Uncertainty quantification of geological model parameters in 3D gravity inversion by Hessian informed Markov chain Monte Carlo

Geological modeling has been widely adopted to investigate underground geometries. However, modeling processes inevitably have uncertainties due to scarcity of data, measurement errors, and simplification of modeling methods. Recent developments in geomodeling methods have introduced a Bayesian framework to constrain the model uncertainties by considering additional geophysical data into the modeling procedure. Markov chain Monte Carlo (MCMC) methods are normally used as tools to solve the Bayesian inference problem. To achieve a more efficient posterior exploration, advances inMCMC methods utilize derivative information. Hence, we introduce an approach to efficiently evaluate second-order derivatives in geological modeling and introduce a Hessian-informed MCMC method, the generalized preconditioned Crank-Nicolson (gpCN), as a tool to solve the 3D model-based gravity Bayesian inversion problem. The result is compared with two other widely applied MCMC methods, random walk Metropolis-Hasting and Hamiltonian Monte Carlo, on a synthetic three-layer geological model. Our experiment demonstrates that superior performance is achieved by the gpCN, which has the potential to be generalized to more complex models.

Geological Modeling Technology and Application Based on Seismic Interpretation Results under the Background of Artificial Intelligence

The development of seismic technology has made seismic data to be widely used in the interpretation of stratigraphic sequence frames, reservoir identification, fluid detection, and other research fields involved in reservoir description. The 3D technology reservoirs have always been the focus, as well as difficulty, of research. With the rapid development of information technology and the continuous improvement of seismic exploration level, people have put forward higher requirements for the accuracy of seismic data interpretation results. Aiming at the large number of structural and unstructured data in seismic, logging, geology, and other disciplines involved in seismic interpretation, how to effectively organize and coordinate analysis to discover the hidden reservoir structure and oil and gas distribution information has always been a geological and important topic for information processing technicians. This thesis is aimed at the current high-water-phase development of Shengtuo Oilfield reservoir and the problems existing in geological research. Based on seismic structural interpretation and attribute analysis, this paper analyzes the reservoir structural characteristics, sedimentary characteristics, and reservoir physical parameter characteristics based on geology, logging interpretation, core analysis, drilling, and seismic interpretation. Using the kriging method with external drift can cooperate with seismic variables to establish a reservoir geological model to study the Shengtuo Oilfield reservoir. We combine artificial intelligence technology with geological modeling technology of seismic interpretation results to explore the best method for predicting earthquakes. The research results in this paper show that the relative error of the model established by the kriging method in the article is relatively small for thinning wells, mainly concentrated around 1%. Examination of the thinning wells of 45 wells shows that the model established is basically good and the example has high accuracy. The research results in this paper have a guiding study of distribution and tapping potentials in the study area, formulating reasonable development and adjustment plans and improving oil recovery.

Resource-environment joint forecasting using big data mining and 3D/4D modeling in Luanchuan mining district, China

The Fourth generation industrial age and 5G + intelligent communication in the "Fourth Paradigm of Science" in the 21st century provide a new opportunity for research on the relationship between mining development and environmental protection. This paper is based on the theory of metallogenic geodynamics background, metallogenic process and quantitative evaluation and chooses the Luanchuan district as a case study, using deep-level artificial intelligence mining and three/four-dimensional (3D/4D) multi-disciplinary, multi-parameter and multi-scale modeling technology platform of geoscience big data (including multi-dimensional and multi-scale geological, geophysical, geochemical, hyperspectral and highresolution remote sensing (multi-temporal) and real-time mining data), carrying out the construction of 3D geological model, metallogenic process model and quantitative exploration model from district to deposit scales and the quantitative prediction and evaluation of the regional Mo polymetallic mineral resources, the aim is to realize the dynamic evaluation of highprecision 3D geological (rock, structure, hydrology, soil, etc.) environment protection and comprehensive development and utilization of mineral resources in digital and wisdom mines, it provides scientific information for the sustainable development of mineral resources and mine environment in the study area. The research results are summarized as follows: (1) The geoscience big data related to mineral resource prediction and evaluation of district include mining data such as 3D geological modeling, geophysics interpretation, geochemistry, and remote sensing modeling, which are combined with GeoCube3.0 software. The optimization of deep targets and comprehensive evaluation of mineral resources in Luanchuan district (500 km2, 2.5 km deep) have been realized, including 6.5 million tons of Mo, 1.5 million tons of W, and 5 million tons of Pb-Zn-Ag. (2) The 3D geological modeling of geology, mineral deposit, and exploration targeting is related to the mine environment. The data of exploration and mining in the pits of Nannihu – Sandaozhuang – Shangfang deposits and the deep channels of Luotuoshan and Xigou deposits show a poor spatial correlation between the NW-trending porphyryskarn deposits and the ore bodies. The NE-trending faults are usually tensional or tensional-torsional structures formed in the post-metallogenic period, which is the migration pathway of hydrothermal fluid of the related Pb-Zn deposit. There is a risk of groundwater pollution in the high-altitude Pb-Zn mining zones, such as the Lengshui and Bailugou deposits controlled by NE-trending faults are developed outside of porphyry-skarn types of Mo (W) deposits in the Luanchuan area. (3) Construction of mineral resources and environmental assessment and decision-making in intelligent digital mines: 3D geological model is established in large mines and associated with ancient mining caves, pit, and deep roadway engineering in the mining areas to realize reasonable orientation and sustainable development of mining industry. The hyperspectral database is used to construct three-dimensional useful and harmful element models to realize the association of exploration, mining, and mineral processing mineralogy for the recovery of harmful elements (As, Sb, Hg, etc.). 0.5 m resolution Worldview2 images are used to identify the distribution of Fe in the wastewater and slag slurry of important tailings reservoirs, so as to protect surface runoff and soil pollution.

Seismic Reservoir Characterisation of Tyumen Formation in Frolov Megadepression

The article discusses the approaches and methods to study the Middle Jurassic deposits of the Tyumen Formation within the Frolov megadepression (West Siberian oil and gas province), which have high hydrocarbon potential. The materials refer to several areas with available 3D seismic data and several dozen oil wells. The problems of seismic interpretation and its application for geological modeling are considered. We also propose several ways to overcome them.