A5056559949- Seismic Imaging and Inversion Techniques
- Reservoir Engineering and Simulation Methods
- Hydrocarbon exploration and reservoir analysis
- Underwater Acoustics Research
- Seismic Waves and Analysis
- Geophysical Methods and Applications
- Geological Modeling and Analysis
- Geology and Paleoclimatology Research
- CO2 Sequestration and Geologic Interactions
- Geothermal Energy Systems and Applications
- Enhanced Oil Recovery Techniques
- earthquake and tectonic studies
Instituto Superior Técnico
2021-2024
TARH (Portugal)
2024
University of Lisbon
2021-2023
GFZ Helmholtz Centre for Geosciences
2021
Istituto di Scienze Marine del Consiglio Nazionale delle Ricerche
2020
Accurate multivariate parametrization of subsurface properties is essential for characterization and inversion tasks. Deep generative models, such as variational autoencoders (VAEs) adversarial networks (GANs), are known to efficiently parametrize complex facies patterns. Nonetheless, the inherent complexity modeling poses significant limitations their applicability when considering multiple simultaneously. Presently, diffusion models (DM) offer state-of-the-art performance outperform GANs...
Predicting groundwater recharge, storage, and transport in mountain environments is challenging due to high spatiotemporal variability limited data. In this context, we report on a time-lapse Electrical Resistivity Tomography (ERT) survey conducted an Alpine catchment October 2024 involving measurements before after major rainfall event. Hydraulic head (yearly variations of ~40 m) water temperature (decreasing long-term trend) have been monitored since 2010 the Val d’Ursé...
Abstract Predicting the subsurface spatial distribution of geological facies from fullstack geophysical data is a main step in geo-modeling workflow for energy exploration and environmental tasks requires solving an inverse problem. Generative adversarial networks (GANs) have shown great potential geologically accurate probabilistic modeling, but existing methods require multiple sequential steps do not account uncertainty facies-dependent continuous properties, linking to observed data....
Accurate prediction of the spatial distribution subsurface permeability is a fundamental task in reservoir characterization and monitoring studies for hydrocarbon production CO 2 geologic storage. Predicting over large areas challenging, due to their high variability anisotropy. Common approaches modeling generally involve deterministic calculations from porosity using precalibrated rock-physics models (RPMs) or geostatistical cosimulation methods that reproduce observed experimental...
The simultaneous prediction of the subsurface distribution facies and acoustic impedance () from fullstack seismic data requires solving an inverse problem is fundamental in natural resources exploration, carbon capture storage, environmental risk management. In recent years, deep generative models (DGM), such as variational autoencoders (VAE) adversarial networks (GAN), were proposed to reproduce complex patterns honoring prior geological information. Variational Bayesian inference using...
Abstract Understanding the thermal behavior of nonsteady state subsurface geosystems, when temperature changes over time, requires knowledge on speed heat propagation and, thus, rock's diffusivity as essential thermo‐physical parameter. Mixing models are commonly used to describe properties polymineralic rocks. A diffusivity‐porosity relation is known from literature that incorporates common mixing into equation and properly works for unconsolidated, clastic clayey, sandy marine sediments...
Geostatistical seismic rock physics amplitude-versus-angle (AVA) inversion allows the joint prediction of and fluid properties from reflection data. In these methods, model perturbation update occur iteratively in petrophysical domain. A facies-dependent precalibrated is applied to simulated calculate elastic properties. Synthetic data are computed models. The models calibrated at well locations act as link between domains, remaining unchanged during procedure: convergence geological...
Summary Predicting the spatial distribution of facies and collocated acoustic impedance (IP) in subsurface from fullstack seismic data is fundamental for assessing mineral energy natural resources potential. In recent years, deep generative models (DGMs) such as variational autoencoders (VAEs) adversarial networks (GANs) were proposed powerful methods to reproduce complex patterns, honoring prior geological data. Variational Bayesian inference using inverse autoregressive flows (IAF) can be...
Abstract Predicting the spatial distribution of geological facies in subsurface from fullstack geophysical data is a main step geo-modeling workflow for energy exploration and environmental tasks requires solving an inverse problem. Generative adversarial networks (GAN) have shown great potential geologically accurate modeling, although with limitations computational costs accounting uncertainty prediction facies-dependent properties. To overcome this limitation, we propose GAN architecture...
Summary Accurate predictions of the spatial distribution permeability in subsurface is fundamental reservoir characterization for several tasks (e.g., CO2 injection and storage monitoring or natural resources characterization). Nonetheless, modelling particularly challenging, due to its strong variability, anisotropy dependency on factors. The most common approaches involve deterministic estimations from rocks’ porosity, using pre-calibrated rock physics models, data-driven stochastic...