Cornell researchers created a machine learning-based model that can forecast, with greater accuracy than current methods, the impact severe wildfire conditions will have on solar electricity generation.