52 list of parameters with quantified influence on the predicted pressure response, which will then guide both the falsification test against observed piezometric data and recommendations for prioritising further site characterisation. Figure 1 - Two side-by-side cross-sections of the Mingomba model showing simulated pore pressure distribution at the start of the dewatering scenario (January 2025) and after five years (January 2030), from one representative Monte Carlo realisation of the H1 prior. The figure illustrates the spatial pattern and magnitude of simulated depressurisation under the planned dewatering scenario for a single set of parameter values. 5. DISCUSSION AND IMPLICATIONS Although the analytical stages of this study are still in progress, the work to date already supports several observations relevant to dewatering design at Mingomba and, more broadly, to pre-operational uncertainty quantification at fractured-rock copper deposits. First, the spread of predicted pressure responses across the 100-realisation ensemble confirms that the uncertainty in pre-operational hydrogeological parameters at Mingomba translates into substantial uncertainty in predicted dewatering response. Decisions about the spatial layout, timing, and capacity of dewatering infrastructure that are based on a single deterministic model would be poorly informed about this true range of possibilities. This observation is consistent with prior modelling of the area, in which sensitivity to neighbouringmine pumping alone has been shown to vary predicted dewatering requirements by 30 to 40 percent (ITASCA, 2024), and reinforces the value of an explicit probabilistic approach. Second, the structure of the workflow itself, testing two competing hypotheses (H1: anisotropic continuum; H2: dual porosity) of increasing complexity, with each hypothesis subject to falsification before further inference, is well suited to the Mingomba context. Preoperational dewatering data does not yet exist at the site, so traditional history matching is not possible. The Popper-Bayes protocol provides a defensible alternative: it directs effort toward identifying the minimum hydrogeological model complexity that the data at the site can support, and it makes the limits of inference explicit at each stage rather than concealing them within a calibrated single model. Third, the upcoming DGSA stage will provide direct operational value once completed by ranking the uncertain parameters according to their influence on predicted pressure response. This ranking will allow limited site characterisation budgets to be directed toward the parameters that most strongly govern dewatering response, and will identify which uncertainties can be tolerated without materially compromising decision quality. The specific recommendations for site characterisation will be derived from the sensitivity ranking once the
RkJQdWJsaXNoZXIy MTM0Mzk2