Raining Bell: Intelligente Regenwarnungen mit Firefly KI-Assistent

“Raining Bell” integrated with “Firefly” Smart Weather Service App Raining Bell is an industry–academia collaborative project that applies real-time radar-monitoring technology to meteorological disaster- prevention services. It integrates nationwide radar observations with FORMOSAT-7 satellite data and employs data-assimilation and radar-echo extrapolation techniques to establish an early-warning system capable of predicting rainfall trends up to three hours in advance. The user interface, designed by the National Science and Technology Center for Disaster Reduction (NCDR), combines location-based services and push notifications to deliver timely, convenient weather information. The latest version incorporates the Firefly smart voice assistant powered by the TAIDE large language model (LLM). Users can interact naturally in Mandarin or Taiwanese to check weather conditions and receive proactive alerts about rainfall and potential disaster risks. By combining real-time forecasts from the Central Weather Administration (CWA) with Raining Bell’s observation data, the system provides intelligent, localized, and personalized weather services. This app was developed under the guidance of the National Science and Technology Council (NSTC) through the joint efforts of NCDR and the National Center for High-Performance Computing (NCHC). The software was produced by ThinkYes Technology Co., Ltd. By integrating AI with disaster-prevention technology, Raining Bell offers a weather service that is both real-time and human-centered. This product represents experimental research results; for official weather forecasts, please refer to announcements from the Central Weather Administration.