Outlier-sensitive flood frequency modelling of the Jhelum river using multi-distribution and multi-estimator analysis

Maximum Likelihood Estimation (MLE), Method of Moments (MOM), and L-Moments (LMOM). Emphasis was placed on analysing the influence of low outliers on flood quantile estimation, for which the Multiple Grubbs–Beck Test was applied, followed by log-linear interpolation to generate corrected datasets. Parameter estimation and quantile calculations were conducted using RMC BestFit, EasyFit Professional, and R software. Goodness-of-Fit (GOF) tests—Kolmogorov–Smirnov, Anderson–Darling, and Chi-Squared, were applied to identify the best-fit models for each station. Results indicate that while the GEV and LP3 distributions were highly sensitive to outlier treatment, especially under LMOM and MLE the Gamma and Gumbel distributions exhibited comparatively stable behaviour. Notably, LP3-LMOM consistently provided the best fit across stations, but GEV-MLE and Gamma-MOM were also strong performers depending on station-specific characteristics. The findings underline the importance of station-wise distribution selection, proper outlier treatment, and the use of multiple estimation methods to improve the robustness of flood risk predictions. The developed framework can assist hydrologists and planners in designing resilient infrastructure and implementing flood management strategies tailored to the hydrological realities of the Jhelum River Basin.