Research On The Reconstruction Technology Of Geomagnetic Data

Geomagnetic data is of great significance to the study of the earth’s surface and the space environment outside the earth.At present,there are many methods for collecting geomagnetic data,such as station monitoring,aerial survey and satellite magnetic survey.In recent years,geomagnetic data are missing or abnormal due to the influence of external environment on geomagnetic stations.In order to ensure the normal use of data and the study of related work,it is particularly necessary to reconstruct geomagnetic data.This paper first introduces the source of the topic,the research background and significance,and the research status at home and abroad,and then describes the main applied algorithms and models.On this basis,the geomagnetic data are reconstructed,and compared with the recorded data of the geomagnetic network.The specific research contents of this paper are as follows:(1)The geomagnetic data of the target station is reconstruct by the trained data of the surrounding stations based on back propagation neural network algorithm.This paper studies from two perspectives include the selection of the number of stations and the days of training which from the input layer of the neural network structure,then give the appropriate range of the days of training and the number of stations for data reconstruction for achieving a low average absolute error.In addition,to calculate the degree of correlation between the geomagnetic data of the stations,the close distance and long distance stations are selected respectively by two kinds of correlation coefficients,one is Pearson correlation coefficient and the other is intraclass correlation coefficients.The factors that affect the correlation of the geomagnetic data besides the distance are analyzed in order to reasonably select the station data in the study of geomagnetic reconstruction.(2)On the basis of gaussian geomagnetic theory,the spherical harmonic coefficient of the first-order spherical harmonic function is solved as the unknown quantity,the space position of the surrounding stations and the geomagnetic observation values at different times are taken as the known quantity into the equation,and the first-order time-varying spherical harmonic coefficient is obtained,and the geomagnetic data of the target station in one day is reconstructed.Compared with the data calculated in the IGRF model,the geomagnetic trend obtained by this method is closer to the real value.On this basis,we explore the influence of the relative position of the target stations and the distance between the reference stations and the target stations on the reconstruction effect.When the target station is in the central position,the absolute residual of reconstruction is low.With the increase of the distance between the reference station and the target station,the deviation between the reconstructed geomagnetic value and the observed value becomes larger and larger.Finally,the missing geomagnetic values of the stations around the earthquake are reconstructed by using the time-varying spherical harmonic coefficient,which is used to calculate the daily variable and diurnal ratio of amplitude of geomagnetic vertical component,and the abnormal phenomena of geomagnetic data before the earthquake in a certain area are successfully extracted.In conclusion,this paper achieved a further research in geomagnetic reconstruction algorithm of back propagation neural network and the time-varying spherical harmonic coefficient.The references on reconstruction of geomagnetic data is given for relevant work of geomagnetism and seismology research.It will provide technical research of the project“The Collection and Analysis of Electromagnetic data and the Construction of Landscape Scene”.