Open Collections

Abstract

Effective surface water management and risk assessment in mining operations require precise meteorological evaluations. This is typically achieved through the application of stochastic precipitation models to runoff-response models, such as GoldSim® water balances or watershed models employing precipitation-runoff models. Importantly, surface water modelling applications in assessments of risk and resiliency require confidence in predicted behaviours at the extremes of statistical distributions. Standard statistical tools are observed to perform well around the centre (mean/median) of the distribution but deviate from observed behaviours at the extremes. This study evaluated the performance of three modelling approaches: First-Order Markov Chain (MC), commonly implemented as WGEN in GoldSim®; a newly modified Markov Chain model; and the CoSMoS framework; in capturing critical hydrological aspects that affect mine operations and planning. This research challenges the status quo of stochastic generators by scrutinizing their actual performance in simulating realistic precipitation scenarios. Each model’s capability was assessed across diverse temporal scales: daily to 30-day intervals for extreme events and annual totals. The findings reveal that while the MC model is widely used, it often underestimates variability in annual precipitation totals, making it less reliable for long-term planning. The Modified Markov Chain model and CoSMoS framework demonstrated improved accuracy in capturing extreme precipitation events, offering more robust solutions for mine water management. The study underscores the importance of selecting appropriate models based on specific site conditions and desired outcomes. Practitioners are encouraged to evaluate model performance against observed data (beyond the 1-day time scale) to ensure reliability.