This paper constructs four meteorological indicators covering atmospheric temperature and humidity characteristics, stratiform stability, dynamic factors, and thermal-dynamic synthesis, based on statistical data of multiple physical quantities and the calculation of atmospheric structure elements using radiosonde data. By analyzing the temporal distribution characteristics of these parameters and employing numerical binary Logistic regression methods, a potential and probability forecasting model for severe convective weather applicable to airport areas has been established. This model can predict the potential for severe convective weather 12–48 h in advance and provide the probability of such events 3–12 h ahead. The study results show that the potential forecast model for 12–48 h can predict the likelihood of severe convective weather occurring relatively early, while the probability forecast model for 3–12 h offers more precise predictions of the occurrence probability, significantly enhancing the accuracy and timeliness of forecasts. This research provides important scientific evidence and technical support for improving the warning capabilities of airport meteorological services, contributing to flight safety and enhancing airport operational efficiency.