Local Severe Storm Tracking And Warning In Pre-convection Stage From The New Generation Geostationary Weather Satellite Measurements

China is located in the eastern part of Asia.Due to the monsoon and the warm and humid air currents of the Pacific Ocean and the Indian Ocean,convective weather often occurs in the summer,causing a large number of casualties and property losses.Timely and accurate identification and forecasting of severe convective weather is an important part of meteorological disaster prevention and reduction work.With the launch of the new generation of geostationary meteorological satellites such as Himawari-8/-9(H08/H09),higher spatial and temporal resolution data provide new information for the identification and tracking of severe convective weather.The empirical prediction model established by the machine learning method enriches the short-term forecasting method of severe convective weather.Both of these have greatly improved the research capabilities on severe convective weather warning and nowcasting.This paper studies the application of observations from the new generation of geostationary meteorological satellite for strong convective identification and short-term prediction,and analyzes the convective systems in East Asia from April to October 2016.Results indicate that the H08 and NWP based statistical model using the Random Forest(R F)algorithm is capable of capturing local burst convective storm systems about 1 ~ 2 hours earlier than the outbreak of heavy rainfall.The final optimal RF model is achieved with an accuracy of 0.79 for classification of all convective storm systems,whi le the Probability of Detection(POD)of this model for severe and medium convections can reach 0.66 and 0.70,respectively.It is found that the classification result of the model combining the satellite observation factor and the numerical model factor i s better than the model established by only the satellite observation factor.The contribution of satellite observation factors to model classification is generally greater than the contribution of numerical model factors.The three water vapor channels of 6.2?m,6.9?m and 7.3?m contributed the most to the model classification.The convective systems in different regions were modeled separately.The study found that the predictive factors contributed by the models in different regions were not exactly the same.The contribution of the 10.4 ?m channel in Guangdong and Guangxi has increased compared with the origin model.The contribution of the K-index and the CIN in the Sichuan Basin has increased compared with the origin model.