| In the context of the rapid development of information technology,there has been an explosive growth of data.How to extract effective information from massive data to ensure the smooth progress of research is a hot topic at this stage.Based on these big data background,data mining technology has come into being.Data mining technology is mainly extraction,transformation,analysis and other modeling processing of a large number of data files,and then efficiently digs out meaningful or valuable data information.In this paper combines with data mining technology to mine and process the data sets of AIRS L3 V6 series and IGRA series from 2002 to 2022,and various data such as atmospheric water vapor and atmospheric temperature that people are most concerned about in daily life are mined and filtered for research and analysis.Since the IGRA dataset is data with unevenly distributed of stations,firstly uses the inverse distance interpolation method to perform preprocessing operations such as interpolation and resolution reconstruction on the IGRA dataset,and transforms the data into 1°×1° a grid point data(longitude and latitude)to make it correspond to the AIRS dataset.Then,the AIRS and IGRA datasets are classified in multiple dimensions such as year,month,and season,mining the various features and implicit information of the data,and then discovering the spatial and temporal distribution characteristics of meteorological parameters.Finally,based on the time series prediction model and neural network model,SARIMA model and combined ARIMA-LSTM model are constructed to compare the prediction of meteorological parameters.The work in this paper can be summarized as follows:(1)Meteorological time series data processing and mining analysis.The AIRS dataset and IGRA dataset files are read,cleaned and stored;the stored dataset is used to calculate the probability of occurrence of contrail cirrus according to the Schmidt–Appleman criterion;the visualization platform of the condensation wake determination model is designed and implemented by combining PyQt5 technology;finally,the accuracy of the two datasets is verified by using Pearson correlation coefficient as the judging criterion.(2)Using the validated AIRS dataset from September 2002 to April 2022,the spatialtemporal characteristics of different meteorological parameters in different regions are summarized through by data mining techniques.The research content mainly includes the analysis of the spatial-temporal characteristics and related factors of atmospheric water vapor and the probability of aircraft condensation wake occurrence.Mainly use empirical orthogonal function,linear regression and M-k mutation test to analyze and study the spatio-temporal modes of different meteorological parmeters;the correlation factor analysis mainly adopts Pearson correlation coefficient as the criterion to explore and test the correlation between various meteorological data.Finally,the influence mechanism of the probability of condensation trails is analyzed based on the spatial and temporal distribution characteristics and correlation analysis.(3)Meteorological time series data estimation and combined prediction model design.Based on the temporal prediction model of machine learning model,two times series prediction models,SARIMA and ARIMA-LSTM,are proposed to predict the meteorological data,and then the combined model is found to be more accurate than the traditional single prediction model.Data mining technology is used to analyze valuable information and patterns from a large amount of meteorological data,which in turn provides references for meteorological research and better protects people’s lives and activities.Figure 48 Table 9 Reference 93... |
PDF Full Text Request