Semiautomatic fault-surface generation and interpretation using topological metrics

Abstract

Seismic fault surfaces are compulsory input for structure modeling that unravels the structural deformation history of the subsurface. Seismic fault attributes provide geoscientists with alternative images of faults. However, seismic fault attributes only highlight possible fault locations and do not directly provide fault surfaces that are required inputs for structural modeling. Interpreters construct seismic fault surfaces using interpreted seismic fault sticks on vertical seismic slices. Interpreting fault sticks on hundreds of seismic slices is time consuming. We have semiautomatically constructed fault surfaces by simulating the procedure of manual seismic fault interpretation. Our algorithm consists of three main steps: (1) obtaining fault sticks in the inline, crossline, and time slices; (2) grouping the fault sticks according to the connectivity and mutual exclusion (topology) between the fault sticks on the inline, crossline, and time slices; and (3) generating the fault surface patches by merging the fault sticks time slice by time slice through the topology analysis. Our algorithm contains one optional step: manually merging the fault patches if needed. We test our algorithm on open access seismic data and our workflow accurately generates fault surfaces for most faults including conjugate faults in the seismic data. Considering that it usually helps to weight the estimation according to the quality of the computed fault attribute, the algorithm computes fault parameters such as fault dip and strike using weighted principal component analysis.