AR模型在大地电磁测深资料处理中的应用  

The application of AR Model in magnetotelluric data processing

作　　者：张春虹[1] 张刚[1] 

机构地区：[1]成都理工大学,地球探测与信息技术教育部重点实验室,成都610059

出　　处：《地球物理学进展》2013年第3期1227-1233,共7页Progress in Geophysics

基　　金：国家科技专项"深部探测技术实验与集成"(SinoProbe-02);国家重大科学仪器设备开发专项"大深度三维电磁探测技术工程化开发"(2011YQ050060);国家自然科学基金项目"扬子地台西缘深部地质结构与油气赋存背景研究"(40839909);中国地质调查局地质调查项目"龙门山及邻近构造带综合地球物理勘查"(1212010914049);中国地质调查局地质调查项目"青藏高原东部及东南部大地电磁剖面探测"(1212011121273;1212011121266)联合资助

摘　　要：大地电磁测深资料数据采集过程中,由于温度、湿度等对仪器的影响或GPS搜星不正常,采集到的数据有时会出现时间序列跳帧或缺失现象.针对这一问题,本文将基于无激励AR(p)模型预测数据的原理引入大地电磁测深数据处理中.根据已知序列确定AR(p)模型阶数以及模型参数,建立正确的预测模型对缺失数据进行预测,并对比经过预测后的数据与实际样本数据的频谱,表明AR(p)预测模型可以解决原始资料的不连续性问题,提高了大地电磁测深野外资料的利用率.In the magnetotelluric sounding raw data acquisition process, due to the influence of temperature, humidity, etc. of the instrument or the GPS satellite is abnormal, the collected time sequence data sometimes occur the phenomenon of frame skipping or deletion. To solve this problem, AR （p） prediction model is introduced to improve the utilization of magnetotelluric deep field observation data, proposing a new method to solve the problem of wild-collected data frame skipping or missing, and applied which into the measured long-period magnetotelluric data. The model order and the model parameters can be determined according to the known sequence, then the AR （p） prediction can be determined, the missing data in the magnetotelluric sounding raw data acquisition process can be predicted and filled to the corresponding position. And the spectrum of predicted data and that of the sample data are contrasted to show that AR（p） model can solve the discontinuity of the original data. this prediction method can make up for the deficiencies caused by the data collection process instrumentation hardware and improve the utilization of magnetotelluric sounding data.

关 键 词：AR(P)模型 预测 大地电磁测深 资料处理 缺失数据 

分 类 号：P318[天文地球—固体地球物理学] P319[天文地球—地球物理学]