Climate change has exacerbated droughts and flash floods, particularly in arid and semi-arid regions, posing significant threats to cultural heritage worldwide. Despite advances in geospatial and hydrodynamic modeling, the vulnerability of Arabian cultural sites to flash floodsespecially those triggered by dam breaches and extreme storms over unprotected watershedsremains largely unexamined. This study integrates Sentinel-2, ALOS-DEM, IMERG-based rainfall data, and ancillary datasets within a Machine Learning-based Support Vector Machine (SVM) framework, coupled with hydrodynamic simulations, to assess cultural susceptibility in Najran, a culturally significant region in southern Saudi Arabia. The Flash Flood Hazard Susceptibility (FFHS) map achieved an AUC of 0.9, with 76.47 % of flood-affected areas concentrated in highsusceptibility zones. The susceptibility classification includes very low (23.79 %), low (32.83 %), moderate (25.16 %), high (13.66 %), and very high (4.57 %) zones. Hydrograph analysis estimated peak discharges of 40,000 m3/s for the dam breach scenario (case a) and 100,000 m3/s when combined with downstream precipitation (case b). The 100-year flood inundation scenario (case b) projected flood depths of 12 m at Al-Aan Palace, 10 m at Al-Lijam Village, 9.5 m at AlAmer, 8.5 m at Al-Manjam Village, and 7 m at Al-Okhdood, with corresponding flow velocities of 3.5 m/s, 9 m/s, 5 m/s, 3 m/s, and 4 m/s, respectively, while Raoum Castle remained unaffected. Notably, Al-Okhdood, the oldest archaeological site dating to the 7th century BC, faces complete submersion. These findings underscore the urgent need for early warning systems and enhanced collaboration between researchers and cultural authorities to protect heritage in Arabian nations.