Improving seismic horizon matching by ordinal measures

Many processing and interpretation tasks of seismic images are based on cross-correlation (CC) as a measure of similarity of reflection patterns. Although, CC has several advantages over other correlation methods, it is very sensitive to nonlinear intensity variations or single outliers, which are however characteristic for seismic data. We show for the example of horizon matching across faults, that more robust ordinal measures can perform better, especially in case of poor data quality.