An interdisciplinary team of researchers from the Universidad Industrial de Santander (UIS) developed a mathematical-computational algorithm that improves the quality of seismic images in emerging basins, with the aim of reducing uncertainty in energy exploration processes, optimizing costs and providing key technological tools for the energy transition in Colombia.
The development is part of the project “Mathematical-computational modeling for the improvement of seismic images in emerging Colombian basins using deep learning with domain adaptation”, an initiative funded by the Vice Rector’s Office for Research and Extension of the UIS, which combines mathematics, geophysics and artificial intelligence to propose a new approach for cleaning post-stack seismic images.
Through deep convolutional neural networks, the team created an algorithm capable of identifying and removing noise present in seismic data, allowing more accurate visualization of geological structures that could contain hydrocarbons, minerals or geothermal energy sources.
The project was developed by an interdisciplinary team of researchers from the HDSP, GIRG and GIGBA research groups, with the support of UIS undergraduate and graduate students, under the direction of professors Henry Arguello Fuentes, from the School of Systems and Computer Engineering and director of the HDSP research group; José David Sanabria, from the School of Physics; and Francisco Velandia, from the School of Geology.
Why study emerging basins?
Emerging basins are structures of the Earth in formation or evolution that have not yet been explored in depth. They represent strategic opportunities for the exploration of subsurface resources, especially in poorly studied areas of Colombia.
According to UIS researchers, the country has a high potential in this type of regions, but available data are often contaminated with noise, which can hinder or distort geological interpretation and lead to costly errors.
“Misinterpretation of a seismic image can lead to erroneous drilling that costs millions of dollars. If we improve the quality of the images, we reduce those risks considerably. There are mature basins in the country with abundant geophysical and well information. We seek to take advantage of this data to train artificial intelligence models that can be applied in emerging basins, where this information does not exist,” explain the researchers.
University-industry alliance to promote energy solutions in the country.
This research responds to a strategic line proposed by the National Hydrocarbons Agency (ANH), which seeks to transfer the knowledge acquired in mature basins, areas already explored, to less studied regions, such as emerging basins. The purpose is to accelerate the identification of sustainable energy sources, aligned with the country’s energy transition.
“The idea is to use quality historical data in well-characterized areas to train our artificial intelligence models and then apply that knowledge in regions where there is very little information,” the research team explains.
The development had the technical support of the companies PETROSEIS LTDA and GIDCO S.A.S., who contributed with their experience in seismic processing and interpretation. Thanks to this collaboration, the algorithm has been integrated into a specialized software, currently in the final testing phase.
“This software will allow industry and academia to access a practical and easy-to-use tool, without the need for advanced programming skills. In addition, it can be continuously updated with new data, improving its performance and adaptability over time,” argue its creators.
Both companies have evaluated the technical feasibility of the development and plan to apply it in real energy exploration projects.
Olga Cecilia Chacón Castaño, processing leader and general manager of PetroSeis Ltda., commented that participating in this type of initiative is an opportunity for science to be reflected in new resources that contribute to their daily business work.
“For a Colombian company like ours, it means a lot to be able to apply algorithms that have never been used in the country before. This type of development not only has an academic value, but also represents a positive contribution by reducing uncertainty in the evaluation of reservoirs. We are just in the implementation of the results of the project and, so far, the progress has been satisfactory. For us, these are significant changes that directly impact the final results,” he added.
With this development, the UIS once again demonstrates its commitment to the generation of applied knowledge and the effective articulation with the productive sector to contribute with the solution to its needs.
