Short-term forecasting of low clouds and other hazardous weather conditions is important to U.S. Navy operations. Exploring data fusion of environmental satellite data and numerical model data utilizing machine learning techniques to improve these forecasts is a goal of this project. This research will deliver a method for improved environmental characterization using large, diverse datasets and will also include the application of existing forecast skill metrics to novel prediction systems and designing novel metrics to assist forecast system development.