Low temperature freezing damage is the main disaster of late maturing citrus in Nanchong. Based on frozen injury experiments and historical disaster data from the agricultural department, combined with the hourly temperature situation of the National Meteorological Monitoring Station, the threshold of low-temperature freezing injury indicators was studied using probability regression and information diffusion theory methods during the winter period under the conditions of no cloud or less cloud sky, the daily temperature range ≥8℃, and the minimum temperature ≤3℃. The results showed that the linear regression relationship between the meteorological temperature established for the two experimental orchards and the measured temperature was statistically tested. When the meteorological temperature was 1.7℃ in Xichong and 2.6℃ in Jialing, the measured temperature of the orchards would reach -2℃, and frost damage began to occur; Based on the current refined weather prediction capability, a composite index threshold system based on the maximum duration (hours), total duration (hours), and total days of low-temperature freezing damage days during a sliding two week period was established. Four frozen damage samples were selected for testing, and the results showed that the frost damage index rating was in good agreement with the measured frost damage level, which had good practical application value.