Analysis of seismic tomography and geological data to identifying spatial relationships between large ore deposits in northern Chile using machine learning methods: Preliminary results

Comte, Diana; Navarro, Felipe; Ojeda, Marcia; Garrido, Felipe; Calle-Gardella, Daniela; Santibañez, Felipe; Egaña, Alvaro; Ehrenfeld, Alejandro; Roecker, Steven; Vargas-Otte, Jimena

Keywords: porphyry, local earthquake tomography, geophysics, seismic tomography, Geophysical exploration


A 3D body wave velocity field in northern Chile was previously obtained using local earthquake tomography (LET) using data recorded by a combined 360 seismic stations deployed at various times over several decades. With the seismic tomography it is possible to know not only the main characteristics of the subduction processes but also, the relationship between the occurrence of large porphyry copper deposits in the north with an associated low Vp/Vs anomaly beneath them, at a regional scale. The relationship between subduction process and mineralization has been recognized in metallogenic studies since the theory of plate tectonics was widely accepted. Important metallogenic belts worldwide are mostly located in subduction zones. Porphyry-type deposit is generally related to arc magmatism or partial melting of subducted plate, with parts of ore-forming fluids ultimately derived from dehydration of the subducted slab, indicating the intimate relationship between subduction process and mineralization. A predictive understanding of the relationships between ore systems, mantle melting, and lithosphere-scale tectonism requires a whole-system approach. In recent years, several types of Machine Learning (ML) methods have been applied by Earth scientists to extract patterns and structures from multi-dimensional feature spaces. In this regard, the aim is to carry out an adequate characterization of the Vp/Vs anomalies present in the study domain, as well as the analysis of the relationship between these anomalies and the mineralized bodies found in these metallogenic strips. For this purpose, it is proposed to use different spatial characterization techniques to segment the areas, which will subsequently allow the use of models based on local spatial information to relate the presence of a given orebody, based on the geological data, satellite multispectral images, and seismic tomography conditions of that site. A different analysis is provided, showing that machine learning techniques are useful to understand the spatial relationships between large porphyry copper deposits within other data sources towards mineral exploration.

Más información

Fecha de publicación: 2021
Año de Inicio/Término: 13-17 December 2021
Idioma: English