Abstract:

As a data synthesis method, Weights of Evidence (WofE) model has been widely used in mineral resource quantitative assessment. On the basis of fuzzy weights of evidence, a new weight of evidence method is developed, where layer construction and data integration are based on vector data and vector method. The complete process of modeling is then given combined with a case study, in which the buffer of mineral deposits is taken as training layer, the unique geological units are obtained as evaluation objects by spatial overlaying of all evidential variable layers, and weights of each evidence are then calculated, at last probability map and delineated target can be gotten based on geological objects by weights integration of all evidential layers. Different from raster-based WofE, vector-based WofE model keeps the natural boundary of a geological feature used as geological units which have explicit geological meanings, and can improve the accuracy of posterior probability; Instead of the rectangular regions resulting from regular division, circular regions are used to represent known ore occurrences, which improves the known ore occurrences representatio. A case study shows that average area error are 0.26% and 6% when prediction unit size is equal and not equal with an integer multiple of initial grid size respectively; When prediction unit size is of an integer multiple of initial grid size, average calculation error for deposits is 4.78%. It can be concluded that Prediction unit partition method based on the weight of evidence of geologic units thought better accuracy than the method based on raster or regular grid. Thus the obtained posterior probability distribution based on vector prediction unit division method is more reliable than grid-based method.

Key words: weights of evidence, unit division, vector data, raster data, data integration, resource valuation