Hail Recognition And Kinetic Energy Prediction Based On Doppler Weather Radar

Due to the limitation of space,the impact on the ground,the sudden occurrence of hail itself and the short life cycle,hailstorm prediction has become a complex and arduous task,and the damage caused by hail to the surface assets should not be ignored.Therefore,accurate identification of hail and prediction of kinetic energy is particularly important to protect people's life and property.The research object of this paper is Aksu area.The data of ground automatic monitoring and artificial observation,altitude sounding data and radar data from 2009 to 2011 are used as data.In order to obtain the recognition characteristics of hail cloud and rain cloud,firstly,the echo top height,maximum reflectivity and basic reflectivity images are extracted from radar detection physical quantity products.Secondly,the relationship between 45 dBZ and 0dBZ is obtained by edge detection,image segmentation and calculation of the specified intensity value of the basic reflectivity map.Secondly,the grayscale difference statistics and grayscale gradient texture features of the basic reflectivity map are calculated.Finally,the calculation is made.Statistical measure of basic reflectivity graph.Through calculation and analysis,it is found that the maximum reflectivity factor and echo top height can be used as important parameters to distinguish hail from rainfall,that 45 dBZ has different characteristics before and after rainfall and hail,and that some variables in gray-scale difference statistics are also different before and after hail fall,and that the maximum reflectivity factor and echo top height can be used as important parameters to distinguish hail from rainfall.Grayscale gradient texture features,mean and variance in statistical measure can also effectively distinguish hail and rainfall.Therefore,the feature extracted in this paper can be used as the discriminant factor of the next recognition model.In order to give full play to the calculated hailstorm characteristics,the effect of hail cloud identification can be improved.In this paper,a hailstorm cloud recognition model is constructed by using decision tree algorithm.The evaluation results are obtained by using 2 ×2 contingency table evaluation criterion and ROC analysis.The result shows that the identification effect of a single decision tree is not good enough.In order to improve this result,this paper combines the Bagging algorithm with the decision tree to get the Bagging decision tree recognition model,and compares the prediction results between the new model and the original decision tree model.Through comparison and analysis,it is found that therecognition effect of a single decision tree is not as good as that of the new Bagging decision tree.Model?In order to further analyze the advantages and disadvantages of the new Bagging decision tree model,the optimal identification model is obtained by comparing its prediction results with those of other commonly used classification models.The results show that the hailstorm recognition algorithm of Bagging decision tree has a good effect,its hit rate POD is90%,false alarm rate is 20%,Heidke skill score is 64%,classification accuracy is 81.43%,and it is better than other methods.This result enables us to identify hail accurately.After the hailstorm has been identified,if the hail can be further judged to be harmful to the surface assets,it will make more contribution to the hail prevention and disaster reduction work.Therefore,by analyzing the weather radar hail data,it is found that the hail kinetic energy < 20JM-2 will not cause damage to crops or infrastructure,but the kinetic energy higher than 20JM-2 will certainly cause damage.In order to distinguish further whether hail is a risk to surface assets and make staff take corresponding hail control measures to reduce hail losses in the future,a method is used to estimate the kinetic energy generated by hail.By analyzing multiple features and hail The relationship between kinetic energy and different hailstone characteristics and hail kinetic energy has been established with the corresponding relationship of the Logistic function.Through the test of the fitting effect of the established function and the statistical significance of the variables introduced in the equation,we can see that the fitting result of the Logistic function established in this chapter is good.According to the established function to predict whether hail will cause damage to ground assets,we can see from the evaluation criterion of 2 × 2 contingency table and the comparison with K nearest neighbor classifier that the prediction result of Logistic function is better,in which the hit rate POD is 83.3%.False alarm rate of 12.5%,FAR skill The score was 74.3%.This result enables us to make a careful preliminary estimate of the damage that may be caused by hail.