TY - JOUR
T1 - Smart Interpretation – automatic geological interpretations based on supervised statistical models
AU - Gulbrandsen, Mats Lundh
AU - Cordua, Knud Skou
AU - Bach, Torben
AU - Hansen, Thomas Mejer
PY - 2017/6/1
Y1 - 2017/6/1
N2 - A method that infers a statistical model, which describes a relation between the knowledge of a geologist (quantified by geological interpretations) and the available information (such as geophysical data, well log data, etc.) that a geologist uses when he/she interprets is proposed and tested. The statistical model is then used to perform automatic geological interpretations wherever the same kinds of information, as used for the initial interpretations, are available. This methodology is named Smart Interpretation (SI). In this study, we look at two different approaches to infer such a model, and we demonstrate the applicability of the model to predict the depth to a low resistivity subsurface layer, based on interpretations from a geological expert, using a 19-layered resistivity model obtained from inversion of airborne electromagnetic (AEM) data. This study shows that SI is capable of making predictions with great accuracy. The method is fast and is able to handle large amounts of data of different origin, which suggest that the method may become a very useful approach to assist in geological modeling based on increasingly large amounts of data of different nature.
AB - A method that infers a statistical model, which describes a relation between the knowledge of a geologist (quantified by geological interpretations) and the available information (such as geophysical data, well log data, etc.) that a geologist uses when he/she interprets is proposed and tested. The statistical model is then used to perform automatic geological interpretations wherever the same kinds of information, as used for the initial interpretations, are available. This methodology is named Smart Interpretation (SI). In this study, we look at two different approaches to infer such a model, and we demonstrate the applicability of the model to predict the depth to a low resistivity subsurface layer, based on interpretations from a geological expert, using a 19-layered resistivity model obtained from inversion of airborne electromagnetic (AEM) data. This study shows that SI is capable of making predictions with great accuracy. The method is fast and is able to handle large amounts of data of different origin, which suggest that the method may become a very useful approach to assist in geological modeling based on increasingly large amounts of data of different nature.
U2 - 10.1007/s10596-017-9621-8
DO - 10.1007/s10596-017-9621-8
M3 - Journal article
SN - 1420-0597
VL - 21
SP - 427
EP - 440
JO - Computational Geosciences
JF - Computational Geosciences
ER -