**Abstract** Random-field reliability analysis (RFRA) is a sophisticated probabilistic method for slope stability assessment that accounts for the spatial variability of soil parameters. Its adoption in geotechnical practice, particularly in regions like East Africa, remains limited due to perceived computational complexity and a lack of validated, accessible workflows. This study aimed to replicate a published RFRA framework for soil-nailed slopes to validate a transparent MATLAB workflow, rigorously assess its reproducibility, and demonstrate its applicability to Ugandan site conditions. The methodology entailed re-implementing the RFRA framework, modelling soil cohesion (c') and friction angle (φ') as cross-correlated lognormal random fields via the Karhunen–Loève expansion. Slope stability was evaluated using the limit equilibrium method within a Monte Carlo simulation scheme, applied to a representative Ugandan case study with local geotechnical parameters. The replication successfully reproduced core theoretical outcomes, confirming a probability of failure (Pf) of approximately 0.12 for the baseline case. A critical finding was the workflow's pronounced sensitivity to the spatial correlation length; a 50% increase in this parameter yielded a 40% rise in the computed Pf, underscoring its importance for site characterisation. The narrow 95% confidence interval for Pf indicated robust numerical convergence. The study confirms the framework's replicability and positions the provided code as a viable, transparent tool for probabilistic assessment, directly applicable to the regional context to demystify random-field modelling.
Okello et al. (Fri,) studied this question.