The Research Of Lightning Near Forecast Based On Clustering And Forecast Results Presentation

Lightning is a kind of destructive phenomenon, which often brings great loss to human society.For many industries, it is of great practical significance to learn the lightning disaster information in advance.Based on the application background and project requirements, this paper analysis the lightning location data in detail, and studies the related prediction and visualization techniques. This study aims to provide a feasible lightning forecast method, and through a certain technical means to present the image of the prediction results, to help the actual lightning disaster protection.First of all, based on the characteristics of spatial data and analysis method, this paper simulated and analyzed the distribution characteristics of lightning location data on different planes, and preliminarily determined the characteristics of the cluster. In order to further explore the intrinsic characteristics of the data, the paper studies the method of cluster analysis.According to the specific selection criteria, the data processing method based on the density based clustering algorithm is determined. Then, this paper discusses the realization method of data processing and the typical DBSCAN algorithm. And through a lot of simulation, the study analyzes the influence of input parameters on the clustering results, offered a basis for the further study.Secondly, in order to determine the lightning prediction method suitable for the data source, this paper makes some research on several typical algorithms, and selects the method based on clustering. Then, the overall implementation process of the method is designed. In order to solve the cluster cloud recognition problem, this paper studies the changes in cloud. According to centroid distance, boundary length and number of lightning and other aspects, the study makes a comprehensive identification of clouds and achieves good results. The evaluation method of forcast result is established. In this study,we obtain the error range of the prediction results,which demonstrates the validity of the prediction method. Also, this paper gives the compensation scheme for the dramatic changes in the cloud.Finally, in order to predict the result image, the data visualization technology is studied. Based on visual C#. Net development platform, this paper designs prediction results presentation system,which makes integrated use of Oracle database, ArcGIS Engine Components and TeeChart component. The system realizes the functions,such astwo-dimensional display,three-dimensional display, time analysis, statistical analysis and so on.