The data used in this study includes cherry phenological observation data and temperature observation data in the cherry planting area for 2018-2020 and the temperature data of the national automatic meteorological observation station in 2006-2020 in Shijiazhuang. The research period is determined by adopting mathematical statistical methods, and the meteorological indicators of cherry fertility are delineated according to the GDD(Growing Degree Days) model. Subsequently, by referring to the “Crop Frost Damage Grade”, the daily minimum temperature, the daily average temperature and duration are selected as the elements of the low-temperature freezing injury index in this paper. The linear regression model between low-temperature freezing injury index and yield reduction rate is established. Compared to the four probability distribution models, including poisson distribution, information diffusion method, normal distribution and Weber distribution, the results show that the information diffusion method passes the chi-square goodness of fit test, and its RMSE value of the model is the minimum. Therefore, the information diffusion method is the most suitable low-temperature freezing injury index distribution model. The study finally determines the weather index insurance net premium rate under different trigger conditions of low-temperature freezing injury in Shijiazhuang. The maximum rate is 2.045%, and the minimum is 0.173%. At the same time, this study finds that in the case of a small amount of data, the distribution pattern of the information diffusion method model is more consistent with the real probability distribution pattern. And this model provides a theoretical model for the crop weather index insurance products, for which long-series observations have yet to be conducted.