LAUSR

Abstract This paper introduces a method for background random noise attenuation in seismic reflection data giving priority to the preservation of coherent seismic events and automation of the algorithm. Since the statistical characteristics of random noise are different than those of coherent events, in the proposed method, after calculating Adaptive Wiener Filter (AWF), with different window sizes, the structure of the input data ware calculated by Fuzzy C-Mean Clustering (FCM). The sorted standard deviation of AWF values ware also used to determine the input data for training of Adaptive Neuro-Fuzzy Inference System (ANFIS). Trained network was generalized to all input data points and the output of ANFIS, alongside with data structure ware used to determine the optimized output by comparing noise level of all outputs. The proposed method was applied on both synthetic and real data sets and the results were compared to those of the conventional methods. The research findings revealed that the method was of a considerably higher performance in random noise attenuation as well as preserving the coherent events.