I. Introduction
Solar generation can display high variability and an accurate solar forecast enables an improved operation of the electric grid. Despite steady improvements of forecasts from Numerical Weather Prediction (NWP) models, forecast skill is still limited by mean and conditional biases. In coastal California, solar production is often hindered by the formation and dissipation of Stratocumulus (Sc) clouds, which are most prominent in late spring and summer months. The majority of installed rooftop systems are located along the coast, posing a significant challenge. Sc clouds form in a shallow planetary boundary layer (PBL) at night and begin to dissipate after sunrise. Typical PBL height is often less than 1 km. Due to computational constraints, state-of-the-art NWP models have limited vertical resolution and parameterize the physical processes governing Sc clouds instead of directly solving them. This often leads to an erroneous representation of Sc clouds and thus an inaccurate solar forecast. While efforts have been made to improve the physical representation of Sc clouds in NWP models, imperfect initial conditions and numerical and physical approximations still affect the accuracy of the forecast.