Probabilistic assessment of windblown sand hazard across a desert photovoltaic site using terrain-downscaled reanalysis winds

Windblown sand hazard at desert photovoltaic (PV) sites is commonly assessed using coarse regional winds or local array scale experiments, leaving dune scale spatial variability poorly resolved. An hourly probabilistic framework was developed to resolve spatial variations in sand-drift potential. Dune morphology was derived from a digital elevation model, and hourly ERA5 winds for 2021–2023 were downscaled to 30 m using a conservation of mass solver. The downscaling method was independently validated and showed good agreement with field observations ( r = 0.949). The downscaled winds were converted into hourly potential sand transport and aggregated into flux potential (FP) and other transport indicators. Uncertainty in threshold friction velocity and representative annual winds was propagated to PV subarray estimates through Monte Carlo simulation. Terrain downscaling produced acceleration over crests and upper slopes and deceleration in interdune and sheltered terrain. Dunes in more exposed surroundings exhibited higher overall FP, while local hotspots clustered near crests and upper windward slopes, where FP reached approximately three times the dune mean. High transport crest zones generally showed greater directional variability than lower leeward areas. Monte Carlo analysis broadened annual FP distributions without obscuring terrain-controlled spatial contrasts. The five activity classes remained clearly separated, with median FP decreasing from 71.88 m 3 m − 1 yr − 1 in Class I to 60.40 m 3 m − 1 yr − 1 in Class V, while the 5th percentile of Class I exceeded the 95th percentile of Class V. The framework supports targeted sand control planning in desert PV development.