A machine learning model, originally trained in Oklahoma, was tested at 15 global sites and showed strong generalizability in predicting rapid changes in surface solar irradiance caused by moving clouds. The model can forecast solar "ramp" events crucial for grid stability, with modifications made for different locations to enhance accuracy. Overall, the model performed well across sites, with half matching or exceeding the original model's predictive performance. A study published in Solar Energy found that most sites have similar or better predictability of solar variability, although extreme environments like mountains, deserts, tropics, and high latitudes showed lower predictability. Collaboration between the University of Colorado Boulder, NOAA Global Monitoring Laboratory, and NOAA Global Systems Laboratory contributed to this research.