Low-Frequency Construction Method for Igneous Rock Reservoir Based on Lithology-Driven and Convolutional Neural Network:A Case Study of H Buried Hill in Huizhou Sag,Pearl River Mouth Basin
WANG Yaosen
LI Li
XU Chao
WANG Shenghao
Abstract：The construction method and accuracy of low-frequency models for seismic inversion are crucial for the accuracy of inversion,especially for buried hill reservoirs with deep burial and strong heterogeneity.The conventional well interpolation low-frequency model may reduce the prediction accuracy of inversion results due to ignoring the heterogeneity between wells.In order to solve the above problems,this paper takes the H buried hill in Huizhou Sag of the Pearl River Mouth Basin as an example to carry out optimal logging evaluation and petrophysical modeling of buried hill.The seismic interval velocity optimized by well control was used as the initial low frequency for prestack simultaneous inversion,and the lithology probability prediction was carried out based on the inversion results.Using elastic parameter bodies,lithology prediction bodies,and logging curves,a low-frequency model driven by lithology was created based on convolutional neural networks as the low-frequency model for subsequent pre-stack inversion.The research results show that the low-frequency model based on lithology-driven and convolutional neural network can effectively solve the problem of difficult construction of low-frequency models in the prediction of igneous rock reservoirs in the Pearl River Mouth Basin,improve the seismic inversion accuracy of igneous rock buried hill reservoirs,and provide strong support for the development of H buried hill.
Keywords：Low-frequency modelConvolutional neural networkPrestack inversionIgneous rock reservoir
Publication Date：2026-02-28
Online Publishing Date：2026-03-03(First online date of this platform, not the publication date of the document)
Pages：8( 131-138 )