Detecting UAV invasion has been identified an important problem from both academy and industry in recent years. However, the countermeasures of detected invading UAVs, if not carefully designed, may also cause controversy. Existing research and practices mostly consider two-stage systems: UAV detection and interdiction/neutralization after detection. While such ``immediate neutralization after detection'' strategy sounds reasonable in mission-critical scenarios (e.g., battlefields), its implementation in civil applications could be complicated. On one hand, personal interdiction of flying UAVs may result in legal consequence as evident from several recent real-world cases. On the other hand, it could be too late to react when a UAV damage is happening. Therefore, a UAV threat shall be predicted in advance and an early alarm system shall be in place before appropriate interdiction/neutralization (e.g., by resorting to law enforcement) is deemed necessary. In this paper, we propose a multi-layered threat analysis approach to UAVs that have been detected in a dynamic environment.