Anticlines Prediction Using Deep Learning

Summary The task of predicting anticlinal uplifts, structural traps with the help of convolutional neural networks and on the basis of morphometric characteristics of landforms is considered in the work. The geological component of this task is an urgent issue in the oil and gas sector - anticlinal uplifts are actually often structural traps for oil and gas. In addition, a very interesting geoinformation component - the use of Deep Learning with U-Net architecture for the location of deep structures, using as input information about the current terrain. Traditionally, anticlinal structures are identified on the basis of an expensive set of field geophysics work, the leading of which (and the most expensive) is 3D seismic survey. The development of alternative methods that allow to solve the same problems with comparable accuracy of the forecast, but are at the same time incomparably less expensive, is an urgent task today.