Shear wave velocity prediction based on long- and short-term memory network: a case study of Paleogene reservoir in Huizhou area, Zhu I Depression, Pearl River Mouth Basin

The shear wave velocity is an important parameter for seismic prestack inversions and accurate prediction of the missing shear wave velocity in the well is a key step of reservoir identification. Traditional methods for predicting shear wave velocity based on petrophysical modeling is often limited by the assumptions of petrophysical models, making it difficult to achieve high prediction accuracy. Therefore, we fully leveraged the advantages of deep learning neural network technology and proposed a shear wave velocity prediction method based on long- and short-term memory networks. This method utilizes long- and short-term memory networks, combined with the unique characteristics of shear wave velocity curves in longitudinal time series, to deeply explore the mapping relationship between logging parameters and shear wave velocity, and successfully built a network model between the two, enabling the model to accurately predict shear wave velocity. This technique was applied to the Paleogene reservoir in Huixinan area, Zhu I Depression, the Pearl River Mouth Basin, South China Sea, and the missing shear wave velocity in the well was predicted. The prediction results show that compared to the estimation of shear wave velocity using traditional petrophysical models, the prediction results based on the long- and short-term memory networks have smaller errors, higher correlation coefficients, and higher prediction accuracy compared to the measured shear wave velocity. This study can lay a solid foundation for reservoir prediction through seismic prestack inversion in the future.