Based on road overwater blocking events and cumulative precipitation data in corresponding period in Hainan Province from 2015 to 2018, the general distribution characteristics of road overwater blocking events in Hainan Province and their correlation with precipitation factors were analyzed. The results show that 78.4% precipitation exceeded 100 mm within 24 hours before the events, and 49.0% precipitation exceeded 100 mm within 12 hours before the events. When the hourly rainfall exceeded 50 mm, the cumulative precipitation exceeded 100 mm within 24 hours and the hourly precipitation exceeded 30 mm, the daily precipitation exceeded 50 mm within 3-5 days and the rainfall exceeded 10 mm within 3-6 hours consecutively, the overwater blocking events might be triggered. The rank-correlation method was used for further quantitative analysis, and it showed that the risk of events increased by 1.3% and 0.7% respectively with the increase of 1 mm of cumulative precipitation in 3 and 24 hours. The Logistic risk prediction models of road overwater blocking events were established based on the cumulative precipitation of 3 hours and 24 hours before the occurrence of the event, respectively. When the two models were applied separately, the prediction ability of short-term heavy rainfall and total cumulative precipitation in early stage was insufficient. When the model was applied comprehensively by using the analytic hierarchy process, the prediction accuracy of the road overwater blocking events could reach 91.7%.

Based on the conventional observation, wind profiling radar, the Doppler radar and the NCEP reanalysis data, the causes of the two rain to snow processes in Liaoning Province and the predictability of the snowfall were analyzed comparatively from the perspective of the weather system, water vapor and thermal dynamical evolution. The results show that it took a long time to warm up with a short frontogenesis time, and the precipitation increased due to the near-ground front zone in the process Ⅰ. The front zone between the 925 hPa and 850 hPa level presented a vertical distribution and the snowfall ended after the frontal passage on the 850 hPa level. While there was a transient strong warming in the process Ⅱ, the cold air wedged into the low layer earlier and the warm and humid air slid up along the cold wedge. The duration of frontogenesis was long and there was no precipitation in the period of the near-ground frontal passage. The occurrence of the heavy snow was closely related to the middle level front zone and low level cold recurrent flow. The slow passage speed of the 850 hPa level front zone led to the persistent snowfall which stopped until the frontal passage on the 700 hPa level. Radar echo displayed that the height of the 0 ℃ layer bright band significantly reduced before the rain changing to sleet, and that height basically maintained during the sleet stage, but the 0 ℃ layer bright band disappeared during the snowfall stage. The effect of preceding warming and the daily temperature change owing to the systemic cooling needed to be concerned when correcting the numerical forecast, then the precipitation period should be determined according to the dynamical characteristics of the precipitation system, finally the snow volume was comprehensively corrected.