Mapeo de prospección de metales base Cu, Mo, Pb, Zn y preciosos Ag mediante análisis de múltiples variables Co-kriging y Regresión Geográficamente Ponderada - GWR

Abstract

Abstract: This thesis addresses the prospectivity mapping of base metals (copper, molybdenum, lead, zinc) and precious metals (silver) through the integration of multivariate estimation techniques and locally varying spatial relationship modeling. The research is based on the application of Cokriging as a multivariate geostatistical method that exploits cross-correlations among elements to improve spatial estimation accuracy, complemented by Geographically Weighted Regression (GWR) to model the spatial non-stationarity of geochemical relationships. A regional geochemical exploration database was processed using routines developed in Google Colab (Python), employing PyKrige, mgwr, GeoPandas, and SciPy libraries. Results demonstrate that Cokriging reduced the estimation variance of silver and copper by incorporating secondary information from correlated metals, while GWR revealed anomalous correlation hotspots where Cu-Mo-Pb-Zn-Ag relationships exhibit significant spatial variations. Cross-validation confirmed the predictive superiority of Cokriging over ordinary univariate Kriging, and the analysis of GWR coefficients allowed the delineation of multifactorial anomaly zones with potential for polymetallic mineral exploration. It is concluded that the integration of both methodologies constitutes a robust framework for prospectivity potential assessment in multielemental data contexts.