Dust storms across the Arabian Peninsula cause significant disruption to aviation, logistics, energy operations, and public health. Existing meteorological forecasting systems provide limited spatial resolution and insufficient lead time for operational decision-making. LaythTech's Earth observation team has developed a satellite-fed prediction model that addresses both limitations.
The Observation System
The forecasting capability draws on a 6U cubesat constellation providing hyperspectral imagery at 30m resolution across the visible and shortwave infrared bands. Dust aerosol characterization using the Aerosol Optical Depth retrieval algorithm provides the core input to the predictive model.
Combined with ECMWF wind field data and a physics-informed neural network trained on 18 months of Arabian Peninsula dust event observations, the system achieves a 4-hour prediction lead time with 78% accuracy at the event-onset threshold.
