The Hong Kong Observatory has developed a lightning nowcasting system for detecting and nowcasting lightning activities over the Hong Kong International Airport (HKIA) with a view to protecting personnel and aircraft on the ground from lightning strikes. The system, named the Airport Thunderstorm and Lightning Alerting System (ATLAS), consists of two components, namely the detection module and the prediction module. In the detection module, data from the Lightning Location Information System (LLIS), which processes data from five lightning sensors, are used for the detection of lightning activities in the neighborhood of HKIA. In the prediction module, all the latest cloud-to-ground (CG) strikes detected by LLIS around HKIA are grouped into clusters in the form of ellipses. The movements of the ellipses are then tracked by the storm motion vectors provided by the Observatory's nowcasting system SWIRLS (Short-range Warning of Intense Rainstorm in Localized System). SWIRLS is a radar-based thunderstorm nowcasting system which identifies and tracks the movements of intense thunderstorms by using the pattern recognition method TREC (Tracking Radar Echoes by Correlation). In ATLAS, the future position of each ellipse is extrapolated based on the TREC vector. If any of these ellipses fall within 1-km or 5-km from the airport boundary, the RED and AMBER alerts will be issued by ATLAS respectively. Weighted ensemble and time-lagged ensemble methods are implemented to improve system performance so that the alert duration is minimized while achieving a high hit rate. This paper presents the system design of ATLAS, its performance and the next steps for further enhancements.