摘要：We propose a methodology for the recovery of lithologiesfrom geological and geophysical modelling results and apply it to fielddata. Our technique relies on classification using self-organizing maps(SOMs) paired with geoscientific consistency checks and uncertainty analysis.In the procedure we develop, the SOM is trained using prior geologicalinformation in the form of geological uncertainty, the expected spatialdistribution of petrophysical properties and constrained geophysicalinversion results. We ensure local geological plausibility in thelithological model recovered from classification by enforcing basictopological rules through a process called “post-regularization”. Thisprevents the three-dimensional recovered lithological model from violatingelementary geological principles while maintaining geophysical consistency.Interpretation of the resulting lithologies is complemented by theestimation of the uncertainty associated with the different nodes of thetrained SOM. The application case we investigate uses data and models fromthe Yerrida Basin (Western Australia). Our results generally corroborateprevious models of the region but they also suggest that the structuralsetting in some areas needs to be updated. In particular, our results suggestthe thinning of one of the greenstone belts in the area may be related to adeep structure not sampled by surface geological measurements and which wasabsent in previous geological models.
