Onyutha, Charles2025-06-232025-06-232025-06-18Onyutha, C. (2025). A multi-hydrological model ensemble prediction uncertainty estimation (e-PRUNE) framework .Hydrology Research.https://doi.org/10.2166/nh.2025.1161998-95632224-7955https://doi.org/10.2166/nh.2025.116https://hdl.handle.net/20.500.12504/2513Among the several hydrological model uncertainty estimation methods, the generalized uncertainty estimation (GLUE) method is popular due to its simplicity. The application of GLUE tends to be limited to the cases when hydrological models are applied individually. Notably, little attention is given to model differences when applying GLUE. This study introduced a framework for multi-hydrological model ensemble prediction uncertainty estimation (e-PRUNE). For demonstration, the framework was applied to real hydrometeorological data while considering three sub-sources of calibration-related uncertainty including the influence of the choice of a sampling scheme, hydrological model (HM), and objective function (OF). Ten SCs, six HMs, and eight OFs were considered. Influences from SCs, OFs, and HMs were combined and assumed to substantially comprise the overall predictive uncertainty (OPU). The sub-uncertainty bound based on HM's choice was larger than that of either SC's or OF's selection. Contributions of sub-uncertainties from HM, SC, and OF to the OPU were additive. Thus, the effect of removing one source of uncertainty (for instance, OF) could easily be realized from the width of OPU's bounds. This study showed the importance of the e-PRUNE framework for insight into the contributions of various sub-uncertainty sources to the OPU.enCalibration-related uncertaintyEnsemble predictive uncertaintyGeneralized likelihood uncertainty estimation (GLUE)Parameter sampling schemesRandomized block quasi-Monte Carlo sampling (RBMC)Single-objective calibrationArticle