Expert-Independent Generalization of Well and Seismic Data Using Machine Learning Methods for Complex Reservoirs Predicting During Early-Stage Geological Exploration

The aim of this study is to develop and apply an autonomous approach for predicting the probability of hydrocarbon reservoirs spreading in the studied area. Autonomy means that after preparing and inputting geological-geophysical information, the influence of an expert on the algorithms is minimized. The study was made based on the 3D seismic survey data and well information on the early exploration stage of the studied field. As a result, a forecast of the probability of spatial distribution of reservoirs was made for two sets of input data: the base set and the set after reverse-calibration, and three-dimensional cubes of calibrated probabilities of belonging of the studied space to the identified classes were obtained. The approach presented in the paper allows for expert-independent generalization of geological and geophysical data, and to use this generalization for hypothesis testing and creating geological models based on a probabilistic representation of the reservoir. The quality of the probabilistic representation depends on the quality and quantity of the input data. Depending on the input data, the approach can be a useful tool for exploration and prospecting of geological objects, identifying potential resources, optimizing and designing field development.