Multiattribute variable-window waveform classification and application
Mingjun Su, Yuan Cheng · 2018
Conventional waveform clustering may be accomplished using single attribute and constant time window. To meet the demand of continental stratigraphic analysis, we present a method for waveform classification using multiple attributes and variable window. We first use PCA for dimension reduction of a number of attributes and then transform these attributes from time domain into frequency domain, where seismic signals within the zone of interest will be interpolated to be of equal trace length. Waveform classification may be obtained through FCM. As per model test, this new method is superior to conventional methods. The case study in the Qibei sag, the Bohai Bay Basin, China shows some sedimentation types, e.g. deltaic front, limy and dolomitic flat, and sand bar, were successfully discriminated. Presentation Date: Monday, October 15, 2018 Start Time: 1:50:00 PM Location: Poster Station 1 Presentation Type: Poster