Using python libraries and k-Nearest neighbors algorithms to delineate syn-sedimentary faults in sedimentary porous media

This paper introduces a methodology based on Python libraries and machine learning k-Nearest Neighbors (KNN) algorithms to create an interactive 3D HTML model (3D_Vertical_Sections_Faults_LRD.html) that combines 2D grain-size KNN-prediction vertical maps (vertical sections) from which syn-sedimentary faults other features in sedimentary porous media can be delineated. The visualized handled with conventional web browsers. physical parameter is measurable, constant over instrumental time, handleable mathematically, its range associated lithological classes. Grain-size input data comes public database of 433 boreholes the Llobregat River Delta (LRD) NE Spain. Four classes were defined: Pre-Quaternary basement, Quaternary gravel, sand, clay–silt. Using new algorithm, seven NW–SE (transversal) three SW–NE (longitudinal) sections created following orientation identified surface detected reflection seismic geophysical surveys. For exploratory K values 1–75 used. around 25 provides general smoothy shape basement top surface, whereas = 1 optimal value represent heterogeneity LRD at short distance. confidence algorithm inspired Similarity Ratio for machine-learning KNN, overall was evaluated as satisfactory. A decreasing trend according density depth inland seaward found. show horizontal interruptions (displacements or steps) continuity coarse bodies (gravel sand) attributable action active faults. These have been linked correlated well-known area related much cases Valencia Trough opening. Moreover, several this by first time revealed fault zones made branches different steps echelon-like arrangement. Faulting seems more evident Pleistocene Lower Detrital Complex less inactive Holocene Upper Complex. Syn-tectonic gravel channels faulty controlled, progradation lobes, lateral migration channel bars also observed. At current development stage, could applied geological environments, making due minor modifications code, especially suitable reduce high (usually unmeasurable) uncertainty qualitative used complex numerical tools aimed modelling lot resources (groundwater, minerals, geothermal, petroleum) Earth phenomena.