The precipitation and raindrop size distribution (RSD) characteristics of the four seasons and different rain types were studied using a PARSIVEL² raindrop disdrometer set in the southwest mountain areas of China from 2019 to 2021. The seasonal precipitation in the southwest mountain areas was mainly stratiform rain. The peaks of the RSD were about 1–2 orders of magnitude higher than those in the plains. The convective rain in spring and autumn was very close to the ocean-like convective mass. The local shape–slope (μ–Λ), radar reflectivity–rain rate (Z–R), and kinetic energy–rain rate (KE–R) relationships were further derived, and the diversity of these relationships was mainly due to the variability of the RSDs. In addition, the differences in the RSD characteristics between the top and the foot of the mountain during a typical precipitation process in the summer of 2020 were further compared. It was found that the number density of the small particles at the top of the mountain was higher than that at the foot of the mountain due to the broken large raindrops caused by the high wind speed, while the high evaporation rate, strong convective available potential energy (CPAE), and water vapor content at the foot of the mountain could strengthen the RSD, making the number density of the large raindrops at the foot of the mountain higher than that at the top.

To address the challenges in observing wire icing, this study took the deep residual network ResNet34 as the baseline model and optimized it using normalization, dropout techniques, and data augmentation methods. A novel wire icing risk level identification model based on deep learning was proposed. The results demonstrated that the optimized ResNet34 model (ResNet34+) achieved an average identification accuracy of 93.3% for wire icing risk levels across different regions and wire orientations. Additionally, the identification accuracy was notably higher between 8:00−11:00 and 15:00−17:00. During a coexisting freezing rain and supercooled fog event on Lushan, the model achieved an average wire icing identification accuracy of 89.4% and 90.8% on east-west and north-south oriented wires, respectively, indicating good generalizability of the model. The application of this model provided a novel approach for identifying wire icing risk levels.