Identification of gas hydrate based on machine learning in the northern South China Sea

Gas hydrate is an important future alternative marine energy resource to fossil fuels, with the advantages of high energy, large reserves, wide distribution, and shallow burial. Accurate identification of gas hydrate reservoirs and estimation of hydrate saturation are the prerequisites for the development and utilization of gas hydrate resources. This research focuses on the difficult issues of hydrate identification, combined with the multidisciplinary technology of ocean-geology-artificial intelligence (AI). The effective hydrate formation identification technology method is studied and put forward based on the geophysical attributes. The method has been verified in the Dongsha area of the northern South China Sea. This study uses machine learning algorithms to analyze whether the sediment contains gas hydrates. Several commonly used machine learning algorithms are selected, such as random forest, Bagging, AdaBoost, and K-Nearest Neighbor (KNN). These algorithms are used to analyze the data of the P-wave velocity and density with high sensitivity to the change of hydrate. The parameters of different algorithm models are optimized through training, and the identification and classification effects of different algorithm models are compared. Finally, the results show that these algorithms could well distinguish whether there is hydrate in the sediment, among those, the KNN algorithm has a good application. The results show method based on machine learning can improve the identification accuracy of gas hydrate. The identification method of this research provides strong technical support for the subsequent exploration and development of hydrates.