Quantitative identification of lithofacies by integrating different geophysical data: False River Point bar Available to Purchase

We automatically extract lithofacies from a modern Mississippi point bar, in False River Louisiana U.S.A. We use K-means clustering with 5 types of data and identify 6 reasonable lithofacies. We integrate different geophysical data such as Electrical Conductivity (EC), permeability calculated from a Hydraulic Profiling Tool (HPT) log, as well as visual core descriptions, and grain size measurements. Multivariate statistical techniques such as the Generalized Additive Model (GAM) allow us to interpolate and increase the limited grain size dataset into sections of the wells where there are no such measurements. GAM is an advanced regression fitting technique which produces robust results and a high R-square ratio. We also enhanced grain size interpolation by implementing a hybrid model which also considers lithology in each well interval.