Figure notes: Each bar represents one year of drilling activity, expressed as the number of drill holes (left axis) and drilled meters (right axis). Project stages reflect variations in data-generating rates throughout the mine lifecycle. Data Quality corresponds to an index derived from database integrity parameters, including completeness, uniqueness, interval consistency, depth validation, and standardized geological coding. Survey Reliability reflects drillhole positional accuracy, including collar coordinates and downhole survey quality. The Final Score represents a weighted composite of these parameters, adjusted to specific priorities while maintaining consistent evaluation criteria. Values are reported as mean ± standard deviation by stage. All scores range from 0 (bad) to 100 (good). The Reserve Definition phase had the lowest overall performance, as reflected in the lowest final score (Figure 5). Although data quality remained relatively high during this period, the pronounced decline in survey reliability significantly affected the composite result, underscoring that positional accuracy is more technically relevant than descriptive consistency alone. In resource estimation, a welldocumented sample has limited value if its spatial location is uncertain, because positional errors directly affect spatial continuity, variography, and resource classification, as well as domain boundary accuracy, thereby introducing errors into tonnage estimates. This stage also coincided with a marked increase in both drillhole count and total drilled meters, including deeper holes, which increases operational complexity and amplifies the need for rigorous collar control and downhole survey accuracy. The results demonstrate that data performance is strongly conditioned by operational intensity and project phase rather than by data age alone. In this case study, early campaigns conducted under more controlled conditions showed high spatial reliability, demonstrating that the assumption that older data are inherently inferior is incorrect The correction of key inconsistencies, such as mismatched sampling and assay intervals, depth discrepancies between relational tables, duplicated records, and non-standardized geological coding, substantially improved internal database coherence. As a result, modeling workflows became more robust, better suited to automated modeling and estimation, and less susceptible to reconciliation conflicts prior to estimation, reducing iterative adjustments driven by structural database issues. Database consolidation enabled the recovery and reintegration of legacy datasets that had been previously excluded from modeling due to quality constraints. This process not only enhanced spatial data coverage but also reduced the need for additional drilling in selected areas, thereby maximizing the value of historical investments. The gridded spatial representation of data quality across the mine allowed the identification of sectors with insufficient data support and lower survey reliability. This spatial diagnosis provided an objective basis for prioritizing infill drilling in areas where incremental improvements in data quality could significantly enhance geological continuity and resource confidence. Rather than uniformly expanding drilling programs, campaigns were strategically directed toward zones where additional information would produce the greatest impact on technical robustness and classification support. The application of GDQM produced measurable operational and technical improvements, including: (i) reduced time required for database reconciliation prior to resource modelling; (ii) fewer model revisions associated with data inconsistencies; (iii) enhanced auditability and reporting transparency; (iv) increased robustness of inputs supporting resource classification; (v) more strategically optimized targeting of subsequent drilling campaigns; and (vi) data that is better suited to automated modelling and estimation workflows. Importantly, many of these gains were achieved without acquiring new data, but by qualifying and consolidating existing information. The case study shows that delays in
RkJQdWJsaXNoZXIy MTM0Mzk2