New paper: AI-based solar radiation retrieval outperforms Heliosat

A new deep-learning satellite retrieval of surface solar radiation developed by Kevin marks a shift from traditional physics-based methods like Heliosat. It not only emulates Heliosat but surpasses it by fine-tuning on ground stations, achieving higher accuracy, especially in cloudy conditions and also in mountainous regions. We demonstrate generalisation across Europe and North Africa and quantify the value of individual satellite channels for solar irradiance estimation. Accurate solar irradiance estimates are key to solar energy planning, forecasting and grid integration. Read more here.