Statistical innovations and artificial neural networks applied to geochemical prospection in Chaparra and Atico quadrangles, Caraveli - Arequipa

• Autor: Bazan Santa Cruz, Orlando
• Formato: artículo
• Fecha de Publicación: 2021
• Institución: Universidad Nacional Mayor de San Marcos
• Repositorio: Revistas - Universidad Nacional Mayor de San Marcos
• Lenguaje: español
• OAI Identifier: oai:ojs.csi.unmsm:article/19212
• Enlace del recurso: https://revistasinvestigacion.unmsm.edu.pe/index.php/iigeo/article/view/19212
• Nivel de acceso: acceso abierto
• Materia: deep learning; geochemistry; prospecting; neural networks; fuzzy overlay; aprendizaje profundo; geoquímica; prospección; redes neuronales; superposición difusa

Geochemical prospecting has always been and will be one of the most effective tools for the mineral resources exploration, that integrates with the base geological mapping and other disciplines like remote sensing and geophysics. This investigation aim is to promote a new geochemical prospecting methodology, which combines the local relative enrichment index (LREI) from mineralization indicator elements (Cu, Au, Ag, Pb, Zn and Mo), range correlation coefficients (RCC) and Principal Components Analysis (PCA), through fuzzy overlays and unsupervised deep learning, to generate geochemical prospectivity maps in the Chaparra (32o) and Atico (33o) quadrangles in the province of Caraveli, Arequipa. The prospectivity map generated by fuzzy modelling and the self-organizing map (SOM) were validated plotting the location of the known deposits and mineralized occurrences, which in most of the cases are plotted within areas with high or moderate prospectivity. Therefore, it is concluded that univariate and multivariate statistical innovations and technological innovations applied to geochemical mapping are effective tools for geochemical prospection.