Connectivity-equipped UAVs offer communication opportunities within CAV ecosystems. This study was motivated by challenges in UAV-based monitoring: (a) at intersections where road user trajectories could be irregular, leading to traffic conflicts or collisions, and (b) during inclement weather where video images are often corrupted by falling rain streaks thus impairing the integrity of the image. In response to these challenges, the study developed and demonstrated the efficacy of (a) an algorithm for UAVs to predict vehicle trajectories at intersections and (b) a 2-stage self-supervised algorithm for enhancing the quality of UAV-sourced images. Both parts of the study highlight the UAV potential and challenges for advanced traffic management in the prospective era of CAV operations. There are some potential practical benefits of this research. First, an enhanced UAV-CAV data domain can help road agencies to improve the reliability of their traffic safety risk assessments and vehicle trajectory tracking. Further, from perspectives of systems control, the study products could enhance CAV operations by providing reliable information for safe and efficient trajectory planning and control at urban intersections.