Arranging 3 years’ worth of airborne precipitation particle images to construct a precipitation particle image dataset in Shan⁃ dong Province. Building a precipitation particle recognition model based on EfficientNet convolutional neural network, named PREN (Precipitation particle Recognition model based on EfficientNet convolutional neural Network).The accuracy rate is 98%, and the multimodel and multi-index evaluation and comparison experiments verify that PREN demonstrates excellent robustness and generalization ability. Taking typical stratiform-cumulus mixed cloud precipitation as two examples (total 3 time periods), PREN is applied to the par⁃ ticle characteristics analysis of generating cells. Combined with airborne Ka-band cloud radar and DMT particle measurement system, an analysis conducted on the shape proportion of precipitation particles inside and outside the generating cells and indifferent intensity generating cells, revealing the precipitation mechanism. The results show that the shapes of precipitation particles in the generating cells are mainly spherical, needle-like, irregular and columnar. Precipitation particles outside the generating cells are mostly spherical and needle-like. The cloud microphysical parameters in the generating cells with different intensities vary. The proportion of graupel and needle particles in the precipitation maturity stage is higher than that in the dissipation stage. The average chord length of precipi⁃ tation particles in the maturity stage is 415 µm. While the average chord length of particles in dissipation stage is 367 µm. The par⁃ ticles on the top of generating cells are mainly spherical and hexagonal, primarily growing through the process of deposition. The ratio of irregular particles and columnar particles in the 0 ℃ are increasing, and the melting process and dynamic conditions favor aggregation and growth, forming irregular particles, while columns mainly originate from the upper levels of the atmosphere.

Based on the record data of water disasters and the 5-min precipitation of 40 meteorological observation stations along the Shuozhou-Huanghua Railway from 2017 to 2019, the characteristics of railway water disasters and precipitation distribution were analyzed, then the three precipitation factors including continuous precipitation, the hourly maximum precipitation and the 24-hour precipitation were counted, the rainfall warning thresholds of no warning, patrol warning, speed limit warning and blockade warning of railway sections in plains and mountainous areas were formulated by using the mean-standard deviation method and the maximum value method. The results show that the water disasters of the Shuozhou-Huanghua Railway mainly occurred in July and August, and the duration of precipitation was mostly within 48 hours. The precipitation types causing water disasters were mainly local rainstorm, short-time heavy precipitation and long-duration precipitation, the railway water disasters in plain sections were mainly caused by local rainstorm, but the main cause of mountainous sections was long-duration precipitation. For railway section in the plain, the accuracy rate of patrol warning was 88.5%, the false rate was 11.5%, the accuracy rate of speed limit warning was 100%, for the railway section in the mountainous, the accuracy of patrols warning was 88.9% and the false rate was 11.1%. The rainfall warning threshold for railway sections in plains and mountainous areas could provide reference for safe running and efficient operation of railway.

Based on traffic accidents and meteorological observation data from December 2015 to November 2018 in Hebei Province, the quantitative relationship between highway traffic accidents and meteorological conditions was analyzed. The highway traffic accidents happened more in summer and autumn than in spring and winter, and they happened more in the daytime than during nighttime. The peak value appeared in August and October, and it appeared during 09:00-11:00 and 14:00-17:00. Highway traffic accidents in Gaocheng of Shijiazhuang and Fengnan of Tangshan occurred most. Take the case of Shijiazhuang area, the response of the relative risk RR of highway traffic accidents to different meteorological elements was analyzed using the Spearman rank correlation and curve fitting method, it was found that there was a significant threshold effect of temperature on the frequency of highway traffic accidents, and the threshold of daily average temperature, the daily maximum temperature and daily minimum temperature  were 20 ℃, 25 ℃ and 15 ℃, respectively. When the daily minimum relative humidity exceeded 80%, the relative risk of accident increased by 3.77% for every 1% increase of relative humidity. When the maximum rain intensity increased 10 mm·h-1, the accident risk increased by 18.8%. When the visibility was less than 1000 meters, the risk of highway traffic accident decreased by 4.14% with the increase of 100 meters in the visibility.

Based on the meteorological observation data at traffic weather stations along the highways of Hebei Province and the highway traffic accidents and closed-control data caused by the meteorological conditions during 2012-2017, the influence factors of the intensity of high impact weather, durations, risk zoning levels, single traffic flow, topography and occurrence period, etc. on the highway traffic passing were selected, firstly. Then the risk level forecast models of fog, road icing and heavy rainfall disasters were established by using the multifactor weighted method. And on this basis, the integrated risk level forecast model of highway traffic based on three weather conditions was built, and the grade standards were defined with the discrimination index of highway closed-control time caused by fog, road icing and heavy rainfall. Through testing, the accuracy rate of the forecast products of highway traffic integrated risk level model based on the meteorological conditions was 76.7%, which can meet the demand of daily traffic meteorological service.