167 Figure 3 – The Return-Risk schematic workflow (Sadeghi, 2020; Sadeghi and Cohen 2023) Here, the Sequential Gaussian Simulation (SGSIM) was applied to the Li-CLR dataset to generate 1000 geostatistical realizations (i.e., scenarios). To quantify the spatial uncertainty represented by the variability among these realizations, the Euclidean distance (- representing the spatial dissimilarity- see Scheidt and Caers, 2009) was calculated for each pair of models. Because the resulting dissimilarity matrix is high-dimensional, Multidimensional Scaling (MDS) was used to reduce its dimensionality and provide a more interpretable representation of the spatial uncertainty structure. The Return–Risk plot derived from these outputs allows the operator to select the subset of models that achieve the highest possible geochemical return with the lowest associated uncertainty. In this example, two scenarios were tested: selecting the 10 best target points and selecting the 30 best target points. After choosing these preferred clusters of models, they were transformed back into their spatial form and averaged to produce an EType map representing the central tendency of the selected realizations. This final output serves as the target map, highlighting the most promising locations for follow-up sampling, field validation, and drilling. A key strength of the method is that the operator—not the algorithm—decides the number of target areas to prioritize. This flexibility allows the selection to be tailored to practical constraints such as available time, budget, field personnel, and broader exploration strategy, making the workflow both robust and operationally realistic (Figure 4).
RkJQdWJsaXNoZXIy MTM0Mzk2