Abstract:The EMD method is the central part for Hilbert-Huang transform.With the method the seismic data can be decomposited into multi-order IMF components,different IMF component has different frequency characteristics which reflects different geological information.The seismic facies analysis which is based on the EMD method and the Kohonen neural network method can be used in fault identification and reservoir prediction.In this paper the EMD method was used in the field seismic data processing in the western of China,waveform seismic facies analysis was conducted based on the IMF components which were obtained by signal reconstruction and decomposition.The model trial calculation and application results of the field data show that using of the interested IMF component could reconstruct signal,and the reconstructed signals can better display fault distribution characteristics and favorite reservoir range,as well as raise both the signal to noise ratio and the resolution of the seismic data,they have important reference values for understanding the fault distribution characteristics and the reservoir prediction.