Numerical three-dimensional forecasting of a river section under abnormal discharge conditions due to a tropical storm: A case study on Santa Catarina River, México

• Comparative assessment of four RANS-based turbulence models in three-dimensional river simulations: Spalart-Allmaras, standard k-ε, realizable k-ε, and standard k-ω models. • Validated predictive capabilities of the URANS conjunction and the VoF multiphase model for river flow patterns. • Extensive sensitivity analyses to ensure robust predictions, balancing computational cost and simulation accuracy. • Identification of the realizable k-e model as a superior performance turbulence model. • Establishment of a methodological reference for turbulence model selection in computational riverine studies. • Findings that provide useful insights for flood mitigation strategies, sediment management, and hydraulic infrastructure design in vulnerable regions. Despite advances in CFD for rivers, turbulence models still face accuracy limits under extreme conditions due to sparse data and high computational costs. This study addresses this gap by performing high-fidelity, three-dimensional RANS simulations of a reach of the Santa Catarina River, Mexico, during Tropical Storm Alberto. Using unsteady Reynolds-Averaged Navier-Stokes equations combined with the Volume of Fluid method, the study evaluates fluvial dynamics under high-density flows induced by intense precipitation and sediment transport. The novelty lies in the systematic comparison of four turbulence models including Spalart-Allmaras, standard k – ε , realizable k – ε , and standard k – ω , to simulate fluvial dynamics accurately. A comprehensive sensitivity analysis and mesh independence verification ensured the robustness of numerical predictions, validated using video-derived velocity data. Findings reveal that the realizable k–ε model adequately captures effective viscosity behavior at low to moderate turbulence intensities (0.25≤T i ≤0.75), achieving peak values near 820 kg/m·s, demonstrating its suitability for flows with strong energy gradients. At higher intensities (T i >1.0), this model maintains greater adaptability, while the standard k–ω model, highly sensitive to dissipation, accurately resolves turbulent dynamics but exhibits a sharp decline in µ eff for T i >3.0, with values ranging between 100 and 500 kg/m·s. The simulations also revealed velocity variations consistent with video-derived data, with an observed maximum velocity of 8.66 m/s and an overall mean velocity of 4.43 m/s, validating the numerical framework for high-density river flows. Although not fully optimized for all rivers, these models effectively simulate complex fluvial hydrodynamics.