A jointly optimized, multi-criteria peaks-over-threshold framework for regime-conditioned extreme wave analysis in the Persian Gulf

Reliable estimation of extreme significant wave height ( H s ) requires threshold and declustering choices that are rarely justified jointly, and Persian Gulf design-wave studies have almost universally been restricted to omnidirectional records despite well-documented directional heterogeneity in the region's wave forcing. This study addresses both gaps through a jointly optimized, multi-criteria peaks-over-threshold framework with simultaneous threshold and declustering interval selection. Candidates are first evaluated against admissibility criteria covering sample size, shape-parameter plausibility, goodness of fit, and storm independence, then ranked using a five-component objective combining predictive cross-validation, goodness of fit, parameter stability, upper-tail accuracy, and sample adequacy. Robustness was confirmed against four alternative weighting scenarios. The framework was applied to hindcast H s records at five Persian Gulf locations, analyzed both omnidirectionally and conditioned on the northwesterly Shamal regime and a secondary directional regime. Selected H s thresholds ranged from 2.49 to 3.30 m, with omnidirectional 100-year H s return levels of 5.27–5.89 m and negative shape parameters of the relevant extreme value distribution throughout. The Shamal-conditioned return level closely matched the omnidirectional 100-year estimate at four of five locations, while a secondary regime provided the closest directional estimate at the north-westernmost location. These results are consistent with the dominant role of Shamal forcing in regional extreme-wave conditions and provide a directionally informed basis for design wave-height estimation in the Persian Gulf.