Automatic Extraction of Layers from Seismic Data in gOcad

Pierre Jacquemin. ( 2007 )
in: 27th gOcad Meeting, ASGA

Abstract

The goal is to automatically partition seismic data into geologic layers by determining the type of material encountered on each node of a seismic data cube. For that, we use horizon extraction to determine a lot of interfaces between layers. The seeds of each horizon may come from wells markers or may be obtained by manual interpretation. Each known node is affected with a value depending on its corresponding horizon. After that, we build for each couple (ix, iy) a set of marks (horizon value, elevation) slipped on a skewer. The obtained skewers are loaded into a 2D table. We also solve the problem of missing values by extrapolating or interpolating their location using several methods based on the neighborhood: • By using the locations of the neighbor • By using the thickness of the layer in the neighborhood. • By using the distance to other known horizons. We also fix the case where the logical succession of horizons is perturbed near faults. The proposed method achieves the following features: • Determination of the layer type everywhere in the seismic cube • An automatic procedure to clean and complete seismic horizons. • A compact way to store the 3D geometry of the sub-surface..

Download / Links

    BibTeX Reference

    @inproceedings{JacqueminRM2007,
     abstract = { The goal is to automatically partition seismic data into geologic layers by determining the type of material encountered on each node of a seismic data cube. For that, we use horizon extraction to determine a lot of interfaces between layers. The seeds of each horizon may come from wells markers or may be obtained by manual interpretation. Each known node is affected with a value depending on its corresponding horizon. After that, we build for each couple (ix, iy) a set of marks (horizon value, elevation) slipped on a skewer. The obtained skewers are loaded into a 2D table. We also solve the problem of missing values by extrapolating or interpolating their location using several methods based on the neighborhood: • By using the locations of the neighbor • By using the thickness of the layer in the neighborhood. • By using the distance to other known horizons. We also fix the case where the logical succession of horizons is perturbed near faults. The proposed method achieves the following features: • Determination of the layer type everywhere in the seismic cube • An automatic procedure to clean and complete seismic horizons. • A compact way to store the 3D geometry of the sub-surface.. },
     author = { Jacquemin, Pierre },
     booktitle = { 27th gOcad Meeting },
     month = { "march" },
     publisher = { ASGA },
     title = { Automatic Extraction of Layers from Seismic Data in gOcad },
     year = { 2007 }
    }