Convective Disaster Early Warning Assessment And Convective Storm System Longevity Prediction

Nowcasting of severe convective weather is an important part of weather forecasting.The convective system that causes severe convective weather has the characteristics of small spatial scale and rapid structural evolution,which makes nowcasting of severe convective weather very difficult.This paper studies strong convective weather.In order to efficiently evaluate the early warning results of existing models,a set of early warning and evaluation systems for severe convective disasters is developed.In order to make nowcasting of the duration of convective systems,a density-based The clustering convective system identification and tracking algorithm is used,and the remaining life of the convective system is analyzed and verified.The main work and research results of this thesis are as follows:(1)In response to the objective evaluation needs of the automatic classification and early warning(recognition)model of severe convection,the "Strong Convective Disaster Assessment System" software was developed.The software reads three types of data:single,early warning and live conditions,extracts time and space information,executes the evaluation algorithm based on pre-designed evaluation rules,visualizes the evaluation results,and finally generates an evaluation report for use by meteorological personnel.This software fills the gap in automated evaluation in the industry.After testing,the system has good response speed and stability,can perform various functions normally,and meets design requirements.(2)In the process of studying the life of convective systems,in order to track convective systems with irregular shapes and scales,a method for identifying and tracking convective systems based on density clustering is proposed.The convective system is composed of convective monomers.Firstly,the tracking of convective monomers is realized by optical flow method,and the adjacent monomers in space are divided into convective systems by density clustering algorithm,and then the results of the individual tracking are used to identify the first time.The convection system is divided and tracked for the second time,and the final recognition and tracking result of the convection system is obtained.Qualitative and quantitative analysis shows that the algorithm proposed in this paper can correctly record the complete process of convective system from birth to disappearance.Compared with the time series of convective monomer,the number of tracked convective system time series is reduced by half,and the average life span is extended by about 14 minutes.(3)In order to guide the time frame of the weather warning results,the remaining life of the convective system is analyzed and verified.First,obtain an effective time series of convective systems.According to the research status of factors affecting the development and evolution of convective systems,four types of features,namely,two-dimensional features,environmental field features,change features,and life features,are designed.The machine learning model is trained,and it is found that the gradient boosting decision tree is used.Making lifetime predictions is valid and outperforms the other two models commonly used in meteorology.