Cloud forecasting remains a significant challenge. Short-term forecasts (0-6 hrs) span a time frame where observation-based techniques typically give way to numerical modeling with regard to forecast skill (outside of the tropics). Observation-based techniques utilizing surface observations such as conditional persistence date back more than fifty years. Various research groups in recent years have applied more sophisticated approaches such as those based on knowledge discovery in database (KDD) techniques using surface observations. In this project, we investigate new, probabilistic short-term cloud forecast techniques that utilize current and historical high-resolution satellite cloud imagery to identify analog ensembles. The approaches we explore range from extensions of the traditional conditional probability-based forecast to automated pattern recognition techniques. Our findings are potentially relevant to a range of applications including civil aviation, military reconnaissance, and variational data assimilation.