The growing use of automated methods in seismic interpretation highlights the importance of data treatment. We analyze the effect of structure-oriented filtering seismic data on machine learning techniques and apply the filter to the data before calculating seismic attributes. We apply this methodology to a migrated section of Buzios Field from the Brazilian presalt in Santos Basin. The analysis is restricted to the case of unsupervised methods, namely self-organized maps and generative topographic mapping. We use four different seismic attributes that are known to be good salt indicators. Two are based on texture, and two are based on coherence. The use of filtering can improve salt identification in clustering when more attributes are considered.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.