Analysis Of Temporal And Spatial Distribution Characteristics Of Inner Mongolia Lightning Based On Big Data Platform

Lightning constitutes a serious threat to people's daily life.It is of great practical significance to analyze the temporal and spatial law of lightning and forecast thunderstorm with higher accuracy.With the rapid development of the Internet and the continuous improvement of meteorological monitoring level,a large amount of lightning data has come into being.When analyzing or forecasting lightning data,a large number of complicated calculations are required,and large-scale calculations become a long time-consuming problem.The traditional single-machine method has become increasingly difficult to meet the storage and processing of massive lightning data.Therefore,how to better mine and research the massive lightning data has become a research hotspot of the meteorological department.However,the emergence of big data technology has provided a new idea for the processing of massive lightning data.The subject focuses on the analysis of the temporal and spatial distribution of ground flashes in Inner Mongolia.The main research contents include the analysis of the spatial and temporal laws of lightning and the application of thunderstorm forecasting.The analysis of the lightning spatiotemporal law mainly uses Spark operators and K-means algorithm to analyze the location of ground flashes;the thunderstorm prediction mainly uses naive Bayes algorithm to further analyze the temporal and Spatial Laws of lightning,so as to predict whether thunderstorm will occur in a certain period of time in a certain region in the future.The main work is as follows.1.Based on Hadoop + spark platform,the division of high incidence lightning area SCK-means algorithm and law of lightning time STime algorithm is designed.Aiming at the problem of the random selection of initial clustering center and K value in the k-means algorithm,Canopy is used to optimize the k-means algorithm,and the optimized algorithm is designed in parallel based on Spark platform,forming the division algorithm of high incidence lightning area SCK means.The algorithm STime is designed based on distribution law of lightning time Spark operator.Finally,SSE,speedup rario and scalability ratio are used as evaluation indexes.The experimental results show that the algorithm of space-time law designed in this paper can accurately and quickly mine the law of lightning and provide decision support for lightning protection and disaster reduction.2.SPNBC thunderstorm prediction model is designed based on the Hadoop + Spark platform.Aiming at the independence assumption problem of the traditional Bayesian model(NBC),the PCA algorithm is employed to optimize the naive Bayes classifier to construct the PNBC.Then the optimized algorithm is parallelized based on spark platform to form the thunderstorm prediction model SPNBC.Finally,the accuracy rate,empty report rate,speedup ratio and scalability ratio are used as evaluation indicators,and experimental comparison is made with the commonly used BP neural network and traditional naive Bayesian thunderstorm forecasting methods.The experimental results show that the thunderstorm prediction model proposed in this paper has comparatively better accuracy and lower misses of forecasts,and has a large performance advantage when processing massive data.Based on the analysis of the spatial and temporal laws of thunder and lightning and the research of thunderstorm forecasting,this paper designs a storage scheme of thunder and lightning data based on the Hadoop platform,an algorithm for mining lightning time and space based on the Spark platform,and a thunderstorm prediction model.And verified by experiment results,the algorithms and models proposed in this paper can effectively improve the efficiency of lightning data mining and the reliability of thunderstorm prediction.The research results of this paper can be used to build a rapid lightning data analysis platform,which can provide a good service for meteorological personnel to quickly analyze the strength and trend of lightning process,and lay a foundation for the further application of meteorological data.