RAINFALL PREDICTION FOR QINGMING FESTIVAL BASED ON ARIMA-LSTM MODEL

Requent rainfall during the Qingming Festival period has impacted public travel, cultural and tourism activities, as well as urban management. Based on nearly 20 years of meteorological data, this study develops an ARIMA-LSTM hybrid model to model and predict rainfall patterns in five representative cities: Xi’an, Turpan, Wuyuan, Hangzhou, and Wuhan. The results indicate that the model demonstrates strong fitting accuracy and stability, with an average R2 of 0.84196 and a prediction accuracy of 89.9%. After incorporating a real-time correction mechanism, the model’s responsiveness to abrupt weather changes improved, with error control enhanced by over 15%. This study provides data support and methodological reference for short-term meteorological services and public travel during the Qingming Festival.