Deadline 7 June 2019
Organisation: BRGM (Bureau de Recherches Géologiques et Minières)
The geological structures resulting from the mechanisms of the salt tectonics, especially at the level of the passive continental margins, are known to be of good reservoir of energy resources. The main difficulty is to recognize structures associated. The seismic imagery allows to recognize these forms / structures by interpretation of the signal. This step always requires an interpretation by a structural geologist expert. This leads to very subjective and highly variable renderings. Recently, computer-based methods of machine learning (in particular "deep learning" branch for image analysis) make it possible to automatically extract the relevant characteristics over large amounts of data. These methods can be applied to geoscientific problems and in particular to the automatic recognition of complex geological structures from relevant descriptive variables: eg. geophysical images and geological informations (rock type, formation type, structural elements, ...).
To do this, it is proposed to (1) to compile and manage all existing data to identify geological structures and more particularly salt tectonics associated; (2) to define a predictive model based on built-up examples and deduce main variables (here type of geological structures); (3) to develop / apply the algorithms of automatic shape recognition methods and this on a case of application on one or more basins affected by salt tectonics mechanisms.