一种可选的电离层TEC区域重构方法

An optional algorithm for ionosphere TEC region reconstruction

电离层总电子含量(total electron content,TEC)区域重构常采用克里金空间插值方法,该方法基于电离层TEC数据的位置关系及相关性等特征,同时满足空间加权估计的无偏性和最优性特征,因此被广泛应用.但克里金空间插值方法拟合变异函数时,传统理论变异函数模型面临函数曲线固定、空间细节变化无法反映以及模型选取人为主观等问题.为解决上述问题,文中提出一种可选的电离层TEC区域重构方法,从TEC数据的空间变化特征出发,采用最小二乘支持向量机(least squares support vector machine,LS-SVM)拟合实验变异函数,能够更精确地刻画变异函数,反映电离层TEC空间变化趋势.为验证本方法的准确性,选用中国陆态网地基GNSS台站三组不同时刻穿刺点垂直TEC(vertical TEC,VTEC)值作为实测数据,同时选用普通克里金空间插值方法中的指数理论变异函数模型、球状理论变异函数模型以及LS-SVM理论变异函数模型进行实验对比.三组实验结果表明,本文方法计算的均方根误差(root mean squared error,RMSE)分别为1.54 TECU、1.76 TECU和2.45TECU,平均绝对误差(mean absolute error,MAE)分别为1.21 TECU、1.23 TECU和1.62 TECU,误差均小于其他两种理论变异函数模型,重构效果最好,为电离层TEC区域重构提供了一种可选的思路.

Ionospheric total electron content(TEC)region reconstruction often adopts Kriging spatial interpolation method,which is based on the position relationship and correlation of TEC,satisfying the unbiasedness and optimality of weighted space estimation.However,the traditional theoretical variogram function model faces the problems of fixed function curve,unreflectable spatial detail change and chosen with subjectivity when Kriging spatial interpolation method is used to fit the variation function.To solve the above problems,this paper proposes an optional algorithm for ionosphere TEC region reconstruction.Starting from the spatial variation trend of TEC data,least squares support vector machine(LS-SVM)is used to fit the experimental variogram function,which can describe the variogram function more accurately and reflect the spatial variation trend of ionospheric TEC.To verify the accuracy of this algorithm,the vertical TEC(VTEC)values of three groups of puncture points at different times at GNSS stations in Crustal Movement Observation Network of China are selected as the measured data.The theoretical variogram function model of exponential and spherical in ordinary Kriging spatial interpolation method and the model in this paper are selected for experiments.Three sets of experimental results show that the root mean squared error(RMSE)results calculated by the ionospheric TEC region reconstruction algorithm provided in this paper are 1.54,1.76 and 2.45TECU,and the mean absolute error(MAE)results are 1.21,1.23 and 1.62 TECU,respectively.The errors are smaller than those of the other two theoretical variogram function models,and the interpolation accuracy is the best.This paper provides an optional algorithm for ionosphere TEC region reconstruction.