A method for well log data generation based on a spatio-temporal neural network

Abstract Well logging helps geologists find hidden oil, natural gas and other resources. However, well log data are systematically insufficient because they can only be obtained by drilling, which involves costly time-consuming field trials. Additionally, missing or distorted common in old oilfields owing to shutdowns, poor borehole conditions, damaged instruments so on. As a workaround, pseudo-data generated from actual data. In this study, we propose spatio-temporal neural network (STNN) algorithm, is built leveraging the combined strengths of convolutional (CNN) long short-term memory (LSTM). The STNN exploits ability CNN effectively extract features related pseudo-well LSTM key along depth direction. method allows full consideration trend with depth, correlation across different series accumulation effect. proved successful predicting acoustic sonic gamma-ray, density, compensated neutron, formation resistivity diameter logs. Results show that proposed achieves higher prediction accuracy it takes into account information