Answers are grounded in this paper and cite their sources.
Streamflow assessment is crucial for effective water resources planning, development, operation, and management. This study aims to develop a hybrid fuzzy optimization framework for the modified Generation Rural 2 M (GR2M) model, incorporating a snowmelt subroutine and representing model parameters as symmetric triangular fuzzy numbers. Fuzzy uncertainty is propagated through the deterministic model equations by applying the extension principle, while parameter central values and semi-widths are calibrated using simulated annealing. For each calibration time step, two auxiliary optimization problems are solved to determine the lower and upper bounds of the fuzzy outflow in addition to the central simulated value. The calibration employs a multi-objective fitness function that simultaneously minimizes the deviation between central simulated and observed flows, penalizes observations falling outside the fuzzy band, and controls the band width. Unlike conventional fuzzy regression, inclusion of observations is therefore treated as a soft rather than an absolute constraint. The methodology was applied to the mountainous Pyli subbasin in central Greece using monthly hydrometeorological data and four levels of the inclusivity factor. Results show that the fuzzy formulation produces parameter ranges and uncertainty bands encompassing most observed outflows, while maintaining central and defuzzified streamflow performances close to those of the deterministic GR2M model. Increasing inclusivity requirement produces wider fuzzy bands whereas lower inclusion weights generate narrower bands approaching the deterministic solution. The proposed framework thus integrates model calibration and fuzzy uncertainty propagation within a single optimization procedure, providing both fitted flows and explicitly quantified ranges of plausible model responses.
No takes yet. Share an insight, caveat, or question.
Stavris et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: