利用ARMA模型短期预报北极地区电离层TEC  

Short-term TEC prediction of the arctic ionosphere based on ARMA model

作　　者：卢建春[1] 乔书波[1] 冯炜[2,3] 祝海峰[1] 

机构地区：[1]信息工程大学,河南郑州450001 [2]中国科学院测量与地球物理研究所,湖北武汉430077 [3]北京卫星导航中心,北京100094

出　　处：《测绘工程》2016年第3期30-34,40,共6页Engineering of Surveying and Mapping

基　　金：国家自然科学基金资助项目(U1431115;41374042);大地测量与地球动力学国家实验室开放式基金资助项目(SKLGED 2014-3-5-E)

摘　　要：北极地区电离层结构分布较为特殊,存在梯度变化。利用时间序列分析中的自回归移动平均模型（Autoregressive Moving Average,ARMA）对欧洲定轨中心（CODE）发布的北纬67.5°~87.5°以及利用反距离加权插值法得到的90°的格网数据逐点进行建模,分别利用7d、10d、20d、30d、40d、50d的电离层TEC值为样本数据采用线性最小方差法进行预报分析。结果表明：90%以上的预报绝对误差小于3TECU,预报精度随TEC样本序列长度的增加而提高,但样本序列增加到一定值后,相对精度提高不大;相同样本数据的预报精度随预报时间长度的增加而降低,起初不是很明显,超过20d后精度降低明显且波动幅度较大。尽管北极地区存在梯度变化,ARMA模型在北极地区具有较高的预报精度,是一种比较理想的预报方法。The Arctic ionosphere structure distribution is special,which embraces the changes of gradient.The autoregressive moving average（ARMA）model is used from the time analysis theory to model the grid data of latitude 67.5°~87.5°provided by Center for Orbit Determination in Europe（CODE）and latitude90°is obtained using Inverse Distance to a Power point by point.The ionosphere TEC of 7days,10 days,20days,30 days,40days,50 days as sample data are adopted plus the linear minimum variance method for prediction.The results show that more than 90% of predicted absolute error is less than 3 TECU.The predicted accuracy will be improve with the increase of TEC sample sequences length,but it will gradually reduce if the length exceeds the optimal length about 40 days.On the other hand,with the same TEC sample,and the predicted days increase,the predicted accuracy decreases.Though it is not very obvious in the beginning,it will be reduced distinctly and the fluctuation will be larger over 20 days.Despite the existence of change of gradient in the Arctic,the ARMA model has good predicted accuracy,which will be a kind of ideal predicted model.

关 键 词：ARMA 北极地区 电离层TEC 线性最小方差预报 

分 类 号：P228[天文地球—大地测量学与测量工程]