We have developed a deep-learning method based on the neural network of the feedforward type to estimate the depth to the basement from potential fields. The data used to train and test the network are related to the Bishop synthetic model. A trial-and-error approach was used to find the hyperparameters that have the best compromise between performance and computation time. The training was organized by associating the depth values of the basement to the data through a moving window, running along profiles in the north–south and east–west directions. In this way, we generated a set of approximately 296,980 examples. We verified the robustness of the trained net by carrying out a test related to another synthetic model, extracted from the Himalaya digital elevation model. The inherent ambiguity of the problem led us to test two hypotheses for the estimation of the basement depth, the first related to a priori information on the density contrast and the shallowest depth and the second assuming the knowledge of the depth at least at two points, but not that of the density contrast. In these cases, our data-driven approach yielded interesting results leading to estimate the maximum depth in the first case and the density contrast in the second one. We finally applied the method to the isostatic anomaly of the Yucca Flat sedimentary basin, Nevada. The results are consistent with previous interpretations of the area, which were based on gravity inversion methods.
12 May 2019 in Open Journal of Mathematical Sciences
Summary: Gravity data inversion is a fundamental tool for geological exploration. A large amount of algorithms have been developed in the past, with different approaches. The choice of the inversion algorithm depends, mainly, on the geological contest, the kind of solution desired, its resolution at the end of the process, the availability of a priori information and on how they can be included in the inversion algorithm. A priori information has, often, a key-role in the inversion process. However, in case of reconstruction of salt and sub-salt structures, seismic information may be poor and unfeasible. So we need to constrain the model with other kind of information. In this work, we present a part of recent Eni R&D activity focused on gravity data inversion. The shown results are mainly based on the Data Space Inversion algorithm (Pilkington 2009), originally presented for the magnetic case, and here extended to the gravity problem. We also putted a strong effort on the computational side of the problem, taking advantage from our experience in HPC (High Performance Computing), in order to speed up the inversion process and so enable its use at industrial level. In this paper we present and discuss some results regarding the application of the methodology to the SEAM (SEG Advance Modeling, 2007) demonstrating that the algorithm allows a consistent depth and density model.
Exploring difficult terrains with complex geology is challenging for any technology, particularly in frontier areas where regional reconnaissance is necessary. Potential field interpretation represents a strong support to extract correct geological information, in addition to that from geological and seismic data. We aim at using this information to build structural models related to the tectonic framework and basement configuration. Many exploration studies are based on source edge mapping analysis from gravity and magnetic data and several semiautomatic methods, based on the use of derivatives of the potential fields, have been developed. Most of these methods have been codified in commercial software, very intensely used in Oil & Gas Industry for exploration studies. We compare six methods of edge detection in the gravity case. Even though the success of each method depends on the quality of data and on the specific geological setting, co-locating the solutions from different methods may often provide an increased confidence in the reliability of a given edge location. A multiscale approach is however decisive in terms of meaningful mapping of source edges, since it provides the finest separation of edges from the largest to the shallowest of the allowed scales.