Development of Region-Specific Rainfall Design Storms Using Machine Learning in Southwestern Saudi Arabia
Year: 2026
Type: article
Abstract: The mountainous southwest of Saudi Arabia exhibits complex, highly seasonal precipitation driven by Indian Ocean monsoon inflows and orographic lifting. To characterize storm hyetographs, cluster analysis was applied to 8972 rainfall events recorded at 151 gauges. Two primary clusters emerged: one with early, intense peaks and another with later peak intensities, broadly reflecting windward versus leeward storm behavior. A locally derived hyetograph profile (AI) was constructed from the cluster centroids and benchmarked against standard design-storm distributions (Uniform, SCS Type II, Huff quartiles). Across fit metrics—cumulative RMSE, Kolmogorov–Smirnov distance, and cosine-intensity similarity—the AI distribution provided the best match for ~46% of storms, markedly outperforming canonical profiles (Uniform and SCS Type II each best-fit only ~11–12%). These results indicate that region-specific rainfall distributions more accurately represent precipitation patterns than conventional profiles, and that tailored hyetographs can improve hydrologic modeling and water-resources assessments in this climatically heterogeneous region.
Source: Atmosphere
Author: Raied Alharbi
Institution: King Saud University
Language: English
Subfield: Atmospheric Science
Field: Earth and Planetary Sciences
Domain: Physical Sciences
SDG: Climate action
Open Access status: gold
Estimated APC paid: $2,896
Funder: King Saud University
Award: G602733861
OpenAlex ID: W7158195763
FWCI:
Cites:
Cited by:
Related to: