Abstract: Based on the early warning signals of meteorological disasters and the data released by “ Flash Message” in Xindu District of Chengdu City from 2015 to 2024, the spatial and temporal distribution characteristics of early warning signals are analyzed by means of classification and statistics. The results show that 9 types of meteorological disaster early warning signals were issued in Xindu District, and the top five were high temperature, heavy rain, strong wind, lightning and fog, among which high temperature warning signals were released the most, accounting for 40% of the total. the early warning signals of meteorological disasters were less in the northwest and more in the southeast. The peak period of release occurred annually from June to August, which was consistent with the actual main flood season. Combined with the application and pilot situation of “Flash Message” mandatory reminder technology, the working mechanism of accurate release of meteorological early warning at the grass-roots level was effectively explored.