Gas chimneys are detected using a supervised neural network trained on reliable examples of gas chimneys. The resultant chimney probability meta-attribute (i.e. the Chimney Cube) can be displayed on key seismic lines. Not all chimneys, detected by neural network training, are related to hydrocarbon migration. The resultant chimneys must be validated based on a set of criteria, such as: pock-mark morphology on time-slices and the spatial relationship with other recognized features of the petroleum system. Valid chimneys can then be output as 3D geo-bodies, and superimposed on 3D reservoir geo-bodies based on seismic facies attributes.

This facilitates the study of spatial relationships between chimneys, faults, traps and other features of the petroleum systems, such as amplitude anomalies, pockmarks, and (paleo-) mud volcanoes. The combined information is compared to analogs that have been classified into a number of categories with different risk profiles for seal and charge.


Chimney Cubes visualize vertical noise trails in seismic records, which facilitates the interpretation of fluid migration paths from seismic data. There are two main application domains:

  1. Geo-hazard interpretation
  2. Deep hydrocarbon exploration & production.

Geo-hazard interpretation focuses on the shallow subsurface. Seismic chimneys are co-visualized with amplitude anomalies, faults and geo-morphological features such as pock-marks and mud-volcanoes. The spatial connection between these features increases our understanding of fluids and fluid migration in the shallow sub-surface. It helps in the identification of potential hazards such as shallow gas pockets that need to be avoided when drilling.

In deep hydrocarbon plays Chimney Cubes are useful instruments to increase our understanding of the petroleum system from source rock to trap and beyond. The main applications in this domain are:

  • Prospect risk assessment for charge and seal
  • Fault seal analysis
  • Basin Analysis
  • Predicting overpressures.