Multiattribute probabilistic neural network for near-surface field engineering application

Unconfined compressive strength (UCS) is an important rock parameter required in the engineering design of structures built on top or within interior formations. In a site investigation project, UCS typically obtained discretely (through point-to-point measurement) and interpolated. This method less than optimal to resolve meter-scale variations heterogenous such as carbonate formations which property changes occur data spacing. We investigate geotechnical application multiattribute analysis based near-surface reflection seismic probe for their attributes at variability. Two Late Jurassic outcrops located central Saudi Arabia serve testing sites: Hanifa Formation Wadi Birk Jubaila Laban. The study uses core 2D profiles acquired both sites, from we constrain UCS, acoustic velocity, density, gamma-ray values. A positive linear correlation between impedance along indicates that can be utilized laterally extrapolate away location. Seismic colored inversion serves input neural network validated with blind test. Results outcrop sites indicate high degree consistency absolute error approximately 5%. also demonstrate applicability predicted interpret mechanical stratigraphy map lateral heterogeneities. These findings provide expensive alternative limited field-scale project.