity data for crustal thicknesses is a nonunique problem. Therefore, additional independent information (e.g., seismic data) is needed to constrain the inversion process. Despite decades of exploration efforts related to mining and the installation of more seismic stations, knowledge on the deep crustal structure of southern Africa remains limited.
In this contribution we present a crustal thickness model for southern Africa: The initial model is determined by inversion of satellite gravity data. Here, we apply seismically constrained non-linear inversion, based on the modified Bott's method and Tikhonov regularization assuming spherical Earth approximation. The inversion hyper-parameters are determined by Monte-Carlo-Marcov-Chain (MCMC) simulation. The data quality of the (active and passive) seismic constraints is high in general, showing e.g. individual uncertainties per point. The problem is that the constraining data points are irregularly distributed, resulting in large areas without constraints. Therefore, in a next step, we want to validate and improve the modelling result for these unconstrained regions.
We use the initial constraining data set to geostatistically simulate a homogeneous crustal thickness model for the investigation region. For this, we apply a Sequential Gaussian Simulation (SGS) based on Ordinary Kriging that includes the uncertainties of the seismic data and allows to characterize uncertainties of the simulated points. The simulated crustal thickness model is then used to qualitatively validate the inversion result. Additionally, we redo the inversion process with a new constraining data set that combines the preexisting constraining points and the simulated model.
5:00pm - 5:15pm
Thermo-compositional models of the West Gondwana cratons
1GFZ Potsdam, Germany; 2Free University Berlin, Germany; 3Schmidt Institute of Physics of the Earth, RAS, Moscow, Russia; 4University of Trieste, Trieste, Italy.; 5University of Utrecht, Utrecht, Netherlands; 6US Geological Survey, Menlo Park, USA.
When Western Gondwana broke apart into the South American and African continents ⁓ 120 Ma ago, some of its cratons were broken apart as well. Following the isopycnic hypothesis, their long-term stability and often neutral to positive buoyancy can be explained by the counteracting effects of cooling (density increase) and iron depletion (density decrease). To separate these effects, we created the presented models following an iterative integrated approach using mainly seismic and gravity data. In the first step, seismic models of Depth to the Moho were created to allow correction of the gravity field and calculation of the residual topography. Second, based on mineral physics and S-wave tomography, we assessed temperature variations in the uppermost mantle and subtracted their effects from both residual gravity and topography. Afterwards, a joint inversion enables determination of potential compositional variations. Adapting the initially juvenile mantle composition leads to a change of thermal effects, thus the process was repeated iteratively until convergence. In result, we obtained self-consistent models of temperature, thermal and compositional density variations, and #Mg, a measure of iron depletion. Our results show deep depleted cratonic roots under the Amazonas, São Francisco, Paranapanema (South America), West African, Northern to Central Congo and Zimbabwe Cratons (Africa). Depletion appears to be mostly absent in the Rio de la Plata Craton of South America and its proposed African counterpart, the Southern Congo Craton as well as the Kaapvaal Craton below 100 km depth and the Tanzania and Uganda Cratons.
5:15pm - 5:30pm
Integrated 3D gravity and geological modelling in the Subhercynian Basin (Germany) – A modelling strategy for the enhanced study of the basins sedimentary and crustal setting
Landesamt für Geologie und Bergwesen Sachsen-Anhalt, Halle(Saale), Germany
3D geological modelling in highly complex areas with sparse or ambiguous information is affected be conceptual uncertainty, which can be significantly reduced by the integration of gravity data. However, gravity modelling itself underlies the non-uniqueness problem, indicating that there is more than one model consistent with the observed gravity field. Therefore, cross-validation of gravity models by integration of regional geologic concepts, geometric and kinematic construction and restoration techniques helps solving this problem.
In this regard, we defined an integrated modelling strategy, which starts with extracting a-priori information from geological maps, 2D seismics, borehole and gravity data, which were independently analysed and conservatively interpreted; i.e. non-unique solutions were completely avoided. Subsequently, geologic interpretations were combined with gravity data, which was analysed by use of gradient calculation and 2D-EULER deconvolution. The resulting combined dataset was validated by use of 2D cross-section balancing techniques considering bed-lengths and area consistency. The resulting serial balanced cross-sections served as solid basis for a 3D gravity modelling.
Our integrated workflow was tested for the less-explored eastern part of the Subhercynian Basin (Saxony-Anhalt, Central Germany). We show that the combination of independently ambiguous data holds the potential to generate new insights into the local fault system, the topography of the crystalline basement (transition of the Mid-German Crystalline Rise and Rhenohercynian Zone) and outlines of salt structures as well as the setting of the base Cenozoic. Furthermore, modelling of long wavelength gravity anomalies provides new information on the crustal setting at the margins of the North German Basin.
5:30pm - 5:45pm
Structure and density configuration of Germany’s subsurface: 3-D-Deutschland, an updated three-dimensional lithospheric-scale model
1Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences, Potsdam, Germany; 2Christian-Albrechts-Universität zu Kiel, Kiel, Germany; 3RWTH Aachen University, Aachen, Germany
With this study we revise and improve the three-dimensional lithospheric-scale structural and density model of Germany (3-D-D). Major shortcomings of this model resulted from joining three regional 3-D models that were poorly covered by data at their margins. Merging into a larger model revealed structural inconsistencies in these “marginal” domains. In order to resolve discrepancies between the units in a more consistent way we integrate newly available data from seismic reprocessing, tomography, and use 3-D gravity modelling to improve the fit between the modelled and observed gravity.
The recently initiated reprocessing of the DEKORP seismic profiles, for example, in the region of the Rhine Graben by the federal geological survey of Hesse (Bär et al. in prep.), indicate that previous assumptions on sediment thickness in certain regions, as well crustal structure and Moho depth need to be revised. We integrate this new structural information together with density variations derived from mantle seismic tomography (CSEM Europe, Fichtner et al., 2018; LSP_Eucrust1.0, Lu et al. 2018) and analyse the updated density distribution against the more detailed Bouguer gravity anomaly map of Germany (Skiba, 2011).
The update of the 3-D-D model is important for ongoing research in seismic hazard assessment in that it serves as a basis for thermal and rheological modelling helping to relate observed seismicity with spatial variations in strength. The model provides a data-consistent background for regional studies on sustainable use of geothermal energy and on the suitability of sites for the underground storage of radioactive waste or of CO2.