Research on spectral inversion method based on Moore-Penrose algorithm

Jun-hua ZHANG,Xiao-hui ZHANG,Qiu ZHANG,Zhi-qiang LI,Yang RAN

DOI: https://doi.org/10.6038/pg20170641

2017-01-01

Abstract:Thin reservoir prediction has been a challenge for geophysicists.Because of the limitation of seismic frequency band,the resolution of existing seismic data is limited,it is difficult to identify thin reservoirs that thickness less than 5 m.The spectral inversion technique utilizes the even and odd components of the reflection coefficient sequence.Compared with conventional inversion theory,it is possible to better identify thin layers.Predecessors have made much substantial research,and achieved a lot of valuable conclusions.Most of them focus on the spectral inversion under sparse condition,but it is difficult to know where there is a reflection coefficient,where there is no reflection coefficient in practical application.In this paper,we apply the Moore-Penrose algorithm to the field of spectral inversion,in order to enhance the resolution of spectral inversion under non-sparse condition.First,we use the model to verify the effectiveness of the proposed method,and then add noise to the model to verify the stability of the method.Finally,it is applied to the actual seismic data,which proves its advantages in expanding the seismic spectrum and improving the seismic resolution,and provides a good data base for the prediction and research of the subsequent thin reservoir.